Artificial Intelligence Automotive Finance Risk Control System Based on Big Data

Abstract

The present invention provides an artificial intelligence automobile financial risk control system based on big data, which relates to the technical field of automobile financial risk monitoring, and aims to achieve more efficient and reliable automobile financial risk control, including: a big data grouping module, used to divide all individual objects into multiple groups; a first data acquisition module, used to collect global automobile financial data based on the groups; a first data analysis module, used to identify local automobile financial risks based on a first neural network; a second data acquisition module, used to collect local automobile financial data for groups judged to have local automobile financial risks; a second data analysis module, used to identify individual automobile financial risks based on a second neural network. The present invention has the advantages of efficient and reliable automobile financial risk control.

Description

Technical Field

[0001] The present invention relates to the technical field of automobile financial risk monitoring, and in particular to an artificial intelligence automobile financial risk control system based on big data. Background Art

[0002] With the development of automobiles, their sales volume is increasing, and automobile financial business is also gradually increasing.

[0003] Auto finance covers credit services, etc., so there will be certain financial risks. In order to ensure the soundness of auto finance, risk monitoring of auto finance is a necessary work link for service providers. The data collected in the field of auto finance are relatively complex in terms of individuals and data types. Existing technologies consume a lot of computing power when processing based on these data. However, reducing the data processing steps will cause problems such as inaccurate data monitoring.

[0004] Therefore, it is necessary to improve automobile financial risk control to achieve more efficient and reliable automobile financial risk control. Summary of the invention

[0005] The purpose of the present invention is to provide an artificial intelligence automobile financial risk control system based on big data, which can achieve more efficient and highly reliable automobile financial risk control.

[0006] The present invention is achieved through the following technical solutions:

[0007] The artificial intelligence automobile financial risk control system based on big data includes:

[0008] A big data grouping module is used to divide all individual objects into multiple groups;

[0009] A first data collection module, used for collecting global automobile financial data based on the groups;

[0010] A first data analysis module, used for identifying local automobile financial risks based on a first neural network;

[0011] A second data collection module is used to collect local automobile financial data for groups judged to have local automobile financial risks;

[0012] The second data analysis module is used to identify individual automobile financial risks based on the second neural network.

[0013] Preferably, the method of dividing all individual subjects into multiple groups is:

[0014] Respectively obtain the total amount and number of installments of the automobile financial loan of the individual object;

[0015] Get the auto finance loan risk parameters of the individual objects respectively:

[0016] ;

[0017] in, For the The auto finance loan risk parameters of the individual object, and Respectively The total amount and number of installments of the automobile finance loan for each individual subject, and is the experience weight, is the total number of individual subjects;

[0018] Get the number of groups ;

[0019] Get the maximum value of the auto finance loan risk parameter of the individual object respectively and minimum value , at the maximum value and minimum value Divide equally between The individual objects falling into the same interval are classified into the same group.

[0020] Preferably, the number of groups to be obtained is The method is:

[0021] ;

[0022] ;

[0023] ;

[0024] in, is the natural exponential function, and These are all intermediate parameters. and are the functions for taking maximum and minimum values ​​respectively, is the ceiling function.

[0025] Preferably, the method for collecting global automobile financial data based on the groups is:

[0026] The total amount of remaining loans, the total amount of theoretical repayments, the total amount of actual repayments, the number of vehicle accidents and the amount of vehicle accident losses of the individual subjects in each group are collected for each month in M ​​consecutive months.

[0027] Preferably, the method for identifying local automobile financial risks based on the first neural network includes:

[0028] Transmitting the global automobile financial data to the input layer;

[0029] Through the feature extraction layer, feature extraction is performed on each group to obtain group feature parameters;

[0030] Performing nonlinear processing of data through at least one hidden layer;

[0031] The output layer outputs whether each group has local automobile financial risk.

[0032] Preferably, the method of extracting features from each group through the feature extraction layer is:

[0033] ;

[0034] ;

[0035] in, and Respectively The first characteristic parameter and the second characteristic parameter of each group, , and Respectively represent The first The total amount of remaining loans, theoretical repayments and actual repayments for the month. and Respectively The first The number of vehicle accidents and the amount of vehicle accident losses in each month, is the intermediate parameter;

[0036] Combining the first characteristic parameter and the second characteristic parameter:

[0037] ;

[0038] in, For the The group characteristic parameters of each group.

[0039] Preferably, the hidden layer adopts a ReLU activation function, and the output layer adopts a Sigmoid activation function.

[0040] Preferably, the method for collecting local automobile financial data for a group determined to have local automobile financial risks is:

[0041] The total amount of remaining loans, the total amount of theoretical repayments, the total amount of actual repayments, the number of vehicle accidents and the amount of vehicle accident losses for each of the individual subjects in the group identified as having local automobile financial risks in each of the consecutive M months are collected.

[0042] Preferably, the second neural network structure is the same as the first neural network structure.

[0043] The technical solution of the present invention has at least the following advantages and beneficial effects:

[0044] The present invention uses big data and artificial intelligence to conduct auto finance risk control, which can conduct a comprehensive risk assessment and identify potential risk factors;

[0045] The present invention processes data in groups, makes a global judgment on advancement and then makes a local judgment. Individuals with different credit amounts face different risks and needs. After grouping, more personalized financial services and management strategies can be provided, while reducing the system computing burden and improving computing efficiency.

[0046] The data grouping of the present invention is more targeted, and can be dynamically divided into different groups according to different data distribution conditions based on the credit amount. The more uneven the data distribution is, the more data groups are provided, which takes into account both the limit on the number of groups and the problem of data density, and helps to improve the reliability of global analysis;

[0047] The local analysis of the present invention can use the same model as the global analysis, further saving training costs and optimizing computing resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A schematic diagram of the principle of an artificial intelligence automobile financial risk control system based on big data provided in Example 1 of the present invention. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0050] Example 1

[0051] This embodiment provides an artificial intelligence automobile financial risk control system based on big data, see Figure 1 ,include:

[0052] A big data grouping module is used to divide all individual objects into multiple groups;

[0053] A first data collection module, used for collecting global automobile financial data based on the groups;

[0054] A first data analysis module, used for identifying local automobile financial risks based on a first neural network;

[0055] A second data collection module is used to collect local automobile financial data for groups judged to have local automobile financial risks;

[0056] The second data analysis module is used to identify individual automobile financial risks based on the second neural network.

[0057] This embodiment first performs global identification by grouping, which is equivalent to treating the data of a group as an individual for judgment. After identifying the risky group, detailed identification and judgment of the individual are performed, which helps to effectively reduce computing power consumption, improve processing efficiency, and ensure the reliability of judgment. It can intelligently and efficiently monitor automobile financial risks and prevent financial problems from occurring. The combination of big data and artificial intelligence provides data support for decision-making and can make more scientific judgments in a complex market environment.

[0058] In this embodiment, the method of dividing all individual objects into multiple groups is:

[0059] Respectively obtain the total amount and number of installments of the automobile financial loan of the individual object;

[0060] Get the auto finance loan risk parameters of the individual objects respectively:

[0061] ;

[0062] in, For the The auto finance loan risk parameters of the individual object, and Respectively The total amount and number of installments of the automobile finance loan for each individual subject, and is the experience weight, is the total number of individual subjects;

[0063] In particular, and The setting can be determined based on the actual application scenario, whether the grouping focuses more on the total loan or the loan per period. and The size of the allocation.

[0064] Get the number of groups ;

[0065] Get the maximum value of the auto finance loan risk parameter of the individual object respectively and minimum value , at the maximum value and minimum value Divide equally between The individual objects falling into the same interval are classified into the same group.

[0066] Furthermore, the number of groups is obtained. The method is:

[0067] ;

[0068] ;

[0069] ;

[0070] in, is the natural exponential function, and These are all intermediate parameters. and are the functions for taking maximum and minimum values ​​respectively, is the ceiling function.

[0071] The number of groups in this embodiment is set by considering the total amount of financial loans and the number of installments of all individual objects, so as to obtain the risk parameters of automobile financial loans, and set the groups according to the uniform distribution of the risk parameters of automobile financial loans. For uniform data, there will be fewer groups, and for unevenly distributed data, the number of groups will be increased in consideration of the situation that a large amount of data will be accumulated locally, so as to prevent problems such as feature loss caused by too much data in one group.

[0072] As a preferred solution, the method for collecting global automobile financial data based on the groups is:

[0073] The total amount of remaining loans, the total amount of theoretical repayments, the total amount of actual repayments, the number of vehicle accidents and the amount of vehicle accident losses of the individual subjects in each group are collected for each month in M ​​consecutive months.

[0074] Next, the method for identifying local automobile financial risks based on the first neural network may include:

[0075] Transmitting the global automobile financial data to the input layer;

[0076] Through the feature extraction layer, feature extraction is performed on each group to obtain group feature parameters;

[0077] Performing nonlinear processing of data through at least one hidden layer;

[0078] The output layer outputs whether each group has local automobile financial risk.

[0079] Specifically, the method of extracting features from each group through the feature extraction layer is preferably:

[0080] ;

[0081] ;

[0082] in, and Respectively The first characteristic parameter and the second characteristic parameter of each group, , and Respectively represent The first The total amount of remaining loans, theoretical repayments and actual repayments for the month. and Respectively The first The number of vehicle accidents and the amount of vehicle accident losses in each month, is the intermediate parameter;

[0083] Combining the first characteristic parameter and the second characteristic parameter:

[0084] ;

[0085] in, For the The group characteristic parameters of each group.

[0086] In specific settings, the hidden layer preferably uses a ReLU activation function, and the output layer can use a Sigmoid activation function for binary classification tasks.

[0087] When performing global identification, it is equivalent to integrating the data of a group together and treating it as an individual for preliminary judgment. The first characteristic parameter and the second characteristic parameter both capture the trend characteristics of the average data of the duration through a sliding window, which is equivalent to obtaining the comprehensive characteristics of data from multiple different continuous periods. The first characteristic parameter can extract the risk of repayment, and the second characteristic parameter can predict the risk caused by failure consumption.

[0088] Then, the method for collecting local automobile financial data for the group determined to have local automobile financial risks is:

[0089] The total amount of remaining loans, the total amount of theoretical repayments, the total amount of actual repayments, the number of vehicle accidents and the amount of vehicle accident losses for each of the individual subjects in the group identified as having local automobile financial risks in each of the consecutive M months are collected.

[0090] In order to reduce the training burden, the second neural network structure is preferably the same as the first neural network.

[0091] Since global recognition is equivalent to treating a group as an individual object, the same neural network can be used to directly capture and judge the features of the individual object data in the same way during local recognition.

[0092] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. The artificial intelligence automobile financial risk control system based on big data is characterized by: include: A big data grouping module is used to divide all individual objects into multiple groups; A first data collection module, used for collecting global automobile financial data based on the groups; A first data analysis module, used for identifying local automobile financial risks based on a first neural network; A second data collection module is used to collect local automobile financial data for the group determined to have local automobile financial risks; A second data analysis module, for identifying individual automobile financial risks based on a second neural network; The method of dividing all individual objects into multiple groups is: Respectively obtain the total amount and number of installments of the automobile financial loan of the individual object; Get the auto finance loan risk parameters of the individual objects respectively: ； in, For the The auto finance loan risk parameters of the individual object, and Respectively The total amount and number of installments of the automobile finance loan for each individual subject, and is the experience weight, is the total number of individual subjects; Get the number of groups ; Get the maximum value of the auto finance loan risk parameter of the individual object respectively and minimum value , at the maximum value and minimum value Divide equally between intervals, the individual subjects falling into the same interval are classified into the same group; The number of groups to obtain The method is: ; ; ； in, is the natural exponential function, and These are all intermediate parameters. and are the functions for taking maximum and minimum values ​​respectively, is the ceiling rounding function; The method for collecting global automobile financial data based on the group is: Collect the total amount of remaining loans, the total amount of theoretical repayments, the total amount of actual repayments, the number of vehicle accidents and the amount of vehicle accident losses of the individual subjects in each group for each of the consecutive M months; The method for identifying local automobile financial risks based on the first neural network includes: Transmitting the global automobile financial data to the input layer; Through the feature extraction layer, feature extraction is performed on each group to obtain group feature parameters; Performing nonlinear processing of data through at least one hidden layer; Through the output layer, each group is outputted to determine whether there is a local automobile financial risk; The method of extracting features from each group through the feature extraction layer is: ; ; in, and Respectively The first characteristic parameter and the second characteristic parameter of each group, , and Respectively represent The first The total amount of the remaining loan, the total amount of the theoretical repayment and the total amount of the actual repayment for the month. and Respectively The first The number of vehicle accidents and the amount of vehicle accident losses in each month, is the intermediate parameter; Combining the first characteristic parameter and the second characteristic parameter: ; in, For the The group characteristic parameters of each group.

2. The big data-based artificial intelligence automobile financial risk control system according to claim 1 is characterized in that: The hidden layer uses the ReLU activation function, and the output layer uses the Sigmoid activation function.

3. The big data-based artificial intelligence automobile financial risk control system according to claim 2 is characterized in that: The method for collecting local automobile finance data for a group determined to have local automobile finance risks is: The total amount of remaining loans, the total amount of theoretical repayments, the total amount of actual repayments, the number of vehicle accidents and the amount of vehicle accident losses for each of the individual subjects in the group identified as having local automobile financial risks in each of the consecutive M months are collected.

4. The big data-based artificial intelligence automobile financial risk control system according to claim 3 is characterized in that: The second neural network structure is the same as the first neural network.

Images