CN112232484B - A method and system for identifying and capturing space debris based on brain-like neural network - Google Patents

Abstract

The invention discloses a method and system for identifying and capturing space debris based on a brain-like neural network, comprising: acquiring image information of space debris; constructing a main network, and optimizing the main network into a target identification network; The distribution complexity of space debris; the network entropy calculation network is introduced to describe the network complexity of the target recognition network; the entropy balance driving factor is obtained according to the network complexity and the distribution complexity of space debris; under the guidance of the game theory framework, the use of entropy The balance driving factor adaptively adjusts the network structure of the target identification network; establishes an information closed loop between the space debris and the space-based detector, and outputs the relative positions of the space-based detector and the space debris in real time until the space debris is captured. The invention has the characteristics of low power consumption, strong generalization ability and high recognition accuracy.

Description

A method and system for identifying and capturing space debris based on brain-like neural network

technical field

The invention relates to the technical field of space debris identification, and more particularly to a method and system for identifying and capturing space debris based on a brain-like neural network.

Background technique

At present, machine learning and brain-like computing methods are usually used to detect space debris. Space debris detection is to detect space debris from the background to facilitate subsequent work. Traditional machine learning has a large demand for data, it relies on a huge amount of training, and requires a huge amount of sample data (hundreds of M, G, and even data of T) to complete the training; and the learning and training process requires accurate feedback , For many problems in the real world, such as aerospace or robotics, there is no good feedback, and there is no way to conduct a large number of simulation experiments to generate a large number of samples for training, or the cost of generating samples is too high, making the application of machine learning. Very difficult. At the same time, the traditional machine learning model has poor generalization ability and high energy consumption; when the system encounters a new situation and obtains new samples, it is often necessary to perform a 0-based model on the data set containing the new samples for the model that has been trained and formed. training, otherwise the already trained model cannot be applied on new samples; a standard computer would consume 250 watts of energy just to recognize 1000 different objects.

The traditional brain-like computing has insufficient analysis of the functional structure of the brain, and the research tools for the brain cannot combine the details and the whole, and cannot perform global imaging of the brain with high spatial and temporal resolution. At the same time, it is difficult to abstract brain function into a mathematical model, the brain map is highly complex and dynamic, and the functional division of brain neurons varies significantly under different spatial scales and spatial distributions. calculations, and cannot meet the real-time requirements.

Therefore, how to provide a method and system for identifying and capturing space debris based on a brain-like neural network with low power consumption, strong generalization ability, and high recognition accuracy is an urgent problem for those skilled in the art to solve.

SUMMARY OF THE INVENTION

In view of this, the present invention provides a method and system for identifying and capturing space debris based on a brain-like neural network. By learning space debris samples, a neural network model sensitive to space debris is obtained, and a space-based detector and debris The relative position between the two planned trajectories to capture space debris. When encountering space debris that may pose a threat, the space debris can be identified, tracked and processed in real time through fast and efficient identification.

In order to achieve the above object, the present invention adopts the following technical solutions:

A method for identifying and capturing space debris based on a brain-like neural network, comprising the following steps:

S1. Obtain image information of space debris observed by space-based detectors;

S2, constructing a main body network, analyzing the connection and disconnection strategy of each network node in the main body network, modeling each network node according to the strategy, and optimizing the main body network as a target identification network;

S3. Introduce an environment entropy calculation network, and use the environment entropy calculation network to describe the distribution complexity of space debris;

S4, introducing a network entropy calculation network, and using the network entropy calculation network to describe the network complexity of the target recognition network;

S5, obtaining an entropy balance driving factor according to the network complexity of the target identification network and the distribution complexity of the space debris;

S6. Introduce a game theory framework, and under the guidance of the game theory framework, use the entropy balance driving factor to adaptively adjust the network structure of the target recognition network, so that the network complexity of the target recognition network and the space debris The distribution complexity matches;

S7. Establish an information closed loop between the space debris and the space-based detector, so that the target recognition network continues to learn, and outputs the relative positions of the space-based detector and the space debris in real time until the space debris is captured.

Preferably, in the above-mentioned method for identifying and capturing space debris based on a brain-like neural network, in S1, an image feature extraction tool is used to extract one or more feature vectors from the image of space debris, and combine them into an overall vector x.

Preferably, in the above-mentioned method for identifying and capturing space debris based on a brain-like neural network, the modeling of each network node in S2 is as follows:

Among them, V represents the set of network nodes; S _i represents the strategy adopted by each node v _i ; p ₁ represents the probability that each network node adopts the "growth"strategy; p ₂ represents the probability that each network node adopts the "fading"strategy; S stands for mixed strategy.

Preferably, in the above-mentioned method for identifying and capturing space debris based on a brain-like neural network, in S2, the network cost and network error of the main network are calculated according to the strategy; optimization is obtained by using the network cost and the network error Objective function, and use the optimization objective function to adjust the connection structure and network accuracy of the main network;

Wherein, the calculation formula of the network cost is:

In the above formula, M _i represents the weight matrix, which is a variable; T represents the matrix transposition; a _ij represents the connection between the i-th node and the j-th node in the connection matrix A; Θ _j represents the relationship between node i and node j. weight; U _i represents the network cost of each node v _i under the current strategy; A' represents the connection matrix of the subject network after modeling; L _{2, t} represents the total network cost of the subject network;

The calculation formula of the network error is:

L _1,t =||X _t -Y _t || ² ;

In the above formula, X _t represents the image feature of the input space debris, Y _t is the output of the main network; L _{1, t} represents the accuracy of the main network;

The calculation formula of the optimization objective function is as follows:

In the above formula, λ _g represents a weight coefficient.

Preferably, in the above-mentioned method for identifying and capturing space debris based on a brain-like neural network, the formula for calculating the complexity of the distribution of space debris in S3 is as follows:

_Hu = f(x);

Among them, f(·) represents the environment entropy calculation network, and H _u represents the environment entropy.

Preferably, in the above-mentioned method for identifying and capturing space debris based on a brain-like neural network, the calculation formula of the network complexity in S4 is:

H _n =g(H _graph );

H_graph表示图形熵，用于计算图论表示下的网络连接复杂度；g(·)表示一个多层感知器，将图形熵H_graph映射到高维特征空间中，得到网络熵H_n。Among them, the network entropy calculation network is regarded as a graph G=(V, E), the node set V={v _i |i=1, 2...k} of the network entropy network, and each node v _i is connected to d _i nodes, and the connection strength is

H _graph represents graph entropy, which is used to calculate the network connection complexity under graph theory representation; g(·) represents a multi-layer perceptron, which maps graph entropy H _graph into high-dimensional feature space to obtain network entropy H _n .

Preferably, in the above-mentioned method for identifying and capturing space debris based on a brain-like neural network, the calculation formula for using the entropy balance driving factor in S5 is:

Q=-λ(H _u -H _n );

Among them, Q is the entropy balance driving factor, which represents the information difference between the environment and the network, which is used to adjust the complexity of the network; λ is the proportional coefficient.

Preferably, in the above-mentioned method for identifying and capturing space debris based on a brain-like neural network, in S6, the entropy balance driving factor is used to adjust the network cost of the target recognition network, and to change the "growth" of the target recognition network with a "fading" strategy to change its network structure.

Preferably, in the above-mentioned method for identifying and capturing space debris based on a brain-like neural network, in S7, the space-based detector includes a controller and a capture device; the target identification network catalogs and locates the space debris, and Real-time output to the controller; the controller controls the capture device to capture the corresponding space debris according to the position information of the space debris; the space-based detector freezes the target identification network during the movement process until After completing the capture task, continue to adjust the structure of the target recognition network.

As can be seen from the above technical solutions, compared with the prior art, the present invention has the following advantages:

1. The present invention evolves into an effective and low-power network by adopting a growing and growing strategy for each node of the main network. Due to the characteristics of network connections, this is a small-world network, and the connection between each node is more efficient. It is sparse, but its function is no less than that of a complex connection or even a fully connected network, and the energy consumption is greatly reduced. It avoids the problem of fixed model and poor generalization ability of traditional machine learning methods.

2. The present invention calculates the environmental entropy and the network entropy respectively by introducing the environmental entropy calculation network and the network entropy calculation network, respectively describes the distribution complexity of the space debris and the network complexity of the target recognition network, and maps the environmental entropy and the network entropy to the same one. In the space, make it have the same physical meaning, and use the entropy difference network to calculate the difference between the two entropies, that is, the entropy balance driving factor. Driven by the entropy balance driving factor, the nodes of the target recognition network and the cost function change, changing the network It can change the network structure and realize the self-adaptive adjustment of the network structure with the change of time and space. At the same time, under the framework of game theory to guide the evolution of the network, the network can continuously learn and change the connection strength through the difference between the external information and the complexity of the network, so that the network can continue to learn, and it can be better for space debris of different shapes. It can greatly improve the accuracy and speed of space debris identification.

The present invention also provides a system for identifying and capturing space debris based on a brain-like neural network, which is suitable for the above-mentioned method for identifying and capturing space debris based on a brain-like neural network, including:

Input module for obtaining image information of space debris observed by space-based detectors

The network optimization module is used to construct the main network and analyze the connection and disconnection strategy of each network node in the main network, model each network node according to the strategy, and optimize the main network as the target identify the network;

an environment entropy calculation module, used for describing the distribution complexity of space debris by using the environment entropy calculation network;

A network entropy calculation module, used to calculate the network complexity of the target identification network by using the network entropy to describe the network

an entropy balance driving factor calculation module, configured to obtain an entropy balance driving factor according to the network complexity of the target identification network and the distribution complexity of the space debris;

The self-adaptive adjustment module is used to adjust the network structure of the target recognition network adaptively by using the entropy balance driving factor under the guidance of the game theory framework, so that the network complexity of the target recognition network is related to the degree of space debris. distribution complexity to match; and

The capture module is used to establish a closed loop of information between the space debris and the space-based detector, so that the target recognition network continues to learn, and outputs the relative positions of the space-based detector and the space debris in real time until the space debris is captured.

It can be known from the above technical solutions that the input of the present invention is the image of the space debris input from the optical camera of the space-based detector, the network optimization module, the environmental entropy calculation module, the network entropy calculation module, the entropy balance driving factor calculation module and the adaptive adjustment module. The module constitutes a brain-like neural network. After the images of space debris pass through the brain-like neural network, the cataloging and positioning of space debris is output, and transmitted to the downstream capture module, that is, the control system, to capture space debris. The controller of the control system is used to adjust the engine parameters of the space-based detector, etc., and control the relative position of the space-based detector and space debris; the controlled object is the capture device, which can be selected from the capture claw or capture net according to actual needs. , and capture the target by adjusting the attitude of the capture device.

Description of drawings

In order to explain the embodiments of the present invention or the technical solutions in the prior art more clearly, the following briefly introduces the accompanying 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 It is an embodiment of the present invention. For those of ordinary skill in the art, other drawings can also be obtained according to the provided drawings without creative work.

1 is a flowchart of a method for identifying and capturing space debris based on a brain-like neural network provided by the present invention;

The accompanying drawing of FIG. 2 is a schematic diagram of the self-adaptive adjustment of the target recognition network provided by the present invention under the guidance of a game theory framework;

3 is a schematic diagram of the optimization process of the main network provided by the present invention;

4 is a structural block diagram of a system for identifying and capturing space debris based on a brain-like neural network provided by the present invention;

FIG. 5 is a working flowchart of a system for identifying and capturing space debris based on a brain-like neural network provided by the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

As shown in Figures 1-2, an embodiment of the present invention discloses a method for identifying and capturing space debris based on a brain-like neural network, which is characterized in that it includes the following steps:

S1. Obtain image information of space debris observed by space-based detectors;

S2. Construct a main network, analyze the connection and disconnection strategy of each network node in the main network, model each network node according to the strategy, and optimize the main network as a target recognition network;

S3. Introduce the environment entropy calculation network, and use the environment entropy calculation network to describe the distribution complexity of space debris;

S4. Introduce the network entropy calculation network, and use the network entropy to calculate the network complexity of the network description target recognition network;

S5. Obtain the entropy balance driving factor according to the network complexity of the target recognition network and the distribution complexity of space debris;

S6. Introduce the game theory framework, under the guidance of the game theory framework, use the entropy balance driving factor to adaptively adjust the network structure of the target recognition network, so that the network complexity of the target recognition network matches the distribution complexity of space debris;

S7. Establish an information closed loop between the space debris and the space-based detector, so that the target recognition network can continue to learn, and output the relative positions of the space-based detector and the space debris in real time until the space debris is captured.

The above steps will be described in detail below.

S1. Obtain image information of space debris observed by space-based detectors:

表示m维的向量空间。In order to better describe the information content of space debris images, it is necessary to convert the acquired images into the form of vectors, that is, to perform feature extraction. In this regard, one or more feature vectors are extracted from the image by using the existing image feature extraction tools, and combined into an overall vector x=(x ₁ , x ₂ , x ₃ . . . x _n ), where x _i represents The i-th eigenvector,

Represents an m-dimensional vector space.

S2. Construct a main network, analyze the connection and disconnection strategy of each network node in the main network, model each network node according to the strategy, and optimize the main network as a target recognition network.

As shown in Figure 3, under the representation of graph theory, the target recognition network is regarded as a graph G=(V, E), and the network nodes are modeled, V={v _i |i=1, 2.. .k}, E represents the set of edges in the graph G, k represents the number of nodes in the graph G, that is, the number of neurons; the strategies that each node can take are "growth" and "fading", the probability is p respectively ₁ , p ₂ , the strategy made by each node i after taking a strategy for the jth node is s _ij :

S _i =(s _i1 , s _i2 , . . . , s _ik ); S _i represents a mixed strategy.

The evolution matrix formed by growth and decline is:

If the connection matrix of the network is A at this time, that is

represents the connection of the network, the new connection matrix is

In order to better describe the effect of the entire network after the network nodes of the main network adopt the strategy under the framework of game theory, it is necessary to calculate the network cost U _i according to the strategy selection. At this time, the network cost U _i is calculated according to the new connection. , assuming that the weight of node i to node j is Θ _j , write it into the weight matrix _Mi = (Θ ₁ , Θ ₂ ,...,Θ _k ):

The weight matrix M _i is a variable variable. By adjusting the weight matrix, it measures the network cost under the current connection strategy and guides the further adjustment of the network structure. T represents the matrix transposition; a _ij represents the connection between the i-th node and the j-th node in the connection matrix A.

The present invention focuses on the adjustment of the network structure, which is a dynamic process, so it is meaningful to calculate the loss function only at a specific time t. Under this problem, we need to calculate the cost of the entire network, namely:

The above formula represents the sum of the node cost of each node after adopting the strategy, that is, the total cost of the network. In specific tasks, it is also necessary to use the network error function to adjust the accuracy of the network, namely:

L _1,t =||X _t -Y _t || ² ;

Among them, X _t is the image feature of the input space debris, Y _t is the output of the main network, and L _{1, t} is used to measure the accuracy of the network.

The optimization objective function is obtained according to the network cost function and the network error function, and its calculation formula is as follows:

Among them, λ _g is a weight coefficient, indicating the proportion of network complexity in the optimization process.

When the main network is working, backpropagation is no longer performed, but the accuracy of the network can still change with the change of the network structure, so by inputting the loss at time t-1 into the network together with the input at the next time, The dual regulation of network accuracy and connection structure is realized, and the final target recognition network is obtained.

As shown in Figure 2, S3, introduce the environment entropy calculation network, and use the environment entropy calculation network to describe the distribution complexity of space debris.

After obtaining the eigenvectors of space debris, input the eigenvectors x=(x ₁ , x ₂ , x ₃ ... x _n ) into the environmental entropy calculation network to calculate the environmental entropy, which is used to describe the complex distribution of space debris degree. The specific calculation method is as follows:

_Hu = f(x)

Among them, f( ) represents the environmental entropy calculation network, which only needs to use a simple multi-layer perceptron; H _u represents the environmental entropy. A multilayer perceptron is a feedforward artificial neural network model that maps multiple input datasets to a single output dataset.

S4, introduce the network entropy calculation network, and use the network entropy to calculate the network complexity of the network description target recognition network.

Using the network entropy calculation network to calculate the network entropy of the target recognition network, the target recognition network can be regarded as a graph G=(V, E). Node set V={v _i |i=1, 2...k}, each node i is connected to d _i nodes, and the connection strength is

H _n =g(H _graph );

In the above formula, H _graph is the graph entropy, which directly calculates the network connection complexity of the target recognition network under the graph theory representation. g(·) is also a multi-layer perceptron, which can map the graph entropy H _graph into a high-dimensional feature space to obtain the network entropy H _n .

S5. According to the network complexity of the target recognition network and the distribution complexity of the space debris, the entropy balance driving factor is obtained.

The information entropy and network entropy after entropy calculation network have the same physical meaning, so they can be subtracted to drive the change of the network structure, and the entropy difference network is used to calculate the entropy balance driving factor.

Q=-λ(H _u -H _n )

Among them, Q is the entropy balance driving factor, which represents the information difference between the environment and the network, and λ is the scaling factor.

S6. Introduce the game theory framework. Under the guidance of the game theory framework, the network structure of the target recognition network is adaptively adjusted by using the entropy balance driving factor, so that the network complexity of the target recognition network matches the distribution complexity of the space debris. Specifically, the connection strategy of the target recognition network is adjusted by using the entropy balance driving factor, that is, the "growth" and "fading" strategies of the target recognition network are changed to change the network structure. The network cost is adjusted by the connection strategy of the target recognition network, and its accuracy needs to be adjusted by calculating the network error. Due to the adjustment of the connection strategy of the network, its network structure will also change. Through the change of the network structure, the network entropy calculation network recalculates the network complexity of the target recognition network, and matches the distribution complexity of the space debris again. The cycle is repeated until the network complexity of the target recognition network matches the distribution complexity of the space debris. Furthermore, the network structure of the target recognition network is adaptively adjusted by using the entropy balance driving factor.

In the following, we will introduce in detail how to use the entropy balance factor to adjust the structure of the target recognition network.

In order to facilitate the explicit time t, the influence of the entropy balance driving factor Q _t on the network structure adjustment of the target recognition network, at time t, U _{i, t} = U _{i, t} (S _{i, t} , Q _t ) is the node v _i in the Cost function under mixed strategy Si _,t . The evolution process of the weight matrix Mi _{, t} (Q _t ) for calculating the network cost is as follows:

Mi _{, t} = Mi _{, t} -1 + f _M (Q _t );

由一个k阶多项式拟合而成：where f _M ( ) is the guiding polynomial that guides the change of the weight matrix.

Fitted by a k-th order polynomial:

In the above formula, γ _i is a coefficient of a polynomial and is a fixed value.

The evolution process of the policy set S _i,t of network nodes with the entropy balance driving factor is as follows, where g _S ( ) is a polynomial function that guides the change of the connection policy.

Si _{, t} = Si _{, t-1} + g _S (Q _t )

The following formula expresses the component form of the connection strategy between node i and node j:

where g _S (Q _t )| _i [j] represents the j-th component of g _S (Q _t )| _i .

有意义，对于g_S(Q_t)|_i[j]需要做出以下的约束：Among them, in order to ensure

Meaningful, for g _S (Q _t )| _i [j] the following constraints need to be made:

Therefore, g _S (Q _t )| _i can be written as:

是一个多项式拟合函数。经过以上的处理，可以看出原本连接强度大的节点继续增强连接会被抑制增幅，保证了网络结构不会趋于固化，灵活可调。通过对熵平衡驱动因子的引入，网络可以实现随着环境的变化调整网络的结构，并且对网络的结构重新赋值，实现自适应调整网络结构的目的。

is a polynomial fitting function. After the above processing, it can be seen that if the nodes with high connection strength continue to strengthen the connection, the increase will be suppressed, which ensures that the network structure will not tend to be solidified and can be adjusted flexibly. By introducing the entropy balance driving factor, the network can adjust the structure of the network with the change of the environment, and reassign the structure of the network to achieve the purpose of adaptively adjusting the network structure.

S7. Establish an information closed loop between the space debris and the space-based detector, so that the target recognition network can continue to learn, and output the relative positions of the space-based detector and the space debris in real time until the space debris is captured. After the adaptive adjustment of the target recognition network,

After realizing a lightweight and effective target recognition network, an information closed loop between space debris and space-based detectors is established, so that the network can continue to learn, and the relative positions of space-based detectors and space debris are output in real time, forming a closed control loop. until the space debris is successfully captured. During this process, as the space-based detector keeps moving and the surrounding environment changes constantly, the target recognition network will continue to receive the images of space debris observed by the space-based detector and continue to learn.

In order to ensure the stability of the recognition during the movement of the space-based detector, the target recognition network needs to be frozen, and the network structure can not be adjusted until the acquisition task is achieved. After obtaining the relative position information and its own attitude information, the space-based detector can perform attitude control and target acquisition through traditional control methods such as Kalman filtering.

Through the above steps, we can know that:

1. The present invention evolves into an effective and low-power network by adopting a growing and growing strategy for each node of the main network. Due to the characteristics of network connections, this is a small-world network, and the connection between each node is more efficient. It is sparse, but its function is no less than that of a complex connection or even a fully connected network, and the energy consumption is greatly reduced. It avoids the problem of fixed model and poor generalization ability of traditional machine learning methods.

2. The present invention calculates the environmental entropy and the network entropy respectively by introducing the environmental entropy calculation network and the network entropy calculation network, respectively describes the distribution complexity of the space debris and the network complexity of the target recognition network, and maps the environmental entropy and the network entropy to the same one. In the space, make it have the same physical meaning, and use the entropy difference network to calculate the difference between the two entropies, that is, the entropy balance driving factor. Driven by the entropy balance driving factor, the nodes of the target recognition network and the cost function change, changing the network It can change the network structure and realize the self-adaptive adjustment of the network structure with the change of time and space. At the same time, under the framework of game theory to guide the evolution of the network, the network can continuously learn and change the connection strength through the difference between the external information and the complexity of the network, so that the network can continue to learn, and it can be better for space debris of different shapes. It can greatly improve the accuracy and speed of space debris identification.

As shown in FIGS. 4-5 , an embodiment of the present invention further provides a system for identifying and capturing space debris based on a brain-like neural network, which is applicable to the above-mentioned method for identifying and capturing space debris based on a brain-like neural network, including:

Input module for obtaining image information of space debris observed by space-based detectors

The network optimization module is used to construct the main network, analyze the connection and disconnection strategy of each network node in the main network, model each network node according to the strategy, and optimize the main network as a target identification network;

The environmental entropy calculation module is used to describe the distribution complexity of space debris by using the environmental entropy calculation network;

The network entropy calculation module is used to use the network entropy to calculate the network complexity of the network description target recognition network

The entropy balance driving factor calculation module is used to obtain the entropy balance driving factor according to the network complexity of the target recognition network and the distribution complexity of space debris;

an adaptive adjustment module for adaptively adjusting the network structure of the target recognition network by using the entropy balance driving factor under the guidance of the game theory framework, so that the network complexity of the target recognition network matches the distribution complexity of the space debris; and

The capture module is used to establish a closed loop of information between space debris and space-based detectors, so that the target recognition network can continue to learn, and output the relative positions of space-based detectors and space debris in real time until the space debris is captured.

The network optimization module, the environment entropy calculation module, the network entropy calculation module, the entropy balance driving factor calculation module and the self-adaptive adjustment module in the present invention form a brain-like neural network, and the input of the brain-like neural network is the optical signal from the space-based detector. The image or video of the space debris input by the camera, after passing through the brain-like neural network, detects and outputs the cataloging and positioning of the space debris, and transmits the location information of the space debris to the downstream capture module. The capture module includes a controller and a capture device. The controller adjusts the space attitude of the capture device according to the position information of the space debris to make it close to the space debris and capture it.

During the movement of the space-based detector, the parameters of the target recognition network are frozen until the acquisition task is completed, and then the structure of the target recognition network is adjusted.

The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same and similar parts between the various embodiments can be referred to each other. As for the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant part can be referred to the description of the method.

The above description of the disclosed embodiments enables any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims (10)

1. A space debris identification and capture method based on a brain-like neural network is characterized by comprising the following steps:

s1, acquiring image information of the space debris observed by the space-based detector;

s2, constructing a main network, analyzing the connection and disconnection strategy of each network node in the main network, modeling each network node according to the strategy, and optimizing the main network into a target identification network;

s3, introducing an environment entropy calculation network, and describing the distribution complexity of the space debris by using the environment entropy calculation network;

s4, introducing a network entropy calculation network, and describing the network complexity of the target identification network by using the network entropy calculation network;

s5, obtaining an entropy balance driving factor according to the network complexity of the target identification network and the distribution complexity of the space debris;

s6, introducing a game theory framework, and under the guidance of the game theory framework, adaptively adjusting the network structure of the target recognition network by using the entropy balance driving factor to enable the network complexity of the target recognition network to be matched with the distribution complexity of the space debris;

the game theory framework guides that:

adjusting the connection strategy of the target identification network by using the entropy balance driving factor so as to change the network structure of the target identification network; through the change of the network structure of the target identification network, the network entropy calculation network recalculates the network complexity of the target identification network, matches the network complexity of the space debris again, and repeats the steps until the network complexity of the target identification network matches the distribution complexity of the space debris;

and S7, establishing an information closed loop between the space debris and the space-based detector, enabling the target recognition network to continuously learn, and outputting the relative positions of the space-based detector and the space debris in real time until the space debris is captured.

2. The method for recognizing and capturing the space debris based on the cranial neural network according to claim 1, wherein in S1, one or more feature vectors are extracted from the image of the space debris by using an image feature extraction tool, and are merged into an overall vector x.

3. The brain-like neural network-based space debris identification capture method according to claim 1, wherein each network node in S2 is modeled as follows:

wherein V represents a set of network nodes; s _i Represents each node v _i The strategy to be adopted; p is a radical of ₁ Representing the probability of each network node adopting a "growth" policy; p is a radical of ₂ Representing the probability of each network node adopting a "fallback" policy; s denotes a mixing strategy.

4. The method for recognizing and capturing the space debris based on the cranial nerve network according to claim 3, wherein in S2, the network cost and the network error of the subject network are calculated according to the strategy; obtaining an optimization objective function by using the network cost and the network error, and adjusting the connection structure and the network precision of the main network by using the optimization objective function;

the calculation formula of the network cost is as follows:

in the above formula, M _i Representing a weight matrix, which is a variable quantity; t represents matrix transposition; a is _ij Representing the connection condition of the ith node and the jth node in the connection matrix A; theta _j Representing the weight of node i to node j; u shape _i Represents each node v _i Network cost under current policy; a' represents a connection matrix of the modeled subject network; l is _2，t Representing a total network cost of the subject network; the lower subscript t denotes time t;

the calculation formula of the network error is as follows:

L _1，t ＝||X _t -Y _t || ² ；

in the above formula, X _t Image features representing input space debris, Y _t Is the output of the host network; l is _1，t Representing the accuracy of the subject network;

the calculation formula of the optimization objective function is as follows:

in the above formula, λ _g Representing a weight coefficient.

5. The method for recognizing and capturing the space debris based on the brain-like neural network as claimed in claim 2, wherein the calculation formula of the distribution complexity of the space debris in the step S3 is as follows:

H _u ＝f(x)；

wherein f (-) represents the ambient entropy calculation network, H _u Representing the ambient entropy.

6. The method for recognizing and capturing the space debris based on the cranial neural network as claimed in claim 5, wherein the calculation formula of the network complexity in S4 is as follows:

H _n ＝g(H _graph )；

wherein, regarding the target identification network as a graph G ═ (V, E), the network computing entropy network node set V ═ { V ═ V } _i I 1, 2.. k, each node v _i And d _i The nodes are connected with each other with the connection strength of

H _graph Representing graph entropy for calculating network connection complexity under graph theory representation; g (-) represents a multilayer perceptron, entropy H of the graph _graph Mapping to a high-dimensional feature space to obtain a network entropy H _n 。

7. The method for recognizing and capturing the space debris based on the brain-like neural network as claimed in claim 6, wherein the calculation formula of the driving factor using entropy balance in S5 is as follows:

Q＝-λ(H _u -H _n )；

wherein Q is an entropy balance driving factor, represents the information difference between the environment and the network, and is used for adjusting the complexity of the network; λ is a proportionality coefficient.

8. The method for spatial debris recognition and capture based on the cranial neural network according to claim 1, wherein the entropy balance driving factor is used to adjust the network cost of the target recognition network in S6, and the strategy of "increase" and "decrease" of the target recognition network is changed to change the network structure of the target recognition network.

9. The brain-like neural network-based space debris identification capture method according to claim 1, wherein in S7, the space-based probe comprises a controller and a capture device; the target identification network catalogs and positions the space debris and outputs the space debris to the controller in real time; the controller controls the capturer to capture the corresponding space debris according to the position information of the space debris; and the space-based detector freezes the target identification network in the moving process until the acquisition task is completed, and then continues to adjust the structure of the target identification network.

10. A brain-like neural network-based space debris identification and capture system, which is suitable for the brain-like neural network-based space debris identification and capture method according to any one of claims 1-9, and comprises:

an input module for obtaining image information of space debris observed by the space-based detector

The network optimization module is used for constructing a main network, analyzing a strategy of connection and disconnection of each network node in the main network, modeling each network node according to the strategy and optimizing the main network into a target identification network;

the environment entropy calculation module is used for calculating the distribution complexity of the network description space debris by utilizing the environment entropy;

a network entropy calculation module for describing the network complexity of the target identification network by using the network entropy calculation network

The entropy balance driving factor calculation module is used for obtaining an entropy balance driving factor according to the network complexity of the target identification network and the distribution complexity of the space debris;

the self-adaptive adjusting module is used for self-adaptively adjusting the network structure of the target identification network by utilizing the entropy balance driving factor under the guidance of the game theory framework so as to enable the network complexity of the target identification network to be matched with the distribution complexity of the space debris; the game theory framework guides that:

adjusting the connection strategy of the target identification network by using the entropy balance driving factor so as to change the network structure of the target identification network; through the change of the network structure of the target identification network, the network entropy calculation network recalculates the network complexity of the target identification network, matches the network complexity of the space debris again, and repeats the steps until the network complexity of the target identification network matches the distribution complexity of the space debris;

and

and the capturing module is used for establishing an information closed loop between the space debris and the space-based detector, enabling the target recognition network to continuously learn, and outputting the relative positions of the space-based detector and the space debris in real time until the space debris is captured.

Images