WO2023000261A1 - Regional traffic prediction method and device - Google Patents

Abstract

Embodiments of the present application disclose a regional traffic prediction method and device. The method comprises the following steps: constructing a road network topological graph of roads and a parking lot topological graph of parking lots in a target region according to road information and parking lot information of the target region; obtaining first feature information of the road network topological graph and second feature information of the parking lot topological graph according to historical average vehicle speed information of the roads and historical parking space occupation information of the parking lots in the target region; fusing the first feature information with the second feature information by means of a multi-channel spatial network, the road network topological graph, and the parking lot topological graph to generate a spatial fusion feature; and predicting average vehicle speed information of each road and parking space occupation information of each parking lot by means of a gated recurrent network and at least two spatial fusion features. By means of the present application, the accuracy of prediction of a road traffic condition can be improved.

Description

A method and device for predicting regional traffic technical field

The present application relates to the technical field of the Internet, and in particular to a method and device for predicting regional traffic.

Background technique

With the rapid development of my country's economy and the acceleration of the modernization process, the number of car ownership and the number of trips continue to increase. The continuous growth of car ownership has led to road traffic congestion and parking difficulties. In-depth research and practice in the field of intelligent transportation in recent years have confirmed its superiority in alleviating traffic congestion, improving road capacity and service levels. Traffic forecasting is an important part of intelligent transportation. It can help managers understand traffic changes in advance, so as to formulate corresponding control strategies. Both road traffic and parking saturation contribute to overall traffic, and together they determine the traffic conditions in an area.

Prediction of current traffic conditions is based on road traffic flow or parking conditions. The prediction methods of traffic flow and parking conditions are basically the same. Early predictions used statistical models, but this method has low anti-interference ability and the prediction results are not accurate. In recent years, due to the strong feature extraction ability and sample space fitting ability of the deep learning model, research on traffic flow and parking space occupancy prediction methods based on the deep learning model has begun, and the prediction effect has improved compared with the early statistical models. But near some popular points of interest, such as hospitals, scenic spots, large shopping malls, etc., the traffic situation is more complicated. On the one hand, the huge traffic flow has brought pressure to the parking lot. On the other hand, limited parking spaces lead to vehicles cruising the road at low speeds unable to find a vacant space. Road traffic flow and parking conditions will affect each other, and there will still be some deviations in predicting traffic conditions through road traffic flow or parking conditions.

Contents of the invention

Embodiments of the present application provide a method and device for predicting regional traffic, which can improve the accuracy of predicting road traffic conditions.

An embodiment of the present application provides a method for predicting regional traffic, which may include:

According to the road information and the parking lot information of the target area, construct the road network topology map of the road in the target area and the parking lot topology map of the parking lot;

According to the historical average vehicle speed information of the road in the target area and the historical parking space occupancy information of the parking lot, the first feature information of the road network topology map and the second feature information of the parking lot topology map are obtained; the first feature information It is used to represent the historical average vehicle speed information of the road in the target area, and the second feature information is used to represent the historical parking space occupancy information of the parking lot in the target area;

Fusing the first feature information with the second feature information through a multi-channel space network, and the road network topology map and the parking lot topology map to generate a space fusion feature;

The average vehicle speed information for each road and the occupancy information for each parking lot are predicted by a recurrent gating network and at least two spatially fused features.

In a feasible implementation manner, the construction of the road network topology map of the road in the target area and the parking lot topology map of the parking lot according to the road information and parking lot information of the target area includes:

Statize the road information of the target area, determine the first connection relationship between each road in the road information according to the natural connection rules of the road, and construct the road network topology map of the target area according to the first connection relationship; the second A connection relationship is used to indicate whether each road is connected in the topology map;

Count the parking lot information in the target area, determine the second connection relationship between each parking lot in the parking lot information according to the shortest path between the parking lots, and construct the parking lot topology in the target area according to the second connection relationship Figure; the second connection relationship is used to indicate whether each parking lot is connected in the topology map.

In a feasible implementation manner, according to the historical average vehicle speed information of the roads in the target area and the historical parking space occupancy information of the parking lot, the first characteristic information of the road network topology map and the first feature information of the parking lot topology map are acquired. Second feature information, including:

Obtain the historical average vehicle speed information of each road in the target area at the target time, generate the average vehicle speed vector corresponding to the road according to the historical average vehicle speed information, and use the average vehicle speed vector as the first feature of the road network topology map at the target time information;

Obtain the historical parking space occupancy information of each parking lot in the target area at the target moment, generate a parking space occupancy vector corresponding to the parking lot according to the historical parking space occupancy information, and use the parking space occupancy vector as the topological map of the parking lot at the target time. Second characteristic information.

In a feasible implementation manner, the first feature information and the second feature information are fused to generate a space fusion features, including:

Input the adjacency matrix of the road network topology map and the first feature information at the target time into the first channel of the multi-channel space network, and obtain the first spatial feature at the target time through the graph convolutional neural network in the first channel ;

Input the adjacency matrix of the topological map of the parking lot and the second feature information of the target time into the second channel of the multi-channel spatial network, and obtain the second spatial feature at the target time through the graph convolutional neural network in the second channel ;

The first spatial feature and the second spatial feature are fused to generate a spatial fusion feature at the target moment.

In a feasible implementation manner, the prediction of the average vehicle speed information of each road and the parking space occupancy information of each parking lot through the loop gating network and at least two spatial fusion features includes:

Input the spatial fusion feature of T ₁ -T _k time into the loop gating network, generate the state information of each time in T ₁ -T _k time and the prediction information of the target area; the state information is the hidden State, used to generate prediction information, the k is a positive integer greater than 1, and the loop gating network includes k loop gating units;

The average vehicle speed information of each road and the parking space occupancy information of each parking lot are predicted according to the prediction information.

In a feasible implementation manner, the input of the spatial fusion feature at time T ₁ -T _k into the loop gating network to generate state information at time T ₁ -T _k and prediction information of the target area includes:

Input the spatial fusion feature at T1 moment into the _first loop gating unit of the loop gating network, and generate the state information _h1 at T1 moment _;

The spatial fusion feature at T2 moment and the state information _h1 at the T1 moment are input into the _second cycle gating unit _of the cycle gating network to generate the state information _h2 at the T2 moment _;

Input the spatial fusion feature at time T _k and the state information h _{k- 1 at time T k-} 1 into the kth loop gating unit of the loop gating network, and generate the state information h _{k at time T k and} the target Forecast information for the area.

In a feasible implementation manner, the prediction of the average vehicle speed information of each road and the parking space occupancy information of each parking lot according to the prediction information includes:

The prediction information includes a first vector and a second vector; the first vector corresponds to the average vehicle speed information of each road, and the second vector corresponds to the parking space occupancy information of each parking lot;

Predicting the average vehicle speed information of each road according to the first vector and the corresponding relationship between each dimension in the first vector and the road;

According to the second vector and the corresponding relationship between each dimension in the second vector and the parking lot, the parking space occupancy information of each parking lot is predicted.

An embodiment of the present application provides an area traffic forecasting device, which may include:

A topology map construction unit, configured to construct a road network topology map of roads in the target area and a parking lot topology map of the parking lot according to the road information and parking lot information in the target area;

A feature information acquisition unit, configured to acquire the first feature information of the road network topology map and the second feature information of the parking lot topology map according to the historical average vehicle speed information of the road in the target area and the historical parking space occupancy information of the parking lot ; The first feature information is used to represent the historical average vehicle speed information of the road in the target area, and the second feature information is used to represent the historical parking space occupancy information of the parking lot in the target area;

A feature fusion unit, configured to fuse the first feature information with the second feature information through a multi-channel space network, and the road network topology map and the parking lot topology map to generate a space fusion feature;

The information prediction unit is used to predict the average vehicle speed information of each road and the parking space occupancy information of each parking lot through the recurrent gating network and at least two spatial fusion features.

In a feasible implementation manner, the topology map construction unit is specifically used for:

Statize the road information of the target area, determine the first connection relationship between each road in the road information according to the natural connection rules of the road, and construct the road network topology map of the target area according to the first connection relationship; the second A connection relationship is used to indicate whether each road is connected in the topology map;

Count the parking lot information in the target area, determine the second connection relationship between each parking lot in the parking lot information according to the shortest path between the parking lots, and construct the parking lot topology in the target area according to the second connection relationship Figure; the second connection relationship is used to indicate whether each parking lot is connected in the topology map.

In a feasible implementation manner, the feature information acquiring unit is specifically configured to:

Obtain the historical average vehicle speed information of each road in the target area at the target time, generate the average vehicle speed vector corresponding to the road according to the historical average vehicle speed information, and use the average vehicle speed vector as the first feature of the road network topology map at the target time information;

Obtain the historical parking space occupancy information of each parking lot in the target area at the target moment, generate a parking space occupancy vector corresponding to the parking lot according to the historical parking space occupancy information, and use the parking space occupancy vector as the topological map of the parking lot at the target time. Second characteristic information.

In a feasible implementation manner, the feature fusion unit is specifically used for:

Input the adjacency matrix of the road network topology map and the first feature information at the target time into the first channel of the multi-channel space network, and obtain the first spatial feature at the target time through the graph convolutional neural network in the first channel ;

Input the adjacency matrix of the topological map of the parking lot and the second feature information of the target time into the second channel of the multi-channel spatial network, and obtain the second spatial feature at the target time through the graph convolutional neural network in the second channel ;

The first spatial feature and the second spatial feature are fused to generate a spatial fusion feature at the target moment.

In a feasible implementation manner, the information prediction unit includes:

The information generation subunit is used to input the spatial fusion feature of T ₁ -T _k time into the loop gating network, and generate the state information of each time point in T ₁ -T _k time and the prediction information of the target area; the state The information is the hidden state at each moment, and is used to generate prediction information, the k is a positive integer greater than 1, and the loop gating network includes k loop gating units;

The information prediction subunit is used to predict the average vehicle speed information of each road and the parking space occupancy information of each parking lot according to the prediction information.

In a feasible implementation manner, the information generating subunit is specifically configured to:

Input the spatial fusion feature at T1 moment into the _first loop gating unit of the loop gating network, and generate the state information _h1 at T1 moment _;

The spatial fusion feature at T2 moment and the state information _h1 at the T1 moment are input into the _second cycle gating unit _of the cycle gating network to generate the state information _h2 at the T2 moment _;

Input the spatial fusion feature at time T _k and the state information h _{k- 1 at time T k-} 1 into the kth loop gating unit of the loop gating network, and generate the state information h _{k at time T k and} the target Forecast information for the area.

In a feasible implementation manner, the information prediction subunit is specifically configured to:

The prediction information includes a first vector and a second vector; the first vector corresponds to the average vehicle speed information of each road, and the second vector corresponds to the parking space occupancy information of each parking lot;

Predicting the average vehicle speed information of each road according to the first vector and the corresponding relationship between each dimension in the first vector and the road;

Predict the parking space occupancy information of each parking lot according to the second vector and the corresponding relationship between each dimension in the second vector and the parking lot.

On the one hand, the embodiments of the present application provide a computer-readable storage medium, where a computer program is stored in the computer-readable storage medium, and the computer program is adapted to be loaded by a processor and execute the above-mentioned method steps.

An embodiment of the present application provides a computer device, including: a processor, a memory, and a network interface; the processor is connected to the memory and the network interface, wherein the network interface is used to provide a network communication function , the memory is used to store program codes, and the processor is used to call the program codes to execute the above method steps.

An embodiment of the present application provides a computer program product or computer program, where the computer program product or computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the above-mentioned method steps.

In this embodiment of the application, the road network topology map of the roads in the target area and the parking lot topology map of the parking lot are constructed according to the road information and parking lot information in the target area, and further based on the history of the roads in the target area The average vehicle speed information and the historical parking space occupancy information of the parking lot, the first feature information of the road network topology map and the second feature information of the parking lot topology map are obtained, and the multi-channel space network is used, and the road network topology map and all The topological map of the parking lot, the first feature information and the second feature information are fused to generate a spatial fusion feature, and finally the average vehicle speed information and each Occupancy information of each parking lot. Using the above method avoids the mutual influence of road traffic flow and parking conditions in areas with complex traffic conditions, which leads to deviations in predicting traffic conditions with a single road traffic flow or parking conditions, and improves the accuracy of road traffic condition prediction .

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments of the present application. Those skilled in the art can also obtain other drawings based on these drawings without creative work.

FIG. 1 is a system architecture diagram of a regional traffic forecast provided by an embodiment of the present application;

FIG. 2 is a schematic flowchart of a method for predicting regional traffic provided by an embodiment of the present application;

Fig. 3a is an exemplary schematic diagram of a target area provided by an embodiment of the present application;

Fig. 3b is an exemplary schematic diagram of road and parking lot visualization provided by the embodiment of the present application;

Fig. 3c is a schematic diagram of an example of a road network topology diagram provided by an embodiment of the present application;

Fig. 3d is a schematic diagram of an example of a parking lot topology provided by an embodiment of the present application;

Fig. 4 is a schematic diagram of an example of a generated spatial fusion feature provided by an embodiment of the present application;

FIG. 5 is a schematic flowchart of a method for predicting regional traffic provided by an embodiment of the present application;

Fig. 6a is a schematic diagram of an example of generating prediction information provided by an embodiment of the present application;

Fig. 6b is an example schematic diagram of the prediction accuracy of a model provided by the embodiment of the present application;

Fig. 6c is an exemplary schematic diagram of the relationship between prediction accuracy and mutual information provided by the embodiment of the present application;

FIG. 7 is a schematic structural diagram of a regional traffic forecasting device provided by an embodiment of the present application;

Fig. 8 is a schematic structural diagram of a computer device provided by an embodiment of the present application.

detailed description

The following will clearly and completely describe the technical solutions in the embodiments of the application with reference to the drawings in the embodiments of the application. Apparently, the described embodiments are only some of the embodiments of the application, not all of them. Based on the embodiments in this application, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the scope of protection of this application.

As shown in Figure 1, the network architecture diagram may include a service server 100 and a user terminal cluster, the user terminal cluster may include a user terminal 10a, a user terminal 10b, ..., a user terminal 10c, wherein there may be communication between the user terminal clusters connection, for example, there is a communication connection between the user terminal 10a and the user terminal 10b, there is a communication connection between the user terminal 10b and the user terminal 10c, and any user terminal in the user terminal cluster can have a communication connection with the service server 100, for example, the user There is a communication connection between the terminal 10 a and the service server 100 , and there is a communication connection between the user terminal 10 b and the service server 100 .

Wherein, the above-mentioned user terminal cluster (also including the above-mentioned user terminal 10a, user terminal 10b, and user terminal 10c) may be integrated with target applications installed. Optionally, the target application may include an application having functions of acquiring map data, processing road network information, and constructing a topology map. The database 10d stores a multi-channel spatial network and a circular gating network. Specifically, the user terminal constructs the road network topology map of the road in the target area and the parking lot topology of the parking lot according to the road information and parking lot information in the target area. Further, according to the historical average vehicle speed information of the road in the target area and the historical parking space occupancy information of the parking lot, the first feature information of the road network topology map and the second feature information of the parking lot topology map are obtained, so The first feature information is used to represent the historical average vehicle speed information of the road in the target area, and the second feature information is used to represent the historical parking space occupancy information of the parking lot in the target area. Through the multi-channel spatial network in the database 10d, and the road The network topology map and the parking lot topology map, the first feature information and the second feature information are fused to generate a space fusion feature, and finally through the loop gating network in the database 10d and at least two space fusion features, predict Average vehicle speed information for each road and parking space occupancy information for each parking lot. It should be noted that the multi-channel spatial network and the loop gating network stored in the database 10d can be stored locally in the user terminal, and the prediction of the average vehicle speed information of each road and the parking space occupancy information of each parking lot can be completed on the user terminal side . Optionally, the above user terminal may be any user terminal selected from the user terminal cluster in the above embodiment corresponding to FIG. 1 , for example, the user terminal may be the above user terminal 10b.

It can be understood that the method provided in the embodiment of the present application can be executed by a computer device, and the computer device includes but is not limited to a terminal or a server. The server 100 in the embodiment of the present application can be a computer device, and the user terminals in the user terminal cluster can also be It can be a computer device, which is not limited here. The above-mentioned business server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, and can also provide cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication , middleware services, domain name services, security services, CDN, and cloud servers for basic cloud computing services such as big data and artificial intelligence platforms. The above-mentioned terminals may include: smart phones, tablet computers, notebook computers, desktop computers, smart TVs, smart speakers, desktop computers, smart watches and other smart terminals with image recognition functions, but are not limited thereto. Wherein, the user terminal and the service server may be connected directly or indirectly through wired or wireless communication, which is not limited in this application.

Further, for ease of understanding, please refer to FIG. 2 , which is a schematic flowchart of a method for predicting regional traffic provided by an embodiment of the present application. The method may be executed by a user terminal (eg, the user terminal shown in FIG. 1 ), or jointly executed by the user terminal and a service server (such as the service server 100 in the embodiment corresponding to FIG. 1 ). For ease of understanding, this embodiment takes the method executed by the above-mentioned user terminal as an example for description. Wherein, the prediction method of the regional traffic may at least include the following steps S101-step S104:

S101, according to the road information and the parking lot information of the target area, construct the road network topology map of the road in the target area and the parking lot topology map of the parking lot;

Specifically, the user terminal constructs the road network topology map of the road in the target area and the parking lot topology map of the parking lot according to the road information and parking lot information in the target area. It can be understood that the target area can be Any area including roads and parking lots, such as a certain area of a city. The road information is the number of roads in the target area and the connection relationship between the roads, and the parking lot information is the number of parking lots in the target area and distance information between the parking lots. The road network topology diagram is a topological structure diagram representing road information. The road network topology diagram includes nodes representing roads and the relationship between nodes. The parking lot topology diagram represents the topology structure diagram of parking lot information. The parking lot topology diagram Including the nodes representing the parking lot, and the relationship between each node.

The specific process of constructing road network topology map and parking lot topology map is as follows:

Please refer to Fig. 3a. Fig. 3a is an example schematic diagram of the target area provided by the embodiment of the present application. and the vector data of the parking lot to visualize the roads and parking lots in the target area, please refer to Fig. 3b. Fig. 3b is an example schematic diagram of the visualization of roads and parking lots provided by the embodiment of the present application, as shown in Fig. 3b, the straight line in the figure is a road, and the dot is a parking lot, where the road has direction information, that is, a two-way street is understood as two roads.

The user terminal collects the road information of the target area, determines the first connection relationship between each road in the road information according to the natural connection rules of the roads, and constructs the road network topology map of the target area according to the first connection relationship. The natural connection rule is whether the roads intersect in the real world, and the first connection relationship is whether the roads are connected in the topology map. For example, if the road R ₁ and the road R ₂ are intersecting in the real world, then the road R ₁ and the road R ₂ are connected in the topological graph, if the road R ₁ and the road R ₂ are not intersecting in the real world, Then the road R ₁ and the road R ₂ are not connected in the topological graph. Please refer to Figure 3c, Figure 3c is an example schematic diagram of the road network topology provided by the embodiment of the present application, as shown in Figure 3c, the target area includes 11 roads, which are respectively marked as R ₁ , R ₂ , ... R ₁₁ , Each road is regarded as a node. If two roads intersect, the two nodes will be connected. The road network topology map of the road can be recorded as G _r = (V _r , E _r ), where V _r represents the Set, V _r ={v ₁ ,v ₂ ,...,v _N }, N is the number of nodes, E _r represents the set of all edges. In addition, the connection relationship between each node is represented by an adjacency matrix A _r , A _r ∈ R ^N×N .

Further, the user terminal collects the parking lot information in the target area, determines the second connection relationship between each parking lot in the parking lot information according to the shortest path between the parking lots, and constructs the target according to the second connection relationship. Parking topology map of the area. The shortest path is the distance of the parking lot in the real world, and the second connection relationship is determined according to whether the shortest distance is less than a preset distance threshold, and the second connection relationship is whether the parking lot is connected in the topology map. For example, if the actual distance between the parking lot P1 and the parking lot _P2 in the real world is less _than the preset distance threshold, then the parking lot _P1 and the parking lot _P2 are connected in the topology map, if the parking lot _P1 and If the actual distance _of the parking lot P2 in the real world is greater _than or equal to the preset distance threshold, then the parking lot P1 and the parking lot _P2 are not connected in the topology map. Please refer to Fig. 3d. Fig. 3d is an example schematic diagram of the topology map of the parking lot provided by the embodiment of the present application. As shown in Fig. 3c, the target area includes 7 parking lots, respectively marked as P ₁ , P ₂ , ... P ₇ , each parking lot is regarded as a node, and the shortest road distance between two parking lot nodes is calculated, and if the distance is less than 600 meters, the two nodes are connected. The parking lot topology map of the parking lot can be recorded as G _p =(V _p ,E _p ), where V _p represents the set of all nodes, V _p ={v ₁ ,v ₂ ,...,v _M }, M is the number of parking lots, and E _p represents the set of all sides. In addition, the connection relationship between each node is represented by A _p with adjacency matrix, A _p ∈ R ^M×M .

S102. According to the historical average vehicle speed information of the road in the target area and the historical parking space occupancy information of the parking lot, acquire the first feature information of the road network topology map and the second feature information of the parking lot topology map;

表示t时刻的所有节点特征，

Specifically, the user terminal obtains the first feature information of the road network topology map and the second feature information of the parking lot topology map according to the historical average vehicle speed information of the road in the target area and the historical parking space occupancy information of the parking lot, so The first feature information is used to represent the historical average vehicle speed information of the road in the target area, and the second feature information is used to represent the historical parking space occupancy information of the parking lot in the target area. It can be understood that the user terminal obtains each The historical average vehicle speed information of the road at the target time, the target time is the sampling time of the historical average vehicle speed information, the average vehicle speed vector corresponding to the road is generated according to the historical average vehicle speed information, and the average vehicle speed vector is used as the road network topology map at the target time The first characteristic information of . For example, to obtain the historical average vehicle speed information x ₁ , x ₂ , ... x _T of T target moments, then the average vehicle speed vector of each road node, that is, the first characteristic information is recorded as x _r =[x ₁ ,x ₂ ,...,x _t ,...,x _T ], T is the length of the historical time series, x _t represents the node characteristics at time t,

Indicates all node features at time t,

表示t时刻的所有节点特征，

Further, the user terminal obtains the historical parking space occupancy information of each parking lot in the target area at the target moment, generates a parking space occupancy vector corresponding to the parking lot according to the historical parking space occupancy information, and uses the parking space occupancy vector as the parking lot topology The second characteristic information of the graph at the target time. For example, to obtain the historical parking space occupancy information x ₁ , x ₂ , ... x _T , at T target moments, then the parking space occupancy vector of each parking lot node, that is, the second characteristic information is recorded as x _p =[x ₁ , x ₂ ,...,x _t ,...,x _T ], T is the length of the historical time series, x _t represents the node characteristics at time t,

Indicates all node features at time t,

S103. Fuse the first feature information with the second feature information through a multi-channel space network, and the road network topology map and the parking lot topology map to generate a space fusion feature;

Specifically, the user terminal fuses the first feature information and the second feature information through a multi-channel space network, and the road network topology map and the parking lot topology map to generate a space fusion feature. It can be understood that , the multi-channel spatial network is called MCSN, including multiple prediction channels, each of which includes a two-layer graph convolutional neural network (GCN), but each prediction channel is heterogeneous and can be To process different node topology graph data, a two-layer GCN model is expressed as follows:

进一步对

进行归一化处理，即

其中

为度矩阵，

W ₀和W ₁是权重矩阵，σ(·)代表激活函数，一般采用Relu()作为激活函数。 Among them, X is the feature matrix and A is the adjacency matrix. In order to preserve its own information in the process of aggregating node features, it is generally necessary to add a self-loop to each node. Specifically, it can be realized by adding the adjacency matrix A and the identity matrix I, namely

further to

to normalize, that is,

in

is the degree matrix,

W ₀ and W ₁ are weight matrices, σ( ) represents the activation function, and Relu() is generally used as the activation function.

The following is an illustration of a multi-channel space network MCSN that contains two channels. MCSN can be expressed as follows:

和

分别为t时刻路网拓扑图和停车场拓扑图的特征矩阵。f(·)表示双层GCN。FC(·)表示全连接层。 Among them, _Ar and A _p are the adjacency matrix of the road network topology map and the parking lot topology map, respectively,

and

are the feature matrices of the road network topology map and the parking lot topology map at time t, respectively. f( ) denotes a two-layer GCN. FC( ) denotes a fully connected layer.

Please refer to Fig. 4. Fig. 4 is an example schematic diagram of the generated spatial fusion feature provided by the embodiment of the present application. As shown in Fig. 4, the multi-channel spatial network includes two channels, that is, the first channel and the second channel, and each channel includes Two-layer graph convolutional neural network GCN.

The user terminal inputs the adjacency matrix of the road network topology map and the first characteristic information of the target moment into the first channel of the multi-channel space network, the adjacency matrix of the road network topology map represents the connection relationship between each road node, and the first channel Process the road network topology graph with N nodes. Further, the first spatial feature at the target moment is obtained through the graph convolutional neural network in the first channel, the first spatial feature is the feature information of the road nodes extracted through the first channel, specifically, through the graph convolution The convolution kernel in the neural network performs feature extraction on the adjacency matrix and the first feature information, and generates the first spatial feature at the target moment through the fully connected layer.

The user terminal inputs the adjacency matrix of the topological map of the parking lot and the second characteristic information of the target moment into the second channel of the multi-channel space network, the adjacency matrix of the topological map of the parking lot represents the connection relationship between each parking lot node, and the second The channel processes the road network topology graph of M nodes. Further, the second spatial feature at the target moment is obtained through the graph convolutional neural network in the second channel, the second spatial feature is the feature information of the parking lot node extracted through the second channel, specifically, through the graph volume The convolution kernel in the product neural network performs feature extraction on the adjacency matrix and the second feature information, and generates the second spatial feature at the target moment through the fully connected layer.

Finally, the user terminal fuses the first spatial feature and the second spatial feature through the multi-channel spatial network to generate a spatial fusion feature at the target moment, specifically, input the first spatial feature and the second spatial feature into the multi-channel spatial network The splicing layer performs vector splicing. For example, if the first spatial feature is an n-dimensional vector and the second spatial feature is an m-dimensional vector, then a vector with a dimension of n+m is generated through the splicing layer. Further, the spliced vectors are passed through a fully connected layer to generate spatial fusion features at the target moment.

It should be noted that the above multi-channel spatial network can be used to generate spatial fusion features at time 1, 2, ... T, that is, spatial fusion features at each time point in the historical time series can be generated.

S104. Predict the average vehicle speed information of each road and the parking space occupancy information of each parking lot through the recurrent gating network and at least two spatial fusion features.

In a feasible implementation, the loop gating network can predict the output at the current moment through the gating mechanism for the input and memory information of the time series. Specifically, the loop gating network can include multiple loop gating units ( GRU), the spatial fusion feature can be generated by the multi-channel spatial network MCSN, therefore, the cyclic gating network and MCSN are combined to generate the MCSTN model, and the MCSTN model can predict the average vehicle speed information of each road and the parking space occupancy information of each parking lot , MCSTN model This is a multi-input and multi-output prediction model. MCSTN can include multiple MCSNs and the same number of GRUs as MCSNs. Each MCSN corresponds to a GRU. The input data of MCSNs are the first feature information and The second feature information, the output data of MCSN is the spatial fusion feature at the target time, the input data of GRU is the spatial fusion feature of the target time output corresponding to MCSN and the output of GRU at the previous time, and the output data of GRU is the target time The status information is also used as the input data of the GRU at the next moment.

Specifically, if the loop gating network includes k loop gating units and _k MCSNs, the user terminal inputs the spatial fusion features at time T _{1 -T k} into the loop gating network to generate State information at each moment and prediction information of the target area; the state information is a hidden state at each moment for generating prediction information, and k is a positive integer greater than 1.

Further, the average vehicle speed information of each road and the parking space occupancy information of each parking lot are predicted according to the prediction information. Specifically, the prediction information includes vectors corresponding to the average vehicle speed information and the parking space occupancy information, and the average vehicle speed information of each road and the parking space occupancy information of each parking lot are predicted according to the above vectors.

In this embodiment of the application, the road network topology map of the roads in the target area and the parking lot topology map of the parking lot are constructed according to the road information and parking lot information in the target area, and further based on the history of the roads in the target area The average vehicle speed information and the historical parking space occupancy information of the parking lot, the first feature information of the road network topology map and the second feature information of the parking lot topology map are obtained, and the multi-channel space network is used, and the road network topology map and all The topological map of the parking lot, the first feature information and the second feature information are fused to generate a spatial fusion feature, and finally the average vehicle speed information and each Occupancy information of each parking lot. Using the above method avoids the mutual influence of road traffic flow and parking conditions in areas with complex traffic conditions, which leads to deviations in predicting traffic conditions with a single road traffic flow or parking conditions, and improves the accuracy of road traffic condition prediction .

Please refer to FIG. 5 . FIG. 5 is a schematic flowchart of a method for predicting regional traffic provided by an embodiment of the present application. The method may be executed by a user terminal (eg, the user terminal shown in FIG. 1 ), or jointly executed by the user terminal and a service server (such as the service server 100 in the embodiment corresponding to FIG. 1 ). For ease of understanding, this embodiment takes the method executed by the above-mentioned user terminal as an example for description. Wherein, the prediction method of the regional traffic may at least include the following steps S201-step S205:

S201. According to the road information and parking lot information in the target area, construct the road network topology map of the roads in the target area and the parking lot topology map of the parking lot;

Wherein, for step S201 in the embodiment of the present invention, refer to the specific description of step S101 in the embodiment shown in FIG. 1 , and details are not repeated here.

S202. Obtain first feature information of the road network topology map and second feature information of the parking lot topology map according to the historical average vehicle speed information of roads in the target area and the historical parking space occupancy information of the parking lot;

For step S202 in the embodiment of the present invention, refer to the specific description of step S102 in the embodiment shown in FIG. 1 , and details are not repeated here.

S203. Fuse the first feature information with the second feature information through a multi-channel space network, and the road network topology map and the parking lot topology map to generate a space fusion feature;

Wherein, for step S203 in the embodiment of the present invention, refer to the specific description of step S103 in the embodiment shown in FIG. 1 , and details are not repeated here.

S204, input the spatial fusion feature at time T ₁ -T _k into the loop gating network, and generate state information at each time of T ₁ -T _k and prediction information of the target area; the state information is each time The hidden state of is used to generate prediction information, and the k is a positive integer greater than 1.

Please refer to Fig. 6a. Fig. 6a is an example schematic diagram of generating prediction information provided by the embodiment of the present application. As shown in Fig. 6a, the MCSTN model generated by combining the cyclic gating network and the multi-channel space network MCSN is shown in the figure, and the cyclic gating network Including k recurrent gating units (GRU), the MCSTN model is a multi-input and multi-output prediction model.

Specifically, the spatial fusion feature at T1 is input into the _first loop gating unit of the loop gating network, and the state information _h1 at T1 is generated _;

The spatial fusion feature at T2 moment and the state information _h1 at the T1 moment are input into the _second cycle gating unit _of the cycle gating network to generate the state information _h2 at the T2 moment _;

Input the spatial fusion feature at time T _k and the state information h _{k- 1 at time T k-} 1 into the kth loop gating unit of the loop gating network, and generate the state information h _{k at time T k and} the target Forecast information for the area.

和历史车位占用信息

生成，因此可以将MCSTN模型作为一个整体，则模型的输入为历史平均车速信息

和历史车位占用信息

通过MCSN完成多通道的特征提取和融合，GRU完成时间序列预测。模型的输出为

和

结合MCSN和GRU的表达式可以得到MCSTN的表达式： Among them, the spatial fusion feature at time T ₁ -T _k is determined by the historical average vehicle speed information

and historical parking space occupancy information

Generated, so the MCSTN model can be taken as a whole, and the input of the model is historical average vehicle speed information

and historical parking space occupancy information

Multi-channel feature extraction and fusion are completed through MCSN, and GRU completes time series prediction. The output of the model is

and

The expression of MCSTN can be obtained by combining the expressions of MCSN and GRU:

为候选隐藏状态，即为当前时刻的记忆信息。h _t为t时刻的输出状态，将被传递到下一节点。W _z为更新门的权重，W _r为重置门的权重，W为候选隐藏状态的权重。 Among them, h _t-1 is the hidden state at time t-1, including the related state of the previous node. _rt is the reset gate, which is used to control the degree of ignoring the state information at the previous moment. z _t is an update gate, which is used to control the degree to which the state information of the previous moment is brought into the current state.

is the candidate hidden state, which is the memory information at the current moment. h _t is the output state at time t, which will be passed to the next node. W _z is the weight of the update gate, W _r is the weight of the reset gate, and W is the weight of the candidate hidden state.

S205. Predict the average vehicle speed information of each road and the parking space occupancy information of each parking lot according to the prediction information.

和

为第一向量，

为第二向量。 Specifically, the prediction information includes a first vector and a second vector, the first vector corresponds to the average vehicle speed information of each road, and the second vector corresponds to the parking space occupancy information of each parking lot. See Figure 6a, the output of the model is

and

is the first vector,

is the second vector.

为N维向量，对应N条道路，即第一向量的第一维对应第一条道路的平均车速信息，第一向量的第N维对应第N条道路的平均车速信息。 Further, predict the average vehicle speed information of each road according to the first vector and the corresponding relationship between each dimension in the first vector and the road, specifically, the first vector

is an N-dimensional vector, corresponding to N roads, that is, the first dimension of the first vector corresponds to the average vehicle speed information of the first road, and the N-th dimension of the first vector corresponds to the average vehicle speed information of the N-th road.

为M维向量，对应M个停车场，即第二向量的第一维对应第一个停车场的车位占用信息，第一向量的第M维对应第M个停车场的车位占用信息。 Further, the parking space occupancy information of each parking lot is predicted according to the second vector and the corresponding relationship between each dimension in the second vector and the parking lot. Specifically, the second vector

is an M-dimensional vector, corresponding to M parking lots, that is, the first dimension of the second vector corresponds to the parking space occupancy information of the first parking lot, and the Mth dimension of the first vector corresponds to the parking space occupancy information of the Mth parking lot.

Using the above method, the MCSTN-based integrated prediction of traffic conditions in the target area is realized. Whether the traffic in the target area is smooth or congested is not only affected by the traffic flow, but also by the parking saturation in the same area. Compared with MCSTN, the existing forecasting models are single-channel. During the forecasting process, they only focus on a single data and ignore other related traffic behaviors, such as only focusing on road traffic conditions or parking conditions. In some cases, there is a strong correlation between road traffic and parking. Especially near some popular points of interest, such as scenic spots, hospitals, and the traffic situation around large shopping malls is very complicated.

In the MCSTN model, the traffic conditions in an area are simultaneously predicted, including road traffic conditions and parking conditions. MCSTN has a wider field of vision, so the prediction effect on road traffic and parking lot conditions is better than that of existing models. .

The following compares the prediction results of the average vehicle speed information and parking space occupancy information between the method in this solution and the method in the prior art according to actual scenarios. The prior art uses a T-GCN model for illustration, and the T-GCN model is a single-channel spatio-temporal model.

The comparison experiment selects several parking lots in district B of city A and several roads around them as the experimental scene for the experiment. The average vehicle speed information of each road and the parking space occupancy information of each parking lot within 30 days are collected. Specifically, the target parking lot and the target road can be selected as prediction objects among many parking lots and roads.

Please refer to Fig. 6b, Fig. 6b is an example schematic diagram of the prediction accuracy of the model provided by the embodiment of the present application, as shown in Fig. 6b, the curve in the figure is the change of the prediction accuracy of the model in one day, and curve 1 is the prediction accuracy of the MCSTN model , Curve 2 is the prediction accuracy of the T-GCN model. Specifically, the prediction accuracy of every 15 time slices is calculated on the test set, and the time-varying curve is obtained. The prediction accuracy Accuracy is shown by the following formula.

为预测的平均车速信息，Y _p为真实的车位占用信息，

为预测的车位占用信息，‖·‖ _F为F范数。 Among them, Y _r is the real average vehicle speed information,

is the predicted average vehicle speed information, Y _p is the real parking space occupancy information,

is the predicted parking space occupancy information, and ‖·‖ _F is the F norm.

The accuracy of the two models is relatively close between 8:00 pm and 6:00 am, while at other times, the prediction accuracy of the MCSTN model is significantly higher than that of the T-GCN model. It can be speculated that the variation of the prediction accuracy of the model is related to the traffic conditions. Between 8:00 p.m. and 6:00 a.m., when there are relatively few cars on the road and parking spaces are plentiful, the prediction accuracy of the two models is very close. But as the parking number increases sharply from 8:00 am, the prediction accuracy of T-GCN decreases, while that of MCSTN increases slightly in this case. Therefore, the difference in prediction accuracy of the two models may be caused by the correlation between road traffic and parking.

In order to confirm the above difference in prediction accuracy, mutual information is used to measure the correlation between the average vehicle speed information on the road and the parking space occupancy information. According to the definition of mutual information, the greater the mutual information value, the stronger the correlation between the two variables. Please refer to Figure 6c, Figure 6c is an example diagram of the relationship between the prediction accuracy and mutual information provided by the embodiment of the present application, as shown in Figure 6c, "○" is the prediction accuracy of the MCSTN model, and "X" is the T-GCN model The abscissa in the figure is the mutual information between the average vehicle speed information and the parking space occupancy information, and the ordinate is the prediction accuracy of the model. It can be seen from the figure that when the mutual information is small, the accuracy of the two types of models is not much different. When the mutual information increases, the prediction accuracy of the MCSTN model is significantly better than that of the T-GCN model. Therefore, the prediction of single data is only suitable for dealing with the low correlation of traffic activities. In the face of complex real traffic environment, integrated prediction can better solve the potential and subtle correlation between different traffic activities in the same space-time environment. sexual issues.

Therefore, using the MCSTN model in this program has a higher accuracy in predicting road traffic and parking conditions.

In this embodiment of the application, the road network topology map of the roads in the target area and the parking lot topology map of the parking lot are constructed according to the road information and parking lot information in the target area, and further based on the history of the roads in the target area The average vehicle speed information and the historical parking space occupancy information of the parking lot, the first feature information of the road network topology map and the second feature information of the parking lot topology map are obtained, and the multi-channel space network is used, and the road network topology map and all The topological map of the parking lot, the first feature information and the second feature information are fused to generate a spatial fusion feature, and finally the average vehicle speed information and each Occupancy information of each parking lot. Using the above method avoids the mutual influence of road traffic flow and parking conditions in areas with complex traffic conditions, which leads to deviations in predicting traffic conditions with a single road traffic flow or parking conditions, and improves the accuracy of road traffic condition prediction .

Please refer to FIG. 7 . FIG. 7 is a schematic structural diagram of an area traffic forecasting device provided by an embodiment of the present application. The prediction device of the regional traffic can be a computer program (including program code) running in the computer equipment, for example, the prediction device of the regional traffic is an application software; this device can be used to execute the method provided by the embodiment of the present application corresponding steps. As shown in FIG. 7 , the regional traffic prediction device 1 of the embodiment of the present application may include: a topology map construction unit 11 , a feature information acquisition unit 12 , a feature fusion unit 13 , and an information prediction unit 14 .

The topology map construction unit 11 is used to construct the road network topology map of the road in the target area and the parking lot topology map of the parking lot according to the road information and parking lot information of the target area;

The feature information acquisition unit 12 is configured to acquire the first feature information of the road network topology map and the second feature of the parking lot topology map according to the historical average vehicle speed information of the road in the target area and the historical parking space occupancy information of the parking lot Information; the first feature information is used to represent the historical average vehicle speed information of the road in the target area, and the second feature information is used to represent the historical parking space occupancy information of the parking lot in the target area;

A feature fusion unit 13, configured to fuse the first feature information with the second feature information through a multi-channel space network, the road network topology map and the parking lot topology map to generate a space fusion feature;

The information prediction unit 14 is used to predict the average vehicle speed information of each road and the parking space occupancy information of each parking lot through the loop gating network and at least two spatial fusion features.

In a feasible implementation manner, the topology map construction unit 11 is specifically used for:

Statize the road information of the target area, determine the first connection relationship between each road in the road information according to the natural connection rules of the road, and construct the road network topology map of the target area according to the first connection relationship; the second A connection relationship is used to indicate whether each road is connected in the topology map;

Count the parking lot information in the target area, determine the second connection relationship between each parking lot in the parking lot information according to the shortest path between the parking lots, and construct the parking lot topology in the target area according to the second connection relationship Figure; the second connection relationship is used to indicate whether each parking lot is connected in the topology map.

In a feasible implementation manner, the feature information acquiring unit 12 is specifically configured to:

Obtain the historical average vehicle speed information of each road in the target area at the target time, generate the average vehicle speed vector corresponding to the road according to the historical average vehicle speed information, and use the average vehicle speed vector as the first feature of the road network topology map at the target time information;

Obtain the historical parking space occupancy information of each parking lot in the target area at the target moment, generate a parking space occupancy vector corresponding to the parking lot according to the historical parking space occupancy information, and use the parking space occupancy vector as the topological map of the parking lot at the target time. Second characteristic information.

In a feasible implementation manner, the feature fusion unit 13 is specifically used for:

Input the adjacency matrix of the road network topology map and the first feature information at the target time into the first channel of the multi-channel space network, and obtain the first spatial feature at the target time through the graph convolutional neural network in the first channel ;

Input the adjacency matrix of the topological map of the parking lot and the second feature information of the target time into the second channel of the multi-channel spatial network, and obtain the second spatial feature at the target time through the graph convolutional neural network in the second channel ;

The first spatial feature and the second spatial feature are fused to generate a spatial fusion feature at the target moment.

Referring to FIG. 7, the information prediction unit 14 of the embodiment of the present application may include: an information generation subunit 141, an information prediction subunit 142;

The information generation subunit 141 is used to input the spatial fusion feature at T ₁ -T _k time into the loop gating network, and generate the state information at each time T ₁ -T _k time and the prediction information of the target area; the The state information is the hidden state at each moment, and is used to generate prediction information, the k is a positive integer greater than 1, and the loop gating network includes k loop gating units;

The information prediction subunit 142 is used to predict the average vehicle speed information of each road and the parking space occupancy information of each parking lot according to the prediction information.

In a feasible implementation manner, the information generating subunit 141 is specifically configured to:

Input the spatial fusion feature at T1 moment into the _first loop gating unit of the loop gating network, and generate the state information _h1 at T1 moment _;

The spatial fusion feature at T2 moment and the state information _h1 at the T1 moment are input into the _second cycle gating unit _of the cycle gating network to generate the state information _h2 at the T2 moment _;

Input the spatial fusion feature at time T _k and the state information h _{k- 1 at time T k-} 1 into the kth loop gating unit of the loop gating network, and generate the state information h _{k at time T k and} the target Forecast information for the area.

In a feasible implementation manner, the information prediction subunit 142 is specifically configured to:

The prediction information includes a first vector and a second vector; the first vector corresponds to the average vehicle speed information of each road, and the second vector corresponds to the parking space occupancy information of each parking lot;

Predicting the average vehicle speed information of each road according to the first vector and the corresponding relationship between each dimension in the first vector and the road;

Predict the parking space occupancy information of each parking lot according to the second vector and the corresponding relationship between each dimension in the second vector and the parking lot.

In this embodiment of the application, the road network topology map of the roads in the target area and the parking lot topology map of the parking lot are constructed according to the road information and parking lot information in the target area, and further based on the history of the roads in the target area The average vehicle speed information and the historical parking space occupancy information of the parking lot, the first feature information of the road network topology map and the second feature information of the parking lot topology map are obtained, and the multi-channel space network is used, and the road network topology map and all The topological map of the parking lot, the first feature information and the second feature information are fused to generate a spatial fusion feature, and finally the average vehicle speed information and each Occupancy information for each parking lot. Using the above method avoids the mutual influence of road traffic flow and parking conditions in areas with complex traffic conditions, which leads to deviations in predicting traffic conditions with a single road traffic flow or parking conditions, and improves the accuracy of road traffic condition prediction .

Please refer to FIG. 8 . FIG. 8 is a schematic structural diagram of a computer device provided by an embodiment of the present application. As shown in FIG. 8 , the computer device 1000 may include: at least one processor 1001 , such as a CPU, at least one network interface 1004 , user interface 1003 , memory 1005 , and at least one communication bus 1002 . Wherein, the communication bus 1002 is used to realize connection and communication between these components. Wherein, the user interface 1003 may include a display screen (Display), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. Optionally, the network interface 1004 may include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 may be a random access memory (Random Access Memory, RAM), or a non-volatile memory (non-volatile memory, NVM), such as at least one disk memory. Optionally, the memory 1005 may also be at least one storage device located away from the aforementioned processor 1001 . As shown in FIG. 8 , the memory 1005 as a computer storage medium may include an operating system, a network communication module, a user interface module, and a data processing application program.

In the computer device 1000 shown in FIG. 8 , the network interface 1004 can provide a network communication function, the user interface 1003 is mainly used to provide an input interface for the user; and the processor 1001 can be used to call the data processing application program stored in the memory 1005 , so as to implement the description of the regional traffic prediction method in any one of the above embodiments corresponding to FIG. 2-FIG.

It should be understood that the computer device 1000 described in the embodiment of the present application can execute the description of the regional traffic prediction method in any one of the embodiments corresponding to Figure 2-6c above, and can also implement the embodiment corresponding to Figure 7 above The description of the forecasting equipment for the regional traffic in , will not be repeated here. In addition, the description of the beneficial effect of adopting the same method will not be repeated here.

In addition, it should be pointed out here that: the embodiment of the present application also provides a computer-readable storage medium, and the computer-readable storage medium stores the computer program executed by the aforementioned regional traffic prediction device, and The computer program includes program instructions. When the processor executes the program instructions, it can execute the description of the regional traffic prediction method in any one of the embodiments corresponding to FIG. 2-FIG. 6c. Therefore, here No further details will be given. In addition, the description of the beneficial effect of adopting the same method will not be repeated here. For the technical details not disclosed in the embodiments of the computer-readable storage medium involved in the present application, please refer to the description of the method embodiments of the present application. As an example, program instructions may be deployed to execute on one computing device, or on multiple computing devices located at one site, or, alternatively, on multiple computing devices distributed across multiple sites and interconnected by a communication network To implement, multiple computing devices distributed in multiple locations and interconnected by a communication network can form a blockchain system.

Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be implemented through computer programs to instruct related hardware. The above programs can be stored in a computer-readable storage medium. During execution, it may include the processes of the embodiments of the above-mentioned methods. Wherein, the above-mentioned computer-readable storage medium may be an area traffic prediction device provided in any one of the foregoing embodiments or an internal storage unit of the above-mentioned device, such as a hard disk or a memory of an electronic device. The computer-readable storage medium may also be an external storage device of the electronic device, such as a plug-in hard disk equipped on the electronic device, a smart memory card (smart media card, SMC), a secure digital (secure digital, SD) card, Flash card (flash card), etc. The above-mentioned computer-readable storage medium may also include a magnetic disk, an optical disk, a read-only memory (read-only memory, ROM) or a random access memory, and the like. Further, the computer-readable storage medium may also include both an internal storage unit of the electronic device and an external storage device. The computer-readable storage medium is used to store the computer program and other programs and quantities required by the electronic device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

The terms "first", "second" and the like in the claims, description and drawings of the present invention are used to distinguish different objects, rather than to describe a specific order. Furthermore, the terms "include" and "have", as well as any variations thereof, are intended to cover a non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units is not limited to the listed steps or units, but optionally also includes unlisted steps or units, or optionally further includes For other steps or units inherent in these processes, methods, products or apparatuses. Reference herein to an "embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the present invention. The presentation of this phrase in various places in the specification are not necessarily all referring to the same embodiment, nor are independent or alternative embodiments mutually exclusive of other embodiments. It is understood explicitly and implicitly by those skilled in the art that the embodiments described herein can be combined with other embodiments. The term "and/or" used in the description of the present invention and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the relationship between hardware and software Interchangeability. In the above description, the composition and steps of each example have been generally described according to their functions. Those skilled in the art may use different methods to implement the described functions for each specific application, but such implementation should not be regarded as exceeding the scope of the present invention.

Each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may physically exist separately, or two or more units may be integrated into one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units.

The above disclosures are only preferred embodiments of the present application, which certainly cannot limit the scope of the present application. Therefore, equivalent changes made according to the claims of the present application still fall within the scope of the present application.

Claims (10)

A method for predicting regional traffic, characterized in that it includes: According to the road information and the parking lot information of the target area, construct the road network topology map of the road in the target area and the parking lot topology map of the parking lot; According to the historical average vehicle speed information of the road in the target area and the historical parking space occupancy information of the parking lot, the first feature information of the road network topology map and the second feature information of the parking lot topology map are obtained; the first feature information It is used to represent the historical average vehicle speed information of the road in the target area, and the second feature information is used to represent the historical parking space occupancy information of the parking lot in the target area; Fusing the first feature information with the second feature information through a multi-channel space network, and the road network topology map and the parking lot topology map to generate a space fusion feature; The average vehicle speed information for each road and the occupancy information for each parking lot are predicted by a recurrent gating network and at least two spatially fused features. The method according to claim 1, characterized in that, according to the road information and parking lot information of the target area, constructing the road network topology map of the road in the target area and the parking lot topology map of the parking lot, comprising: Statize the road information of the target area, determine the first connection relationship between each road in the road information according to the natural connection rules of the road, and construct the road network topology map of the target area according to the first connection relationship; the second A connection relationship is used to indicate whether each road is connected in the topology map; Count the parking lot information in the target area, determine the second connection relationship between each parking lot in the parking lot information according to the shortest path between the parking lots, and construct the parking lot topology in the target area according to the second connection relationship Figure; the second connection relationship is used to indicate whether each parking lot is connected in the topology map. The method according to claim 1, characterized in that, according to the historical average vehicle speed information of the road in the target area and the historical parking space occupancy information of the parking lot, the first characteristic information and the parking information of the topological map of the road network are obtained. The second feature information of the field topology map, including: Obtain the historical average vehicle speed information of each road in the target area at the target time, generate the average vehicle speed vector corresponding to the road according to the historical average vehicle speed information, and use the average vehicle speed vector as the first feature of the road network topology map at the target time information; Obtain the historical parking space occupancy information of each parking lot in the target area at the target moment, generate a parking space occupancy vector corresponding to the parking lot according to the historical parking space occupancy information, and use the parking space occupancy vector as the topological map of the parking lot at the target time. Second feature information. The method according to claim 3, characterized in that the first feature information and the second feature information are fused through the multi-channel space network, the road network topology map and the parking lot topology map , to generate spatial fusion features, including: Input the adjacency matrix of the road network topology map and the first feature information at the target time into the first channel of the multi-channel space network, and obtain the first spatial feature at the target time through the graph convolutional neural network in the first channel ; Input the adjacency matrix of the topological map of the parking lot and the second feature information of the target time into the second channel of the multi-channel spatial network, and obtain the second spatial feature at the target time through the graph convolutional neural network in the second channel ; The first spatial feature and the second spatial feature are fused to generate a spatial fusion feature at the target moment. The method according to claim 1, characterized in that, predicting the average vehicle speed information of each road and the parking space occupancy information of each parking lot through the loop gating network and at least two spatial fusion features, comprising: Input the spatial fusion feature of T ₁ -T _k time into the loop gating network, generate the state information of each time in T ₁ -T _k time and the prediction information of the target area; the state information is the hidden State, used to generate prediction information, the k is a positive integer greater than 1, and the loop gating network includes k loop gating units; The average vehicle speed information of each road and the parking space occupancy information of each parking lot are predicted according to the prediction information. The method according to claim 5, characterized in that the spatial fusion features at T ₁ -T _k time are input into the loop gating network to generate state information at T ₁ -T _k time and prediction information of the target area ,include: Input the spatial fusion feature at T1 moment into the _first loop gating unit of the loop gating network, and generate the state information _h1 at T1 moment _; The spatial fusion feature at T2 moment and the state information _h1 at the T1 moment are input into the _second cycle gating unit _of the cycle gating network to generate the state information _h2 at the T2 moment _; Input the spatial fusion feature at time T _k and the state information h _{k- 1 at time T k-} 1 into the kth loop gating unit of the loop gating network, and generate the state information h _{k at time T k and} the target Forecast information for the area. The method according to claim 5, wherein said predicting the average vehicle speed information of each road and the parking space occupancy information of each parking lot according to said prediction information includes: The prediction information includes a first vector and a second vector; the first vector corresponds to the average vehicle speed information of each road, and the second vector corresponds to the parking space occupancy information of each parking lot; Predicting the average vehicle speed information of each road according to the first vector and the corresponding relationship between each dimension in the first vector and the road; Predict the parking space occupancy information of each parking lot according to the second vector and the corresponding relationship between each dimension in the second vector and the parking lot. A forecasting device for regional traffic, characterized in that it includes: A topology map construction unit, configured to construct a road network topology map of roads in the target area and a parking lot topology map of the parking lot according to the road information and parking lot information in the target area; A feature information acquisition unit, configured to acquire the first feature information of the road network topology map and the second feature information of the parking lot topology map according to the historical average vehicle speed information of the road in the target area and the historical parking space occupancy information of the parking lot ; The first feature information is used to represent the historical average vehicle speed information of the road in the target area, and the second feature information is used to represent the historical parking space occupancy information of the parking lot in the target area; A feature fusion unit, configured to fuse the first feature information with the second feature information through a multi-channel space network, and the road network topology map and the parking lot topology map to generate a space fusion feature; The information prediction unit is used to predict the average vehicle speed information of each road and the parking space occupancy information of each parking lot through the recurrent gating network and at least two spatial fusion features. A computer device, characterized in that it includes: a processor, a memory, and a network interface; The processor is connected to the memory and the network interface, wherein the network interface is used to provide a network communication function, the memory is used to store program codes, and the processor is used to call the program codes to execute The method according to any one of claims 1-7. A computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium, and the computer program is adapted to be loaded by a processor and execute the method according to any one of claims 1-7.

Images