A multi-level model method and system for early warning of violation risks in the transportation industry

Abstract

The present invention relates to a multi-level model method and system for early warning of violation risks in the transportation industry. With transportation enterprises as the supervision object, the risk behavior text and the risk level corresponding to the risk behavior text are extracted by obtaining historical expert experience documents on the handling of traffic accidents in the transportation industry, and the correlation between the two is calculated, so that it can classify the risk level of new illegal behavior descriptions; construct an enterprise portrait index system, extract enterprise characteristics, use the enterprise characteristics as input, and output a violation level training early warning model. The early warning model is used to predict the level and probability of risks generated by transportation enterprises in the future. Taking into account the differences of the enterprises themselves, the present invention conducts in-depth mining of the basic information and historical violation records of transportation enterprises, constructs a comprehensive enterprise portrait index system, and uses the portrait index as the input training of the model, further improving the prediction ability of the model to realize intelligent monitoring of the enterprise.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a multi-level model method and system for early warning of violation risks in the transportation industry. Background Art

[0002] At present, most of the enterprise risk data comes from the description of the punishment for illegal and irregular behaviors in the fields such as the internal punishment data of the enterprise. It is not easy to know which punishments have a higher level of risk. In addition, there is a lack of a large number of models with high accuracy in the industry. The traditional enterprise risk warning model uses some qualitative expert experience rules to establish the model, that is, to give a given weight to the indicators of the enterprise, sum up to get the enterprise score, and determine the risk level for the enterprise according to the score. Due to the huge amount of data, the workload of manually integrating data, analyzing and summarizing, and obtaining results is large. The manual screening task is heavy and inefficient, and it is impossible to screen the information comprehensively. At the same time, the expert experience model has a large subjective factor and cannot unify the judgment standard, so the accuracy is low. The lack of analysis of the enterprise's own data makes the applicability poor. Summary of the invention

[0003] In order to address the deficiencies of the prior art, the present invention proposes a multi-level model method and system for early warning of violation risks in the transportation industry. By obtaining historical expert experience documents on the handling of traffic accidents in the transportation industry, extracting risk behavior texts and risk levels corresponding to the risk behavior texts, and calculating the correlation between the two, it is possible to classify new illegal behavior descriptions into risk levels; construct an enterprise portrait index system, extract enterprise characteristics, use enterprise characteristics as input, and output a violation level training early warning model. The early warning model is used to predict the level and probability of risks that transportation companies will generate in the future. Taking into account the differences among companies themselves, the present invention conducts in-depth mining of the basic information and historical violation records of transportation companies, constructs a comprehensive enterprise portrait index system, and uses portrait indicators as input training for the model, further improving the model's predictive ability.

[0004] To achieve the above objectives, the technical solutions adopted by the present invention include:

[0005] A multi-level model method for early warning of traffic industry violation risks, characterized by comprising:

[0006] Obtain historical expert experience documents on traffic accident handling in the transportation industry; the historical expert experience documents include risk behavior texts and risk levels corresponding to the risk behavior texts;

[0007] Obtain a description of the offense;

[0008] Segment the risk behavior text and illegal behavior description into words and remove duplicates to obtain risk words and illegal behavior words, and generate risk word text and illegal behavior word text;

[0009] Mapping each of the risk words into a risk word vector using one-hot encoding;

[0010] Get the risk word vectors included in the risk behavior text, and generate a risk word vector set X = {x1, x2, x3, ..., x n};

[0011] Use one-hot encoding to map each of the illegal behavior words into an illegal behavior word vector, generating an illegal behavior word vector set Z = {z1, z2, z3, ..., z n};

[0012] Use one-hot encoding to map the risk levels corresponding to the risk behavior texts into risk level word vectors, respectively, to generate a risk level word vector set Y = {y1, y2, y3, y4};

[0013] Calculate the weighted average h of the risk word vector i ,

[0014]

[0015] Where W is the first initial weight matrix, n is the number of risk word vectors;

[0016] Calculate each risk word vector u i ,

[0017]

[0018] Where W' is the second initial weight matrix, is the jth column of the matrix W';

[0019] Calculate the risk level word vector y i with u i The loss is obtained by modifying the first initial weight matrix W and the second initial weight matrix W' until convergence, and obtaining the first weight matrix Q and the second weight matrix Q';

[0020] Calculate the weighted average ki of the illegal behavior word vector,

[0021]

[0022] Where n is the number of illegal behavior word vectors;

[0023] Calculate the word vector p for each illegal behavior i ,

[0024]

[0025] in, is the jth column of the matrix Q';

[0026] The p i After normalization, we can get the risk level probability yi corresponding to the illegal behavior word vector.

[0027]

[0028] Output the illegal behavior word vector Z with the largest probability value i The risk level corresponding to the illegal behavior word;

[0029] Constructing a risk profile indicator system for transportation enterprises and marking the risk levels of transportation enterprises, and extracting enterprise characteristics from the risk profile indicator system for transportation enterprises;

[0030] The enterprise characteristics are input into the early warning model, and the early warning model outputs the enterprise violation risk prediction result, wherein the early warning model is trained with the enterprise risk portrait indicator system as a sample.

[0031] Furthermore, the risk levels include A, B, C and D.

[0032] Furthermore, the segmentation method includes an N-gram algorithm and a Jieba word segmenter;

[0033] Further, the value of N in the N-gram algorithm is 2;

[0034] Further, the risk word vector, illegal behavior word vector and risk level word vector are binary word vectors;

[0035] Furthermore, the risk profile indicator system of the transportation enterprise includes the type of organization, operating period, business scope type, number of inspection behaviors, number of historical changes in registered address and / or business address, number of inspection forms, frequency of inspection dates, and enterprise violation level and reporting time;

[0036] The institution category is the second digit of the “unified social credit code”;

[0037] The operating period is the period from the operating end date to the operating start date;

[0038] The types of business scope described include non-farmers' professional cooperatives and farmers' professional cooperatives and their branches;

[0039] The number of inspection behaviors includes the number of inspection behaviors without duplicates, the number of inspection behaviors without duplicates in the past three months, the number of inspection behaviors without duplicates in the past six months, the number of inspection behaviors without duplicates in the past year, and the number of inspection behaviors without duplicates in the past three years;

[0040] The number of historical changes in the registered address and / or business address includes the number of addresses without duplicates in the past three months, the number of addresses without duplicates in the past six months, the number of addresses without duplicates in the past year, and the number of addresses without duplicates in the past three years;

[0041] The number of inspection form types includes the number of inspection forms without duplicates and the number of inspection types without duplicates; the number of inspection forms without duplicates includes the number of inspection forms without duplicates in the past three months, the number of inspection forms without duplicates in the past six months, the number of inspection forms without duplicates in the past year, and the number of inspection forms without duplicates in the past three years; the number of inspection types without duplicates includes the number of inspection types without duplicates in the past three months, the number of inspection types without duplicates in the past six months, the number of inspection types without duplicates in the past year, and the number of inspection types without duplicates in the past three years;

[0042] The inspection date frequency includes the time interval from the most recent assessment date to the present, the time interval from the most recent three months assessment date to the present, the time interval from the most recent six months assessment date to the present, the time interval from the most recent one year assessment date to the present, and the time interval from the most recent three years assessment date to the present;

[0043] The enterprise violation level and reporting time include the number of risk level A, B, C and D behaviors that occurred in the past 7 days, 14 days, 30 days, 60 days, 120 days and 180 days.

[0044] Furthermore, the marking of the risk level of transportation enterprises includes marking enterprises that have engaged in Class A risk behavior in the past year as high-risk enterprises, that is, y=1; marking enterprises that have engaged in Class B risk behavior in the past year as medium-high risk enterprises, that is, y=2; marking enterprises that have engaged in Class C risk behavior in the past year as medium-risk enterprises, that is, y=3; marking enterprises that have engaged in Class D risk behavior in the past year as medium-low risk enterprises, that is, y=4; otherwise, the enterprise is marked as y=0.

[0045] Furthermore, the enterprise characteristics include organizational type, operating period and business scope type.

[0046] Furthermore, the early warning model is obtained by training with the lightGBM algorithm.

[0047] The present invention also relates to a multi-level model system for early warning of traffic industry violation risks, characterized in that it includes:

[0048] The data acquisition module is used to obtain historical expert experience documents on traffic accident handling in the transportation industry, descriptions of illegal behaviors, and risk profile indicators of transportation enterprises;

[0049] A processing module is used to segment and remove duplicate words from risk behavior text and illegal behavior description, and map them and risk levels into binary word vectors respectively;

[0050] A feature extraction module, used to extract enterprise features from the risk profile indicator system of the transportation enterprise;

[0051] The prediction module is used to calculate the risk level corresponding to the illegal behavior words and use the lightGBM model to predict the probability of illegal behavior by transportation companies in the next year.

[0052] The present invention also relates to a computer-readable storage medium, characterized in that a computer program is stored on the storage medium, and the computer program implements the above method when executed by a processor.

[0053] The present invention also relates to an electronic device, characterized in that it comprises a processor and a memory;

[0054] The memory is used to store historical expert experience documents on traffic accident handling in the transportation industry, descriptions of illegal behaviors, and risk profile indicators of transportation enterprises;

[0055] The processor is used to execute the above method by calling historical expert experience documents, illegal behavior descriptions and transportation enterprise risk profile indicators, using the lightGBM model.

[0056] The beneficial effects of the present invention are:

[0057] The multi-level model method and system for warning of violation risks in the transportation industry described in the present invention are adopted. Transportation enterprises are taken as the supervision objects. Through in-depth mining of the basic information and historical violation records of transportation enterprises, a comprehensive enterprise portrait training lightGBM model is constructed to improve the accuracy of supervision.

[0058] In order to further achieve differentiated supervision, the present invention proposes to calculate the risks of enterprises with higher levels and the probability of generating risks, and extract the relationship between the description of traffic risk behavior and the corresponding risk level determined by expert experience, so that it can perform risk classification on new illegal and irregular behavior descriptions. In this way, the level and probability of future risks of traffic enterprises can be predicted in the model, achieving the purpose of differentiated supervision. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 This is a flow chart of a multi-level model method for early warning of violation risks in the transportation industry according to the present invention.

[0060] Figure 2 This is a schematic diagram of the structure of a multi-level model system for warning of violation risks in the transportation industry according to the present invention. DETAILED DESCRIPTION

[0061] In order to more clearly understand the content of the present invention, it will be described in detail with reference to the drawings and embodiments.

[0062] The first aspect of the present invention relates to a process of steps as follows Figure 1 The multi-level model approach shown to warn of traffic industry violation risks includes:

[0063] Obtain historical expert experience documents on traffic accident handling in the transportation industry; the historical expert experience documents include risk behavior texts and risk levels corresponding to the risk behavior texts, and the risk levels include A, B, C, and D. As shown in Table 1,

[0064]

[0065] Table 1

[0066] Obtain the description of the illegal behavior. The illegal behavior here is the classified illegal behavior that has not been recorded in the above-mentioned historical expert experience documents. It can be a new illegal behavior that occurred in a transportation company within a certain period of time in the past, such as overloading of passengers in carpooling and chartering, and illegal parking on the road.

[0067] The risk behavior text and illegal behavior description are segmented and deduplicated. The segmentation method includes N-gram algorithm and Jieba word segmenter, where the value of N is 2; and all risk words are deduplicated and filtered. For example, "drivers overload, speeding, answering and making phone calls while driving, privately changing the operating route and other illegal driving behaviors" is segmented and deduplicated and filtered into [speeding, driving, process, answering and making phone calls, mobile phone, privately changing, operation, route, violation, behavior];

[0068] Obtain risk vocabulary and illegal behavior vocabulary, and generate risk vocabulary text and illegal behavior vocabulary text;

[0069] Use one-hot encoding to map each of the risk words into a binary risk word vector. For example, after segmentation, there are 100 risk words in total, and the binary risk word vector of the 11th risk word is (1011).

[0070] Get the risk word vectors included in the risk behavior text, and generate a risk word vector set X = {x1, x2, x3, ..., x n};

[0071] Use one-hot encoding to map each of the illegal behavior words into an illegal behavior word vector, generating an illegal behavior word vector set Z = {z1, z2, z3, ..., z n}

[0072] Use one-hot encoding to map the risk levels A, B, C and D corresponding to the risk behavior text to risk level word vectors, that is, to (1,0,0,0), (0,1,0,0), (0,0,1,0) and (0,0,0,1), respectively, to generate a risk level word vector set Y = {y1, y2, y3, y4};

[0073] Calculate the weighted average h of the risk word vector i ,

[0074]

[0075] Where W is the first initial weight matrix, n is the number of risk word vectors;

[0076] Calculate each risk word vector u i ,

[0077]

[0078] Where W' is the second initial weight matrix, is the jth column of the matrix W';

[0079] Calculate the risk level word vector y i with u i The loss is obtained by modifying the first initial weight matrix W and the second initial weight matrix W' until convergence, and obtaining the first weight matrix Q and the second weight matrix Q';

[0080] Calculate the weighted average ki of the illegal behavior word vector,

[0081]

[0082] Where n is the number of illegal behavior word vectors;

[0083] Calculate the word vector p for each illegal behavior i ,

[0084]

[0085] in, is the jth column of the matrix Q';

[0086] The p i Normalize it to get the risk level probability y corresponding to the illegal behavior word vector i ,

[0087]

[0088] Output the illegal behavior word vector Z with the largest probability value i The risk level corresponding to the illegal behavior word;

[0089] Construct a risk profile index system for transportation enterprises and mark the risk levels of transportation enterprises, and extract enterprise characteristics from the risk profile index system for transportation enterprises; specifically, the risk profile index system for transportation enterprises includes institutional type, operating period, business scope type, number of inspection behaviors, number of historical changes in registered addresses and / or business addresses, number of inspection forms, frequency of inspection dates, and enterprise violation levels and reporting times; extract enterprise characteristics from the above indicators, including institutional type, operating period, and business scope type.

[0090] The institution category is the second digit in the "unified social credit code", the operating period is the period from the end date of the operation to the start date of the operation, and the business scope types include non-farmer professional cooperatives and farmers professional cooperatives and their branches.

[0091] In order to further enrich the data dimensions and increase the amount of sample data, the frequency of changes in registration items is counted in different time periods. The number of historical changes in registered addresses and / or business addresses includes the number of addresses without duplicates in the past three months, the number of addresses without duplicates in the past six months, the number of addresses without duplicates in the past year, and the number of addresses without duplicates in the past three years. The number of types of inspection forms includes the number of inspection forms without duplicates and the number of inspection types without duplicates. The number of inspection forms without duplicates includes the number of inspection forms without duplicates in the past three months, the number of inspection forms without duplicates in the past six months, the number of inspection forms without duplicates in the past year, and the number of inspection forms without duplicates in the past three years. The number of inspection types without duplicates includes the number of inspection types without duplicates in the past three months, the number of inspection types without duplicates in the past six months, the number of inspection types without duplicates in the past year, and the number of inspection types without duplicates in the past three years. The frequency of inspection dates includes the time interval from the most recent assessment date to the present, the time interval from the most recent three months assessment date to the present, the time interval from the most recent six months assessment date to the present, the time interval from the most recent one year assessment date to the present, and the time interval from the most recent three years assessment date to the present. The enterprise violation level and reporting time include the number of risk level A, B, C and D behaviors in the past 7 days, 14 days, 30 days, 60 days, 120 days and 180 days.

[0092] The enterprise characteristics are input into the early warning model, and the early warning model outputs the enterprise violation risk prediction result, wherein the early warning model is trained with the enterprise risk portrait indicator system as a sample.

[0093] Enterprises that have engaged in risk-level A behaviors in the past year are marked as high-risk enterprises, that is, y=1; enterprises that have engaged in risk-level B behaviors in the past year are marked as medium-high risk enterprises, that is, y=2; enterprises that have engaged in risk-level C behaviors in the past year are marked as medium-risk enterprises, that is, y=3; enterprises that have engaged in risk-level D behaviors in the past year are marked as medium-low risk enterprises, that is, y=4; otherwise, the enterprise is marked as y=0.

[0094] Specifically, the data is divided into a training set and a test set. For example, among the more than 500,000 pieces of data, there are 234,789 companies that have illegal or irregular behaviors in the next year, accounting for about 0.278. The sample data can be divided into a training set and a test set in a ratio of 7:3. The test set data will be used to evaluate the model effect in the end, and the training set data can be cross-validated with 5-fold cross validation to obtain the best model parameters.

[0095] The early warning model is obtained through the lightGBM algorithm training. The above-mentioned enterprise risk prediction model can be obtained by training the machine learning component through sample data. In order to make the model highly interpretable and effective, the model learned by the lightGBM algorithm is selected as the above-mentioned enterprise risk prediction model. The enterprise risk prediction model can be continuously optimized through continuous data parameter adjustment during use.

[0096] Finally, the portrait indicators are input into the above-mentioned early warning model, and the early warning model outputs the level and probability of risks that transportation companies will encounter in the future (for example, within the next year or half a year).

[0097] Another aspect of the present invention also relates to a multi-level model system for early warning of traffic industry violation risks, the structure of which is as follows: Figure 2 As shown, including:

[0098] The data acquisition module is used to obtain historical expert experience documents on traffic accident handling in the transportation industry, descriptions of illegal behaviors, and risk profile indicators of transportation enterprises;

[0099] A processing module is used to segment and remove duplicate words from risk behavior text and illegal behavior description, and map them and risk levels into binary word vectors respectively;

[0100] A feature extraction module, used to extract enterprise features from the risk profile indicator system of the transportation enterprise;

[0101] The prediction module is used to calculate the risk level corresponding to the illegal behavior words and use the lightGBM model to predict the probability of illegal behavior by transportation companies in the next year.

[0102] By using this system, the above-mentioned operation processing method can be executed and the corresponding technical effects can be achieved.

[0103] An embodiment of the present invention also provides a computer-readable storage medium capable of implementing all the steps in the multi-level model method for police traffic industry violation risks in the above-mentioned embodiment, and a computer program is stored on the computer-readable storage medium, which, when executed by a processor, implements all the steps in the multi-level model method for police traffic industry violation risks in the above-mentioned embodiment.

[0104] An embodiment of the present invention also provides an electronic device for executing the above method. As an implementation device of the method, the electronic device has at least a processor and a memory. In particular, the memory stores data and related computer programs required for the execution of the method, such as historical expert experience documents on traffic accident handling in the transportation industry, descriptions of illegal behaviors, and risk profile indicators for transportation enterprises. The processor calls the data and programs in the memory to execute all steps of the implementation method and obtains the corresponding technical effects.

[0105] Preferably, the electronic device may include a bus architecture, the bus may include any number of interconnected buses and bridges, and the bus will include various circuits linked together by one or more processors and memories. The bus may also link together various other circuits such as peripherals, voltage regulators, and power management circuits, which are all well known in the art and are therefore not further described herein. The bus interface provides an interface between the bus and the receiver and transmitter. The receiver and transmitter may be the same component, i.e., a transceiver, providing a unit for communicating with various other systems over a transmission medium. The processor is responsible for managing the bus and general processing, while the memory may be used to store data used by the processor when performing operations.

[0106] In addition, the electronic device may further include components such as a communication module, an input unit, an audio processor, a display, and a power supply. The processor (or controller, operation control) used may include a microprocessor or other processor device and / or a logic device, which receives input and controls the operation of various components of the electronic device; the memory may be one or more of a cache, a flash memory, a hard drive, a removable medium, a volatile memory, a non-volatile memory, or other suitable devices, which may store the above-mentioned related data information, and may also store a program for executing related information, and the processor may execute the program stored in the memory to realize information storage or processing, etc.; the input unit is used to provide input to the processor, for example, it may be a key or a touch input device; the power supply is used to provide power to the electronic device; the display is used to display display objects such as images and text, for example, it may be an LCD display. The communication module is a transmitter / receiver that sends and receives signals via an antenna. The communication module (transmitter / receiver) is coupled to the processor to provide input signals and receive output signals, which may be the same as the case of a conventional mobile communication terminal. Based on different communication technologies, multiple communication modules may be provided in the same electronic device, such as a cellular network module, a Bluetooth module and / or a wireless local area network module, etc. The communication module (transmitter / receiver) is also coupled to a speaker and a microphone via an audio processor to provide an audio output via the speaker and receive an audio input from the microphone, thereby realizing a common telecommunication function. The audio processor may include any suitable buffer, decoder, amplifier, etc. In addition, the audio processor is also coupled to a central processor, so that recording can be performed on the machine through the microphone, and the sound stored on the machine can be played through the speaker.

[0107] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0108] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A system that specifies the functions of a box or multiple boxes.

[0109] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a product including an instruction system, which is implemented in the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0110] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps of the functions specified in a box or multiple boxes. Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they know the basic creative concept. Therefore, the attached claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0111] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

Claims

1. A multi-level model method for early warning of traffic industry violation risks, characterized in that: include: Access to historical expert experience documentation on traffic accident handling in the transportation industry; The historical expert experience document includes risk behavior text and risk level corresponding to the risk behavior text; Obtain a description of the offense; Segment the risk behavior text and illegal behavior description into words and remove duplicates to obtain risk words and illegal behavior words, and generate risk word text and illegal behavior word text; Mapping each of the risk words into a risk word vector using one-hot encoding; Get the risk word vectors included in the risk behavior text, and generate a risk word vector set X = {x1, x2, x3, ..., x i ,...,x n }; Use one-hot encoding to map each of the illegal behavior words into an illegal behavior word vector, generating an illegal behavior word vector set Z = {z1, z2, z3, ..., z i ,...,z n }; Use one-hot encoding to map the risk levels corresponding to the risk behavior texts into risk level word vectors, respectively, to generate a risk level word vector set Y = {y1, y2, y3, y4}; Calculate the weighted average h of the risk word vector i , Where W is the first initial weight matrix, n is the number of risk word vectors; Calculate each risk word vector u i , Where W' is the second initial weight matrix, is the jth column of the matrix W'; Calculate the risk level word vector y i with u i The loss is obtained by modifying the first initial weight matrix W and the second initial weight matrix W' until convergence, and obtaining the first weight matrix Q and the second weight matrix Q'; Calculate the weighted average k of the illegal behavior word vectors i , Where n is the number of illegal behavior word vectors; Calculate the word vector p for each illegal behavior i , in, is the j-th column of the matrix Q′; The p i Normalize it to get the risk level probability y corresponding to the illegal behavior word vector i , Output the illegal behavior word vector Z with the largest probability value i The risk level corresponding to the illegal behavior word; Constructing a risk profile indicator system for transportation enterprises and marking the risk levels of transportation enterprises, and extracting enterprise characteristics from the risk profile indicator system for transportation enterprises; The enterprise characteristics are input into the early warning model, and the early warning model outputs the enterprise violation risk prediction result, wherein the early warning model is trained with the enterprise risk portrait indicator system as a sample.

2. The method according to claim 1, characterized in that The risk levels include A, B, C and D. The risk level A is illegal driving behavior, the risk level B is inadequate safety work, the risk level C is the existence of an absolute speed difference, and the risk level D is a large difference between the inner and outer wheels.

3. The method according to claim 2, characterized in that The segmentation includes N-gram algorithm and Jieba word segmenter.

4. The method according to claim 3, characterized in that The value of N in the N-gram algorithm is 2.

5. The method according to claim 4, characterized in that The risk word vector, illegal behavior word vector and risk level word vector are binary word vectors.

6. The method according to claim 5, characterized in that The risk profile indicator system for transportation enterprises includes the type of organization, operating period, business scope type, number of inspections, number of historical changes in registered address and / or business address, number of inspection forms, frequency of inspection dates, and enterprise violation level and reporting time; The institution category is the second digit of the "unified social credit code"; The operating period is the period from the operating end date to the operating start date; The types of business scope described include non-farmers' professional cooperatives and farmers' professional cooperatives and their branches; The number of inspection behaviors includes the number of inspection behaviors without duplicates, the number of inspection behaviors without duplicates in the past three months, the number of inspection behaviors without duplicates in the past six months, the number of inspection behaviors without duplicates in the past year, and the number of inspection behaviors without duplicates in the past three years; The number of historical changes in the registered address and / or business address includes the number of addresses without duplicates in the past three months, the number of addresses without duplicates in the past six months, the number of addresses without duplicates in the past year, and the number of addresses without duplicates in the past three years; The number of inspection form types includes the number of inspection forms without duplicates and the number of inspection types without duplicates; the number of inspection forms without duplicates includes the number of inspection forms without duplicates in the past three months, the number of inspection forms without duplicates in the past six months, the number of inspection forms without duplicates in the past year, and the number of inspection forms without duplicates in the past three years; the number of inspection types without duplicates includes the number of inspection types without duplicates in the past three months, the number of inspection types without duplicates in the past six months, the number of inspection types without duplicates in the past year, and the number of inspection types without duplicates in the past three years; The inspection date frequency includes the time interval from the most recent assessment date to the present, the time interval from the most recent three months assessment date to the present, the time interval from the most recent six months assessment date to the present, the time interval from the most recent one year assessment date to the present, and the time interval from the most recent three years assessment date to the present; The enterprise violation level and reporting time include the number of risk level A, B, C and D behaviors that occurred in the past 7 days, 14 days, 30 days, 60 days, 120 days and 180 days.

7. The method according to claim 6, characterized in that The marking of the risk level of the transportation enterprise includes marking the enterprise that has committed Class A risk behavior in the past year as a high-risk enterprise, that is, y=1; Enterprises that have engaged in risk-level B behaviors in the past year are marked as medium-high risk enterprises, that is, y=2; enterprises that have engaged in risk-level C behaviors in the past year are marked as medium-risk enterprises, that is, y=3; enterprises that have engaged in risk-level D behaviors in the past year are marked as medium-low risk enterprises, that is, y=4; otherwise, the enterprise is marked as y=0.

8. The method according to claim 7, characterized in that The enterprise characteristics include organizational type, operating period and business scope type.

9. The method according to claim 8, characterized in that The early warning model is trained by the lightGBM algorithm.

10. A multi-level model system for early warning of traffic industry violation risks, using the method according to any one of claims 1 to 9, characterized in that: include: The data acquisition module is used to obtain historical expert experience documents on traffic accident handling in the transportation industry, descriptions of illegal behaviors, and risk profile indicators of transportation enterprises; A processing module is used to segment and remove duplicate words from risk behavior text and illegal behavior description, and map them and risk levels into binary word vectors respectively; Feature extraction module, used to extract enterprise features from the risk profile indicator system of transportation enterprises; The prediction module is used to calculate the risk level corresponding to the illegal behavior words and use the lightGBM model to predict the probability of illegal behavior by transportation companies in the next year.

11. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 9 is implemented.

12. An electronic device, characterized in that: including a processor and a memory; The memory is used to store historical expert experience documents on traffic accident handling in the transportation industry, descriptions of illegal behaviors, and risk profile indicators of transportation enterprises; The processor is used to execute the method described in any one of claims 1 to 9 by calling historical expert experience documents, illegal behavior descriptions and transportation enterprise risk profile indicators, using the lightGBM model.

Images