CN106411998A - Prediction method for UBI (Usage-Based Insurance) system based on internet of vehicles big data - Google Patents

Abstract

The invention discloses a UBI system prediction method based on big data technology. The main steps include: Step 1: Apply the intelligent vehicle-mounted terminal OBD to collect and store vehicle driving data and driving behavior information, and perform data processing; Step 2: Analysis and modeling; Step 3: Based on the analysis and modeling, give a reasonable The car insurance prediction scheme, and conduct modular system analysis and processing according to the user's personalized service requirements; Step 4: Based on the analysis and research of driving behavior, give the car insurance prediction model and UBI car insurance pricing strategy. The prediction method of the car networking insurance system based on big data proposed in the present invention, that is, the vehicle insurance system (Usage-Based Insurance, UBI) based on driving behavior, the method is applied in the application of intelligent vehicle terminal OBD, vehicle data collection, driving behavior On the basis of information storage and processing, data analysis and modeling, a reasonable auto insurance prediction scheme is given, and a modular system analysis and processing is carried out for user personalized services.

Description

Prediction method of UBI system based on big data of Internet of Vehicles

technical field

The present invention relates to UBI system research in the era of big data, that is, data sources, data processing, data analysis and forecasting models, etc., and especially provides the auto insurance forecasting model and UBI auto insurance pricing strategy.

Background technique

In 2013, my country's property insurance industry broke through the 100 million yuan mark, an increase of 21.3% over 2009. Despite this, the profitability of the insurance industry is still not satisfactory. Since the traditional motor vehicle insurance only considers the purchase price and type of the vehicle, the vehicle insurance model is extremely single and does not consider the impact of driving behavior on motor vehicle insurance. The user who settles the claim pays the bill, which makes the auto insurance premium setting of the policyholder seriously unreasonable.

In contrast, foreign insurance rates are more flexible, and the insurance rates for unmarried and young people in the United States are the highest (lack of responsibility, prone to vehicle accidents); German beginners have high rates (high probability of accidents); Canadians use cars on weekends more than cars for work Low rate (low probability of accident). Foreign countries are actively promoting UBI (Usage-Based Insurance, UBI, a vehicle insurance system based on driving behavior) insurance, and have achieved certain results. In the future, UBI's car networking insurance model will continue to be promoted and applied.

With the advent of the Internet era and the development of technology globalization, the mobile Internet is constantly penetrating into various fields of society and economy. Similarly, the Internet of Vehicles under the Internet is also penetrating into the auto insurance industry. Therefore, the auto insurance industry based on the Internet of Vehicles has a huge potential. development prospects. Among them, Internet of Vehicles technology and big data technology are the core driving forces for the future development of the insurance industry.

The invention patent with the publication number CN105389864A and titled "A Method for Extracting Automobile UBI Information" discloses a data extraction method based on automobile UBI, which belongs to the field of automobile UBI insurance system, including automobile UBI data collection, collection method, pre-processing Methods, processing flow, cloud computing and data mining, etc. However, this patent only involves the collection of some parameters during the operation of the car by the intelligent vehicle perception terminal, and the research on the UBI system, and does not involve the prediction of auto insurance costs.

Contents of the invention

The purpose of this invention is to solve the problem that the vehicle insurance model is extremely single, without considering the impact of driving behavior on motor vehicle insurance, and propose a UBI system in the era of big data, data sources, data processing, data analysis and prediction models, and carry out various modules. Detailed analysis. Combined with the analysis results of the data, a reasonable auto insurance forecasting model is established.

In order to solve the above problems, the present invention proposes a UBI system prediction method based on the big data of the Internet of Vehicles based on the research on the insurance of the Internet of Vehicles and the UBI system in the era of big data. The specific technical scheme is as follows:

The UBI system prediction method based on big data technology includes the following steps:

Step 1: Apply the intelligent vehicle-mounted terminal OBD to collect and store vehicle driving data and driving behavior information, and perform data processing;

Step 2: Analytical modeling;

Step 3: On the basis of analysis and modeling, a reasonable auto insurance prediction scheme is given, and a modular system analysis and processing is carried out according to the user's personalized service requirements;

Step 4: Based on the analysis and research of driving behavior, the auto insurance prediction model and UBI auto insurance pricing strategy are given.

Further, the data processing described in step 1 includes two parts: data preprocessing and data storage. Data preprocessing can obtain valuable data information for vehicle insurance prediction schemes. Data storage is to filter out UBI by analyzing data related to driving behavior. The data required by the system, and then classify and merge these data, and store them in a distributed database. The collection and storage is to use the Internet of Vehicles, through OBD, GPS and other devices to complete the collection of the vehicle's own status and environmental information data, and through The Internet transmits the collected data to the central processing unit.

Further, the analysis and modeling process in step 2 is to give different insurance rates to different driving behaviors for the data features extracted by the pre-processing.

Beneficial effect:

1. The present invention obtains the UBI system in the Internet of Vehicles big data era by processing big data.

2. Under the action of this system, analyze the big data processing process in detail, and combine the actual data to explain the impact of driving behavior on auto insurance from different angles.

3. According to the results of the analysis, a simple analysis model and reasonable suggestions for the prediction model of auto insurance are given.

Description of drawings

Figure 1 is the car networking model in OBD mode.

Fig. 2 is the UBI system of the Internet of Vehicles big data era of the present invention.

Figure 3 is a graph of the results of the analysis and processing of the collected data.

Fig. 4 is a flowchart of the present invention.

detailed description

The specific embodiment of the present invention will be further described in detail in conjunction with the accompanying drawings. The present invention proposes a UBI system prediction method based on the big data of the Internet of Vehicles, studies the insurance of the Internet of Vehicles, and innovatively proposes a UBI system prediction method in the era of big data. This method comprehensively analyzes the car owner's driving habits, driving mileage, purchase price and vehicle type. The second-generation car insurance that considers driving safety (PHYD, Pay How You Drive) proposes a multi-mode car insurance plan based on the combination of cars and people, breaking the traditional analysis model that only focuses on cars or people. The data analyzed and processed by the present invention are all real driving behavior data collected by the vehicle-mounted terminal OBD.

As shown in Figure 1, the Internet of Vehicles (IOV) completes the collection of the vehicle's own state and environmental information data through OBD, GPS and other devices, and transmits the collected data to the central processing unit through the Internet and processes the data. Analysis and processing, effective supervision and comprehensive service system for vehicles with different needs, to realize intelligent control of vehicles.

On-Board Diagnostics (OBD, On-Board Diagnostics) is the core technology of the Internet of Vehicles. It integrates the vehicle's intelligent perception module, the connection module between the vehicle and the Internet, and the monitoring module of the vehicle system and components (engine, emission control system, etc.) to realize vehicle status. real-time recording and reporting. The car networking system in OBD mode is composed of three main parts: OBD terminal, background system, and mobile phone APP. The car networking model in OBD mode, the built-in sensor of the vehicle has the function of intelligent perception, and the on-board diagnosis OBD controls the local network (CAN) , Controller Aver Network) is connected to the bus to obtain the vehicle status information in the electronic control unit (ECU, Engine Control Unit). This model system is similar to the logical composition of the Internet of Things, consisting of data collection, data analysis and processing, and data reporting.

As shown in Figure 2, the UBI system in the era of big data mainly consists of data sources, data processing, data analysis, and prediction models. The following is a detailed analysis and overview of the big data UBI auto insurance system.

The OBD installed in the motor vehicle monitors the various systems of the vehicle in real time. The application of the Internet of Vehicles realizes the successful connection from the client to the server (Client/Server). The server is the resource center of the entire application system, and the data sent by the client is transmitted to The database server and the client can also access the database. The data source of the present invention is stored in the relational database MySQL and transmitted to the distributed database management system through the data gateway. MySQL has the characteristics of small size, fast speed, and low cost. It is suitable for quickly generating data in vehicle conditions, updating data in the database in a timely manner, removing redundant data information, and reducing the waste of network resources.

Data processing includes data preprocessing and data storage. Data preprocessing can obtain valuable data information for car insurance prediction schemes. By analyzing the data related to driving behavior, the data required by the UBI system proposed by the present invention are screened out, such as the number of times of four emergency (sudden braking, sudden acceleration, sudden deceleration, sharp turn) per day, mileage, travel time, and speeding times and other data, and then classify and merge these data, and store them in the distributed database HBase. HBase is a Hadoop-based project, also known as Hadoop Distributed File System (HDFS, Hadoop Distributed FileSystem). It is a distributed database for unstructured data storage. It uses Zookeeper to manage the cluster. It is divided into Master (the leader in Zookeeper) and multiple regional servers (RS, RegionServer) at the architectural level. The basic structure is as follows. RS is a node in the cluster. Each RS can manage multiple Regions. Each Region can only be served by one RS. HBase needs multiple Regions to store data. HBase defines a certain value for each Region. Range, the data that falls within the specified range will be allocated to the specified Region, so as to distribute the load to each node. This is the process and advantages of distributed storage. YARN (Yet Another Resource Negotiator) is a resource manager for distributed clusters . The MapReduce1 architecture is to execute Map and Reduce tasks on the entire cluster and report the results, but in a large cluster, when the number of cluster nodes exceeds a certain number, cascading failures will occur, and cascading failures will cause serious deterioration of the entire cluster through network flooding . In order to overcome this defect of MapReduce1, the technology of YARN hierarchical cluster management framework can be used to make the cluster shared, scalable and more reliable. The YARN hierarchical structure is that the resource management program ResourceManager transfers various resources to the basic node agent program NodeManager, and the NodeManager starts and monitors basic application program execution and resource management (CPU, memory and other resource allocation).

Spark is a cluster computing system based on memory computing. Its core is Resilient Distributed Datasets (RDD, Resilient Distributed Datasets). All operations of Spark are based on RDD. RDD is a fault-tolerant and parallel data structure. RDD is an unmodifiable A collection of distributed objects. Each RDD is composed of multiple partitions, and each partition can be calculated on different nodes in the cluster at the same time. The partition characteristics and parallel computing capabilities of RDD enable Spark to make better use of scalable hardware resources. If partitioning and persistence are combined, massive amounts of data can be processed more efficiently.

The present invention collects the data of 1,000 vehicles, and analyzes and processes the data information related to driving behavior, such as four times, mileage, maximum instantaneous speed and travel time. As shown in Figure 3, it is a histogram of the data obtained from the four aspects of the driving behavior based on the driving distance per day, the sum of the four emergency times per day, the maximum speed and the latest travel time. Through the analysis of these data, the corresponding The driving behavior processing results provide strong evidence for the UBI car insurance scheme in the era of big data in the present invention.

Data modeling and analysis is to obtain the desired results for the data features extracted by preprocessing. After data extraction, the Spark algorithm is often used. Spark commonly used applications include Spark SQL, Spark Streaming, MLLib, Graph, etc. Spark SQL uses RDD to implement SQL query; Spark Streaming streaming computing provides real-time computing functions; GraphX graph computing framework realizes basic graph computing functions, common graph algorithms and pregel graph programming framework; MLLib machine learning library provides common classification, Clustering, regression, cross-validation and other machine learning algorithms are implemented in parallel, such as naive Bayesian, logistic regression, decision tree, neural network, TFIDF, collaborative filtering and other algorithms, which already exist in ML lib, and only need to bring the data into the call More convenient.

The UBI system proposed by the present invention gives different insurance rates to different driving behaviors and provides personalized value-added services. After big data analysis and processing, the implementation plan of the motor vehicle insurance provided by the system is as follows: set a base total score value for each user every day, such as 100 points, four urgent/total mileage per day/daily speeding times/daily The night driving time is assigned the total score according to 5:2:2:1, that is, 50 points/20 points/20 points/10 points.

The following table 1 is the scoring rule formulated according to the driving behavior. The quality of a person's driving behavior can be judged by the accumulated scores.

Table 1

According to the plan, the cumulative score of one year is recorded as Sum, the number of days of driving is included in the calculation of the score as Day, and the average score is recorded as Avg:

Avg=Sum/Day

In order to prevent malicious fraud, the number of days Day has certain regulations: if Day<100 days, it is regarded as the lowest level; if 100≤Day<250, a certain percentage will be multiplied by 50% on the original Sum; if Day≥250, the original Sum will be used calculate.

According to the Avg analysis, different levels are divided into different grades. Avg≥80 is a five-star customer, 60≤Avg<80 is a four-star customer, 40≤Avg<60 is a three-star customer, 20≤Avg<40 is a two-star customer, and 0≤Avg<20 is a one-star customer. level customers. Customers with different star ratings can bear different car insurance rates. Insurance companies should reward high-quality customers (ie customers with high star ratings) and give preferential activities in the insurance for the next year. ), which can increase the insurance premium rate for vehicles insured in the coming year. In addition, the acquired data can also provide customers with personalized services, such as timely recommendation of local features and store activity information based on driving habits and frequented places, and timely reminders for users with bad driving behavior.

Fig. 4 is a flow chart of the UBI system prediction method based on the big data technology of the present invention, from which it can be seen that the prediction method of the present invention is first based on the technical framework of the big data of the Internet of Vehicles, using the intelligent vehicle-mounted terminal OBD to monitor the vehicle driving Collect and store data and driving behavior information, and perform data processing; then analyze and model, and give a reasonable car insurance prediction plan based on the analysis and model, and conduct modular system analysis based on the user's personalized service requirements and processing. Finally, based on the analysis and research of driving behavior, the auto insurance prediction model and UBI auto insurance pricing strategy are given.

Claims (3)

1. Based on the UBI system prediction method under the big data technology, it is characterized in that, comprising the steps: Step 1: Apply the intelligent vehicle-mounted terminal OBD to collect and store vehicle driving data and driving behavior information, and perform data processing; Step 2: Analytical modeling; Step 3: On the basis of analysis and modeling, a reasonable auto insurance prediction scheme is given, and a modular system analysis and processing is carried out according to the user's personalized service requirements; Step 4: Based on the analysis and research of driving behavior, the auto insurance prediction model and UBI auto insurance pricing strategy are given. 2. The UBI system prediction method based on big data technology according to claim 1, wherein the data processing in step 1 includes two parts, data preprocessing and data storage, and the data preprocessing can obtain the prediction of vehicle insurance The solution provides valuable data information. Data storage is to filter out the data required by the UBI system by analyzing the data related to driving behavior, and then classify and merge these data, and store them in a distributed database. Networking, through OBD, GPS and other devices, complete the collection of the vehicle's own status and environmental information data, and transmit the collected data to the central processor through the Internet. 3. The UBI system prediction method based on big data technology according to claim 1, characterized in that, the analysis and modeling process described in step 2 is aimed at the data features extracted by preprocessing, and different driving behaviors are given different driving behaviors. insurance rate.