Method, system and program product for evaluating driving safety of official vehicle

Abstract

The invention belongs to the technical field of official vehicle management, and particularly discloses an official vehicle driving safety assessment method and system and a program product, and the method comprises the steps: collecting driving data in real time in an official vehicle use process, carrying out the data processing and feature extraction, and carrying out the risk recognition through employing a corresponding risk recognition network model based on the extracted features. According to the method, the corresponding risk category and probability are determined, then the risk level, the risk feature and the suggested measure are further judged, and finally the risk information is summarized to generate the safety assessment report for output, so that efficient driving safety assessment of the official vehicle can be realized, the risk identification capability is improved, and data-driven decision support is provided for safety management personnel; the safety consciousness of a driver can be enhanced, and the traffic safety accident rate of official vehicles is reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of official vehicle management, and specifically relates to a driving safety assessment method, system and program product for official vehicles. Background Art

[0002] With the rapid development of information technology, the official vehicle management system has gradually transformed into an intelligent and digital one. Through various sensors and communication equipment, it can realize functions such as real-time monitoring, dispatch management and fault monitoring of vehicles, thereby significantly improving the management efficiency of official vehicles. However, although the existing official vehicle network system has been improved in terms of function and efficiency, there are still the following problems in the management of official vehicles:

[0003] Lack of timeliness: Existing technologies often adopt a periodic security assessment approach and lack real-time monitoring and feedback mechanisms. This makes it impossible for the system to respond in a timely manner when facing real-time security threats, resulting in potential risks not being identified and handled in a timely manner.

[0004] Poor dynamic risk identification capabilities: Existing technologies are insufficient in their ability to process dynamic data. The complexity of the network environment and the frequency of data interaction make it difficult for traditional security solutions to effectively identify potential risks, thus affecting the security protection effect of the system.

[0005] Lack of effective risk detection mechanism: Existing technologies are weak in risk detection capabilities and usually rely on post-analysis, making it difficult to detect and respond to abnormal situations in a timely manner. This delayed response not only reduces the security of the system, but can also lead to serious data loss and service interruption.

[0006] Insufficient overall security assessment: Existing technologies often only focus on security measures in one aspect, but lack a comprehensive assessment of the security situation. This one-sided security protection strategy affects the overall security protection performance of the system and is prone to weak links.

[0007] In summary, the existing official vehicle network management system has obvious deficiencies in safety assessment and is in urgent need of improvement. Summary of the invention

[0008] The purpose of the present invention is to provide a method, system and program product for evaluating the driving safety of official vehicles, so as to solve the above-mentioned problems existing in the prior art.

[0009] In order to achieve the above object, the present invention adopts the following technical solutions:

[0010] In a first aspect, a driving safety assessment method for an official vehicle is provided, comprising:

[0011] The driving data of the target official vehicle is collected in real time through the official vehicle network, and the driving data are pre-processed to obtain a driving data set;

[0012] Extract features from the driving data set to obtain a feature set;

[0013] Inputting the feature set into a preset risk identification network model to perform risk identification and obtain a risk identification result, wherein the risk identification result includes each risk category and the risk probability corresponding to each risk category;

[0014] Calculate the comprehensive risk score corresponding to each risk category based on each risk category and the risk probability corresponding to each risk category;

[0015] Determine the risk level of each risk category based on the comprehensive risk score corresponding to each risk category, determine the recommended measures for the corresponding risk category based on the risk level, and determine the corresponding risk characteristics based on the risk category;

[0016] Summarize each risk category and its corresponding risk probability, comprehensive risk score, risk level, risk characteristics and recommended measures to generate a safety assessment report for the target official vehicle;

[0017] Output the safety assessment report of the target official vehicle.

[0018] In a possible design, the preprocessing of each driving data to obtain a driving data set includes:

[0019] Data cleaning and data formatting are performed on various driving data to obtain corresponding formatted data, and data integration is performed on various formatted data after data cleaning and data formatting to obtain a driving data set, wherein the driving data set includes time information, position information, speed information, acceleration information, direction information, fuel consumption information, fault information, driver information, task information, environment information and path information.

[0020] In a possible design, the feature extraction is performed on the driving data set to obtain a feature set, including:

[0021] Select each feature data based on the driving data set, perform feature construction calculation on each feature data, and obtain the corresponding feature value;

[0022] Scaling each eigenvalue to obtain a corresponding nominal feature, wherein the scaling includes normalization or standardization;

[0023] The nominal features are used to form a feature set, which includes position features, speed features, acceleration features, driving behavior features, environment features and task features.

[0024] In one possible design, the risk identification network model includes a convolutional neural network model that has been trained, verified and tested, and the loss function during training adopts the cross-entropy loss function, and the evaluation indicators during verification and testing adopt accuracy, precision, recall rate and F1 score.

[0025] In one possible design, the risk identification network model includes a recurrent neural network model that has been trained, verified and tested, and the loss function during training adopts the cross entropy loss function, and the evaluation indicators during verification and testing adopt accuracy, precision, recall rate and F1 score.

[0026] In a possible design, the step of calculating the comprehensive risk score corresponding to each risk category according to each risk category and the risk probability corresponding to each risk category includes:

[0027] Retrieve the preset comprehensive risk scoring formula and determine the corresponding historical risk factor and environmental impact factor based on the risk category;

[0028] The risk probability, historical risk coefficient and environmental impact coefficient corresponding to the risk category are substituted into the comprehensive risk scoring formula for calculation to obtain the corresponding comprehensive risk score. The comprehensive risk scoring formula is: R=P×H×E, where R is the comprehensive risk score, P is the risk probability, H is the historical risk coefficient, and E is the environmental impact coefficient.

[0029] In a possible design, determining the risk level of each risk category according to the comprehensive risk score corresponding to each risk category includes:

[0030] The comprehensive risk score corresponding to each risk category is substituted into the preset risk rating rules for matching to determine the corresponding risk level. The risk rating rules include several risk levels and the comprehensive risk score interval corresponding to each risk level.

[0031] In a second aspect, a driving safety assessment system for official vehicles is provided, comprising a data acquisition unit, a feature extraction unit, a risk identification unit, a score calculation unit, a grade determination unit, a comprehensive assessment unit and a result output unit, wherein:

[0032] A data collection unit is used to collect various driving data of the target official vehicle in real time through the official vehicle network, and pre-process the various driving data to obtain a driving data set;

[0033] A feature extraction unit, used for extracting features from the driving data set to obtain a feature set;

[0034] A risk identification unit, used for inputting a feature set into a preset risk identification network model to perform risk identification and obtain a risk identification result, wherein the risk identification result includes each risk category and the risk probability corresponding to each risk category;

[0035] A scoring calculation unit, used to calculate a comprehensive risk score corresponding to each risk category according to each risk category and the risk probability corresponding to each risk category;

[0036] A level determination unit is used to determine the risk level of the corresponding risk category according to the comprehensive risk score corresponding to each risk category, determine the recommended measures for the corresponding risk category according to the risk level, and determine the corresponding risk characteristics according to the risk category;

[0037] A comprehensive assessment unit is used to summarize each risk category and the corresponding risk probability, comprehensive risk score, risk level, risk characteristics and recommended measures to generate a safety assessment report for the target official vehicle;

[0038] The result output unit is used to output the safety assessment report of the target official vehicle.

[0039] In a third aspect, a driving safety assessment system for official vehicles is provided, comprising:

[0040] A memory for storing instructions;

[0041] A processor is used to read the instructions stored in the memory and execute any one of the methods described in the first aspect according to the instructions.

[0042] In a fourth aspect, a computer-readable storage medium is provided, wherein instructions are stored on the computer-readable storage medium, and when the instructions are executed on a computer, the computer executes any one of the methods described in the first aspect. In addition, a computer program product is provided, and when the computer program product is executed on a computer, the computer executes any one of the methods described in the first aspect.

[0043] Beneficial effects:

[0044] 1. It can improve security identification capabilities: The present invention can effectively identify potential security threats, ensure real-time monitoring of the use of official vehicles and rapid response to abnormal situations, thereby reducing security risks.

[0045] 2. Efficient risk assessment can be achieved: The present invention introduces a comprehensive risk scoring mechanism, which can quantitatively evaluate various potential risks. Through real-time monitoring and risk scoring, managers can promptly identify high-risk behaviors and take corresponding measures to effectively prevent accidents.

[0046] 3. Provide data-driven decision support: The safety assessment report generated by the present invention can provide detailed risk identification results, risk levels and targeted recommended measures, providing data-driven decision support for managers. This systematic report makes the management of official vehicles more scientific and reasonable, and improves management efficiency.

[0047] 4. It can reduce the incidence of safety accidents: The present invention can significantly improve the management level of official vehicles for official use, facilitate timely intervention measures for identified risk behaviors, effectively reduce the occurrence of safety accidents, and ensure the safety of drivers and passengers.

[0048] 5. It can enhance the driver's safety awareness: The present invention can effectively enhance the driver's safety awareness and promote the formation of a safe driving culture by identifying risk situations during the use of official vehicles and providing feedback. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0050] Figure 1 This is a schematic diagram of the steps of the method in Example 1 of the present invention;

[0051] Figure 2 This is a schematic diagram of the structure of the system in Example 2 of the present invention;

[0052] Figure 3 Schematic diagram of the system structure in Example 3 of the present invention. DETAILED DESCRIPTION

[0053] It should be noted that the description of these embodiments is used to help understand the present invention, but does not constitute a limitation of the present invention. The specific structures and functional details disclosed herein are only used to describe the exemplary embodiments of the present invention. However, the present invention can be embodied in many alternative forms, and it should not be understood that the present invention is limited to the embodiments set forth herein.

[0054] It should be understood that, unless otherwise clearly specified and limited, the term "connection" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, or it can be the internal communication of two elements. For ordinary technicians in this field, the specific meanings of the above terms in the embodiments can be understood according to specific circumstances.

[0055] In the following description, certain details are provided to facilitate a complete understanding of the example embodiments. However, it will be appreciated by those of ordinary skill in the art that the example embodiments may be implemented without these certain details. For example, devices may be shown in block diagrams to avoid obscuring the examples with unnecessary details. In other embodiments, well-known processes, structures, and techniques may not be shown in unnecessary detail to avoid obscuring the embodiments.

[0056] Embodiment 1:

[0057] This embodiment provides a driving safety assessment method for official vehicles, which can be applied to corresponding official vehicle network servers, such as Figure 1 As shown, the method comprises the following steps:

[0058] S1. Various driving data of the target official vehicle are collected in real time through the official vehicle network, and various driving data are pre-processed to obtain a driving data set.

[0059] During specific implementation, various driving data of the target official vehicles can be collected in real time through the official vehicle network. In order to ensure the accuracy and real-time nature of data collection, the following technical means can be used: 1. Wireless transmission protocol: Data is transmitted in real time through wireless networks such as LTE / 5G to ensure that the information can be quickly transmitted to the server after collection. 2. Combination of timed collection and event-driven collection: For dynamically changing data, such as position and speed, a timed collection method is used; and for key events (such as failures, changes in mission instructions, etc.), an event-driven method is used to ensure the immediacy of data. 3. Data encryption and integrity verification: During data transmission, the SSL / TLS protocol is used for data encryption to ensure the security of transmitted data, and the integrity of the data is verified through methods such as checksums.

[0060] Various driving data can be derived from the following aspects:

[0061] 1. Vehicle-mounted sensors:

[0062] GPS module: used to obtain the vehicle's real-time location information, including longitude, latitude, altitude and other data.

[0063] Speed ​​sensor: monitors vehicle speed in real time and provides speed change information.

[0064] Acceleration sensor: monitors the acceleration and deceleration of the vehicle and provides acceleration data.

[0065] Gyroscope: used to detect the vehicle's steering angle and posture changes and obtain the vehicle's dynamic stability data.

[0066] Fuel consumption sensor: monitors the vehicle's fuel consumption and provides fuel consumption data per kilometer.

[0067] Temperature sensor: monitors the temperature of the engine and other key components to ensure the safety of vehicle operation.

[0068] 2. On-board computer system:

[0069] OBD-II interface data: collects vehicle fault codes, engine operating status, driving behavior data and other information.

[0070] Driving recorder data: records driving trajectory, parking conditions, idling time, etc., providing data support for subsequent analysis.

[0071] 3. Dispatch center data:

[0072] Task instruction information: Collect the vehicle's task assignment information, including task start time, task content, expected completion time, etc.

[0073] Vehicle status information: including whether the vehicle is online, status (idle, running, faulty, etc.), current location, etc.

[0074] Dispatch command history: record the instructions and feedback issued by the dispatch center, and analyze task response status.

[0075] 4. User interaction data:

[0076] Driver identity verification information: includes the driver's ID, identity verification time, verification method (such as fingerprint, password), etc., which is used to analyze the reliability of the user's identity.

[0077] Operation record: records the operations performed by the driver during driving, including the time and frequency of acceleration, braking, lane changing, etc.

[0078] 5. Environmental monitoring data:

[0079] Meteorological data: Obtain real-time weather information through external interfaces, including temperature, humidity, precipitation, wind speed, etc., to assess the impact of the environment on vehicle driving.

[0080] Traffic flow data: Obtain the current traffic flow situation on the road and analyze the potential threat of traffic congestion to driving safety.

[0081] Exemplarily, the various driving data finally collected may include: location data: such as longitude, latitude, and altitude; speed data: such as instantaneous speed and average speed; acceleration data: such as instantaneous acceleration and deceleration; steering data: such as steering angle and steering rate; fuel consumption data: such as instantaneous fuel consumption and cumulative fuel consumption; temperature data: such as engine temperature and oil temperature; fault information: such as fault code, fault description, and fault occurrence time; driving behavior data: such as acceleration events, sudden braking events, and lane change frequency; task information: such as task type, task start time, task end time, and task status; driver information: such as driver ID, identity authentication time, and verification method; environmental data: such as real-time weather conditions, temperature, humidity, and traffic flow.

[0082] After completing the collection of various driving data, it is necessary to pre-process the driving data. Data pre-processing is a key step to ensure data quality and accuracy, laying the foundation for subsequent analysis. The pre-processing process includes:

[0083] 1. Data cleaning. Identify and remove invalid data and outliers. The specific process includes:

[0084] Identify missing values: Check whether there are missing values ​​in each data item. Missing values ​​may lead to inaccurate results of subsequent analysis. Statistical methods (such as mean interpolation, nearest neighbor interpolation, etc.) can be used to fill missing values ​​to ensure data integrity.

[0085] Remove outliers: Use statistical methods (such as Z-score or IQR method) to detect outliers. For key indicators such as speed and position, set reasonable upper and lower limits. Data outside the range will be marked as outliers and processed. Outliers can be processed by removing, correcting or marking them as special data.

[0086] Format consistency check: Check whether the formats of each data item are consistent. For example, the date format should be unified as YYYY-MM-DD to ensure that data from different sources are formatted consistently.

[0087] 2. Data formatting. Unify data from different sources into a processable format. The specific process includes:

[0088] Timestamp normalization: All time data is standardized into UTC format to facilitate time series analysis.

[0089] Unit unification: Convert data in different units into a unified standard, for example, unify speed into kilometers per hour (km / h) and fuel consumption into liters per kilometer (L / km) to ensure data consistency.

[0090] Feature encoding: For categorical data (such as driver ID, task type), one-hot encoding or label encoding is used to convert it into numerical features for the input of subsequent models.

[0091] 3. Data integration. After data cleaning and data formatting, various formatted data are merged to form a unified driving data set. The specific process includes:

[0092] Data merging: Merge data from different sources (such as on-board sensors, dispatch centers, and user interactions) to ensure that each record contains relevant contextual information. For example, the location information, speed, mission instructions, and driver information of each vehicle are integrated into a complete record.

[0093] Data storage: Store the integrated data in a database, using a relational database or a NoSQL database, for quick retrieval and analysis.

[0094] Finally, the integrated and stored driving data set may include time information, location information, speed information, acceleration information, direction information, fuel consumption information, fault information, driver information, task information, environment information and path information. Among them:

[0095] 1. Time information:

[0096] Timestamp: UTC timestamp in a unified format, recording the collection time of each piece of data.

[0097] 2. Location information:

[0098] Longitude: The longitude information of the vehicle’s current location.

[0099] Latitude: The latitude information of the vehicle’s current location.

[0100] Altitude (optional): The altitude of the vehicle's current location.

[0101] 3. Speed ​​information:

[0102] Current Speed: The current speed of the vehicle (unit: km / h).

[0103] Speed ​​Change: The difference between the current speed and the previous speed, used to identify sudden acceleration or deceleration.

[0104] 4. Acceleration information:

[0105] Acceleration: The acceleration value of the vehicle, used to judge driving behavior (unit: m / s 2 ).

[0106] 5. Direction information:

[0107] Steering Angle: The steering angle of the vehicle, reflecting the direction of the vehicle.

[0108] 6. Fuel consumption information:

[0109] Current Fuel Consumption: Current fuel consumption value (unit: L / km).

[0110] Total Fuel Consumption: The cumulative fuel consumption since the last refueling.

[0111] 7. Fault information:

[0112] Fault Codes: Vehicle fault diagnostic information, recording engine and system fault codes.

[0113] Error Description: Detailed description of the error.

[0114] 8. Driver information:

[0115] Driver ID: The driver's unique identification code.

[0116] Authentication Time: The timestamp of the driver's authentication.

[0117] Authentication Method: The authentication method used (such as password, fingerprint, etc.).

[0118] 9. Task information:

[0119] Task Type: The type of task currently performed by the vehicle (such as pick-up, transportation).

[0120] Task Status: The status of the current task (e.g. in progress, completed).

[0121] Task Start Time: The timestamp when the task starts.

[0122] Task End Time (optional): The timestamp when the task is completed.

[0123] 10. Environmental information:

[0124] Weather Information (Weather Data): Real-time weather conditions, including temperature, humidity, precipitation, etc.

[0125] Traffic Data: Traffic flow information on current roads, used to assess traffic conditions.

[0126] 11. Path information:

[0127] Driving Path: The path data of the vehicle, including path length, path curvature and other characteristics.

[0128] S2. Extract features from the driving data set to obtain a feature set.

[0129] In specific implementation, feature extraction can extract features from driving data sets that are critical to subsequent model training and analysis. These features can be used to identify potential safety risks and driving behaviors, providing effective information support for subsequent threat identification and risk assessment. The feature extraction process includes:

[0130] 1. Feature selection and definition. Select each feature data based on the driving data set, such as:

[0131] Location feature data:

[0132] Current longitude and latitude: Directly use the longitude and latitude in the intermediate data as location features.

[0133] Driving path length: The driving path length of the vehicle is obtained by calculating the distance between GPS track points.

[0134] Speed ​​characteristic data:

[0135] Instantaneous speed: Directly use the instantaneous speed in the intermediate data.

[0136] Speed ​​change rate: Calculates the difference between the current speed and the speed at the previous moment, reflecting the speed fluctuation.

[0137] Acceleration characteristic data:

[0138] Acceleration value: Directly use the acceleration data in the intermediate data.

[0139] Acceleration event count: Counts the number of acceleration events within a certain time window.

[0140] Driving behavior characteristic data:

[0141] Rapid acceleration event: set a threshold (such as acceleration> 3m / s 2), when the acceleration exceeds the threshold, it is marked as a rapid acceleration event.

[0142] Emergency braking event: set a negative threshold (such as acceleration <-3m / s 2 ), when the acceleration is lower than the threshold, it is marked as an emergency braking event.

[0143] Environmental characteristic data:

[0144] Weather impact factor: convert weather information into numerical features (such as rainy day = 1, sunny day = 0) to evaluate the impact of weather on driving behavior.

[0145] Traffic flow index: Calculates the degree of traffic congestion based on traffic flow data. Example: Traffic flow index is: 5 (1-10, 1 is smooth and 10 is congested).

[0146] Task characteristic data:

[0147] Task type encoding: One-hot encoding of the task type, such as pick-up task, transportation task, etc. Example: Pick-up task = 1, transportation task = 0, the current task type is pick-up.

[0148] Task duration: Calculate the difference between the task start time and end time to get the task duration. Example: Task start time: 08:00, end time: 09:00, then the duration is: 1 hour

[0149] 2. Feature construction and calculation. Perform feature construction calculation on each feature data to obtain the corresponding feature value. After selecting the feature data, the feature data can be substituted into the corresponding feature construction calculation model or formula for calculation to obtain the corresponding feature value. For example, for calculation of driving path length: use the Haversine formula to calculate the distance between GPS coordinates to obtain the driving path length; for counting sudden acceleration and sudden braking events: use conditional judgment to count sudden acceleration and sudden braking events; for calculation of task duration: use time information to calculate the duration of the task.

[0150] 3. Feature scaling: In order to ensure that different eigenvalues ​​are at the same level and help the convergence of subsequent models, standardization or normalization methods can be used to process the eigenvalues.

[0151] Standardization (Z-score standardization):

[0152]

[0153] Among them, X′ is the standardized eigenvalue, μ is the mean of the feature, and σ is the standard deviation.

[0154] Normalization (Min-Max Normalization):

[0155]

[0156] The eigenvalues ​​are scaled to the interval [0, 1] to ensure that different eigenvalues ​​are on the same scale.

[0157] 4. Use each nominal feature to form a feature set. For example, the final feature set may include location features: longitude and latitude, driving path length; speed features: instantaneous speed, speed change rate; acceleration features: acceleration value, sudden acceleration event count, sudden braking event count; driving behavior features: events related to driving behavior; environmental features: weather influencing factors, traffic flow index; task features: task type code, task duration.

[0158] S3. Input the feature set into a preset risk identification network model to perform risk identification and obtain a risk identification result, wherein the risk identification result includes each risk category and the risk probability corresponding to each risk category.

[0159] In specific implementation, machine learning / deep learning models can be used for risk prediction to improve the ability to identify abnormal behaviors and potential risks. That is, the feature set is input into the preset risk identification network model for risk identification to obtain risk identification results, which include each risk category and the risk probability corresponding to each risk category. Risk categories such as "sudden acceleration", "sudden braking", "abnormal position", etc., and the corresponding risk probabilities are such as sudden acceleration: 0.85, sudden braking: 0.10, and normal behavior: 0.05.

[0160] The risk identification network model can adopt a convolutional neural network model or a recurrent neural network model that has been trained, verified and tested. The convolutional neural network model (CNN) is suitable for processing data with spatial structure and can effectively identify complex patterns and relationships. The recurrent neural network model (RNN) is suitable for processing time series data and can capture the temporal dependencies in the data.

[0161] The convolutional neural network model structure includes:

[0162] Input layer: receives feature vectors with shape N, T, F, where N is the number of samples, T is the time step, and F is the number of features. For example, the input features may be: 65, 10.5, 2.5, 0, 1 (instantaneous speed, driving path length, acceleration, weather impact factor, task type code).

[0163] Convolutional layer: extract features through multiple convolution kernels.

[0164] Pooling layer: down-samples to reduce feature dimensions and computational complexity.

[0165] Fully connected layer: Outputs classification probability and uses softmax activation function to convert the output into probability.

[0166] The recurrent neural network model structure includes:

[0167] Input layer: receives feature vectors, such as time series data.

[0168] Hidden layer: The core part of RNN, responsible for transmitting hidden states.

[0169] Output layer: Outputs the classification results and uses the softmax function to convert the hidden state into probability.

[0170] Before training the model, you need to prepare the corresponding feature sample data. Assume that the feature sample data is 1000 records, which can be divided into training set (700 records), validation set (150 records) and test set (150 records). For each feature sample data, the corresponding normal behavior can be marked as 0, and the corresponding positive abnormal behavior (such as sudden acceleration, sudden braking, etc.) can be marked as 1. Example feature sample data: Normal behavior: 65, 10.5, 0.5, 0, 1-> label is 0; Abnormal behavior (sudden acceleration): 80, 12.0, 3.0, 1, 0-> label is 1.

[0171] When training the model, the cross entropy loss function can be used to measure the difference between the probability distribution predicted by the model and the true label; the Adam optimizer is used to update parameters to accelerate convergence and improve model performance; after training, the validation set and test set are used to evaluate the model performance. The evaluation indicators include: Accuracy, Precision, Recall and F1-score, etc., to comprehensively evaluate the performance of the model; according to the evaluation results, the hyperparameters (such as learning rate) can be adjusted and the model can be retrained.

[0172] S4. Calculate the comprehensive risk score corresponding to each risk category based on each risk category and the risk probability corresponding to each risk category.

[0173] In specific implementation, a preset comprehensive risk scoring formula can be retrieved, and the corresponding historical risk coefficient and environmental impact coefficient are determined based on the risk category. For example, according to the accident records in the past year, the proportion of accidents caused by sudden acceleration is 30%, so the historical risk coefficient of the sudden acceleration risk category is 0.3. If it is determined that the current weather is rainy when suddenly accelerating, the corresponding environmental impact coefficient is 1, and so on. Then, the risk probability, historical risk coefficient, and environmental impact coefficient corresponding to the risk category are substituted into the comprehensive risk scoring formula for calculation to obtain the corresponding comprehensive risk score. The comprehensive risk scoring formula is: R = P × H × E, where R is the comprehensive risk score, P is the risk probability, H is the historical risk coefficient, and E is the environmental impact coefficient. Exemplarily, if the risk probability corresponding to the sudden acceleration risk category is 0.85, the historical risk coefficient is 0.3, and the environmental impact coefficient is 1, then its comprehensive risk score R = 0.85 × 0.3 × 1 = 0.255.

[0174] S5. Determine the risk level corresponding to each risk category based on the comprehensive risk score corresponding to each risk category, and determine the recommended measures corresponding to each risk category according to the risk level, and determine the corresponding risk characteristics according to the risk category.

[0175] In specific implementation, substitute the comprehensive risk scores corresponding to each risk category into the preset risk rating rules for matching to determine the corresponding risk level. The risk rating rules include several risk levels and the comprehensive risk score intervals corresponding to each risk level. Such as high risk level: R > 0.6; medium risk level: 0.3 < R ≤ 0.6; low risk level: R ≤ 0.3R. Recommended measures for the high risk level of sudden acceleration: Immediately notify relevant management personnel and take emergency measures, such as restricting vehicle travel or increasing the monitoring frequency; Recommended measures for the medium risk level of sudden acceleration: It is recommended to conduct regular inspections and provide safety training for drivers to reduce potential risks; Recommended measures for the low risk level of sudden acceleration: Maintain the current monitoring strategy, record and continue to observe. When R = 0.255 for sudden acceleration, it corresponds to the low risk level, and the corresponding recommended measure is: Maintain the current monitoring strategy, record and continue to observe. At the same time, determine the corresponding risk characteristics according to the risk category. For example, the risk characteristics corresponding to sudden acceleration are: the acceleration exceeds 3m / s 2 , and the instantaneous speed is above 60km / h; the risk characteristics corresponding to sudden braking are: the acceleration is below -3m / s 2 , and the instantaneous speed is below 30km / h; the risk characteristics corresponding to abnormal position are: the vehicle position deviates from the set route by more than 50 meters; and so on.

[0176] S6. Summarize each risk category and the risk probability, comprehensive risk score, risk level, risk characteristics, and recommended measures corresponding to each risk category to generate a safety assessment report for the target official vehicle.

[0177] In specific implementation, after the risk assessment is completed, a safety assessment report for the target official vehicle can be generated, which aims to integrate the identified threats, assessed risk levels and corresponding recommended measures to provide decision support for managers. That is, each risk category and the corresponding risk probability, comprehensive risk score, risk level, risk characteristics and recommended measures are summarized to generate a safety assessment report for the target official vehicle.

[0178] S7. Output the safety assessment report of the target official vehicle.

[0179] During specific implementation, the safety assessment report of the target official vehicle can be stored and visualized to provide decision-making support for managers, allowing them to take timely intervention measures based on identified risk behaviors, effectively reduce the occurrence of safety accidents, and ensure the safety of drivers and passengers.

[0180] Embodiment 2:

[0181] This embodiment provides a driving safety assessment system for official vehicles. Figure 2 As shown, it includes a data acquisition unit, a feature extraction unit, a risk identification unit, a score calculation unit, a grade determination unit, a comprehensive evaluation unit and a result output unit, wherein:

[0182] A data collection unit is used to collect various driving data of the target official vehicle in real time through the official vehicle network, and pre-process the various driving data to obtain a driving data set;

[0183] A feature extraction unit, used for extracting features from the driving data set to obtain a feature set;

[0184] A risk identification unit, used for inputting a feature set into a preset risk identification network model to perform risk identification and obtain a risk identification result, wherein the risk identification result includes each risk category and the risk probability corresponding to each risk category;

[0185] A scoring calculation unit, used to calculate a comprehensive risk score corresponding to each risk category according to each risk category and the risk probability corresponding to each risk category;

[0186] A level determination unit is used to determine the risk level of the corresponding risk category according to the comprehensive risk score corresponding to each risk category, determine the recommended measures for the corresponding risk category according to the risk level, and determine the corresponding risk characteristics according to the risk category;

[0187] A comprehensive assessment unit is used to summarize each risk category and the corresponding risk probability, comprehensive risk score, risk level, risk characteristics and recommended measures to generate a safety assessment report for the target official vehicle;

[0188] The result output unit is used to output the safety assessment report of the target official vehicle.

[0189] Embodiment 3:

[0190] This embodiment provides a driving safety assessment system for official vehicles. Figure 3 As shown, at the hardware level, it includes:

[0191] Data interface, used to establish data connection between the processor and the external data terminal;

[0192] A memory for storing instructions;

[0193] The processor is used to read the instructions stored in the memory and execute the driving safety assessment method for official vehicles in Example 1 according to the instructions.

[0194] Optionally, the system further includes an internal bus, through which the processor, the memory and the data interface can be interconnected, and the internal bus can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc.

[0195] The memory may include, but is not limited to, random access memory (RAM), read only memory (ROM), flash memory, first input first output (FIFO) and / or first in last out (FILO) memory, etc. The processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0196] Embodiment 4:

[0197] This embodiment provides a computer-readable storage medium, on which instructions are stored, and when the instructions are executed on a computer, the computer executes the driving safety assessment method for official vehicles in Embodiment 1. The computer-readable storage medium refers to a carrier for storing data, which may include but is not limited to a floppy disk, an optical disk, a hard disk, a flash memory, a USB flash drive, and / or a memory stick, etc. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.

[0198] This embodiment also provides a computer program product, which, when executed on a computer, executes the driving safety assessment method for official vehicles in Embodiment 1. The computer may be a general-purpose computer, a special-purpose computer, a computer network or other programmable devices.

[0199] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention. 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. A driving safety assessment method for official vehicles, characterized in that: include: The driving data of the target official vehicle is collected in real time through the official vehicle network, and the driving data are pre-processed to obtain a driving data set; Extract features from the driving data set to obtain a feature set; Inputting the feature set into a preset risk identification network model to perform risk identification and obtain a risk identification result, wherein the risk identification result includes each risk category and the risk probability corresponding to each risk category; Calculate the comprehensive risk score corresponding to each risk category based on each risk category and the risk probability corresponding to each risk category; Determine the risk level of each risk category based on the comprehensive risk score corresponding to each risk category, determine the recommended measures for the corresponding risk category based on the risk level, and determine the corresponding risk characteristics based on the risk category; Summarize each risk category and its corresponding risk probability, comprehensive risk score, risk level, risk characteristics and recommended measures to generate a safety assessment report for the target official vehicle; Output the safety assessment report of the target official vehicle.

2. A driving safety assessment method for official vehicles according to claim 1, characterized in that: The preprocessing of various driving data to obtain a driving data set includes: Data cleaning and data formatting are performed on various driving data to obtain corresponding formatted data, and data integration is performed on various formatted data after data cleaning and data formatting to obtain a driving data set, wherein the driving data set includes time information, position information, speed information, acceleration information, direction information, fuel consumption information, fault information, driver information, task information, environment information and path information.

3. A driving safety assessment method for official vehicles according to claim 1, characterized in that: The feature extraction of the driving data set is performed to obtain a feature set, including: Select each feature data based on the driving data set, perform feature construction calculation on each feature data, and obtain the corresponding feature value; Scaling each eigenvalue to obtain a corresponding nominal feature, wherein the scaling includes normalization or standardization; The nominal features are used to form a feature set, which includes position features, speed features, acceleration features, driving behavior features, environment features and task features.

4. A driving safety assessment method for official vehicles according to claim 1, characterized in that: The risk identification network model includes a convolutional neural network model that has been trained, verified and tested, and the loss function during training adopts the cross-entropy loss function, and the evaluation indicators during verification and testing adopt accuracy, precision, recall rate and F1 score.

5. A driving safety assessment method for official vehicles according to claim 1, characterized in that: The risk identification network model includes a recurrent neural network model that has been trained, verified and tested, and the loss function during training adopts the cross entropy loss function, and the evaluation indicators during verification and testing adopt accuracy, precision, recall rate and F1 score.

6. A driving safety assessment method for official vehicles according to claim 1, characterized in that: The calculating of the comprehensive risk score corresponding to each risk category according to each risk category and the risk probability corresponding to each risk category includes: Retrieve the preset comprehensive risk scoring formula and determine the corresponding historical risk factor and environmental impact factor based on the risk category; The risk probability, historical risk coefficient and environmental impact coefficient corresponding to the risk category are substituted into the comprehensive risk scoring formula for calculation to obtain the corresponding comprehensive risk score. The comprehensive risk scoring formula is: R=P×H×E, where R is the comprehensive risk score, P is the risk probability, H is the historical risk coefficient, and E is the environmental impact coefficient.

7. A driving safety assessment method for official vehicles according to claim 1, characterized in that: Determining the risk level of each risk category according to the comprehensive risk score corresponding to each risk category includes: The comprehensive risk score corresponding to each risk category is substituted into the preset risk rating rules for matching to determine the corresponding risk level. The risk rating rules include several risk levels and the comprehensive risk score interval corresponding to each risk level.

8. A driving safety assessment system for official vehicles, characterized in that: It includes a data collection unit, a feature extraction unit, a risk identification unit, a score calculation unit, a grade determination unit, a comprehensive evaluation unit and a result output unit, wherein: A data collection unit is used to collect various driving data of the target official vehicle in real time through the official vehicle network, and pre-process the various driving data to obtain a driving data set; A feature extraction unit, used for extracting features from the driving data set to obtain a feature set; A risk identification unit, used for inputting a feature set into a preset risk identification network model to perform risk identification and obtain a risk identification result, wherein the risk identification result includes each risk category and the risk probability corresponding to each risk category; A scoring calculation unit, used to calculate a comprehensive risk score corresponding to each risk category according to each risk category and the risk probability corresponding to each risk category; A level determination unit is used to determine the risk level of the corresponding risk category according to the comprehensive risk score corresponding to each risk category, determine the recommended measures for the corresponding risk category according to the risk level, and determine the corresponding risk characteristics according to the risk category; A comprehensive assessment unit is used to summarize each risk category and the corresponding risk probability, comprehensive risk score, risk level, risk characteristics and recommended measures to generate a safety assessment report for the target official vehicle; The result output unit is used to output the safety assessment report of the target official vehicle.

9. A driving safety assessment system for official vehicles, characterized in that: include: A memory for storing instructions; A processor is used to read the instructions stored in the memory and execute the driving safety assessment method for official vehicles described in any one of claims 1-7 according to the instructions.

10. A computer program product, characterized in that When the computer program product is run on a computer, the driving safety assessment method for official vehicles according to any one of claims 1 to 7 is executed.

Images