Method for dynamically adjusting score based on driving behavior of commercial vehicle

Abstract

The invention provides a method for dynamically adjusting a score based on a driving behavior of a commercial vehicle. The method comprises the following steps: S1, a data acquisition module acquires original vehicle operation data according to an acquisition frequency; s2, sending the original vehicle operation data to a data processing module; s3, the data processing module sends the standardized key driving behavior characteristic data to a score calculation module; s4, the score calculation module calculates a driving behavior score according to the feature data; s5, the score calculation module sends a score result to the client platform; according to the invention, the total score of the driver in four aspects of vehicle speed control, accelerator control, idle speed control and brake control is calculated according to the score calculation module, and then the dynamic environment weight model is used to dynamically adjust the total score according to the driving environment change dynamic adjustment environment, so that the obtained driving behavior score is more real and comprehensive, and the driving behavior score is more accurate. And a more scientific and objective basis is provided for motorcade management.

Description

Technical Field

[0001] The present invention relates to the field of computers, and in particular to a method for dynamically adjusting and scoring driving behaviors based on commercial vehicles. Background Art

[0002] According to market research, the global commercial vehicle market size is expected to reach $500 billion in 2025, and the market share related to driving behavior management exceeds 10%. Especially in China and North America, the number of commercial vehicles is huge, and there is a strong demand for driving behavior management systems.

[0003] With the wide application of commercial vehicles, driving behavior directly affects the operation efficiency, fuel economy, traffic safety and environmental friendliness of vehicles. However, traditional driving behavior analysis mostly focuses on a single dimension, such as speeding or hard braking, lacking a comprehensive and scientific evaluation of driving behavior. Summary of the Invention

[0004] The purpose of the present invention is to solve the deficiencies existing in the prior art, and to propose a method for dynamically adjusting and scoring driving behaviors based on commercial vehicles.

[0005] To achieve the above purpose, the present invention adopts the following technical solutions: A method for dynamically adjusting and scoring driving behaviors based on commercial vehicles, including the following steps: S1: The data acquisition module collects original vehicle operation data according to the acquisition frequency; The data acquisition module includes an OBD device, a CAN bus, and an in-vehicle terminal device; The original vehicle operation data includes real-time vehicle speed V, acceleration a, rotation speed f, throttle opening, gear data, and braking data.

[0006] S2: The data acquisition module sends the collected original vehicle operation data to the data processing module; Specifically, it includes the following steps: S21: The data processing module cleans the original vehicle operation data to obtain vehicle operation data; The cleaning includes removing outliers, duplicate data, and incomplete data; S22: The data processing module extracts key driving behavior feature data from the vehicle operation data; The key driving behavior feature data includes speeding duration T n , idle duration T m , total low-speed time G, total stable time H, total number of frequent idles P, number of hard accelerations N, number of hard brakes M, number of smooth driving times Z, and number of frequent brakes F; Speeding duration T n: The data processing module is provided with a speed threshold V1. The data processing module extracts the real-time vehicle speed V and the corresponding real-time time T from the original vehicle operation data. When V - V1 ≥ 0, the real-time vehicle speed V is speeding. When V - V1 changes from less than 0 to greater than 0, the real-time time T1 at this time is the starting time of the speeding time; when V - V1 changes from greater than 0 to less than 0, the real-time time T2 at this time is the ending time of the speeding time, and the speeding time is T2 - T1. The original vehicle operation data contains multiple periods of speeding time, and the total speeding time T n is the sum of multiple periods of speeding time, T n = (T2 1 - T1 1 ) + (T2 2 - T1 2 ) + …… + (T2 n - T1 n ); Idle time duration T m : When the real-time vehicle speed V becomes 0, the corresponding real-time time is recorded as the idle start time T3; when the real-time vehicle speed V changes from 0, the corresponding real-time time is marked as the idle end time T4, and the idle time duration is recorded as T4 - T3. The original vehicle operation data contains multiple periods of idle time, and the total idle time T m is the sum of multiple periods of idle time, T m = (T4 1 - T3 1 ) + (T4 2 - T3 2 ) + …… + (T4 m - T3 m ); Total low-speed time G: The data processing module is provided with a threshold A. When the real-time vehicle speed V ≤ A, the current vehicle speed is marked as low speed. When the real-time vehicle speed V changes from greater than the threshold A to less than the threshold A, the corresponding time is marked as T5. When the real-time vehicle speed V changes from less than the threshold A to greater than the threshold A, the corresponding time is marked as T6. Then the low-speed time is T6 - T5, and the total low-speed time is G = (T6 1 - T5 1 ) + (T6 2 - T5 2 ) + …… + (T6 g - T5 g ); Total stable time H: When A ≤ V ≤ V1, mark the real-time vehicle speed V as the stable vehicle speed. When the real-time vehicle speed V changes from less than A to greater than A, the corresponding time is marked as the stable starting time T7. When the real-time vehicle speed V changes from greater than A to less than A, the corresponding time is marked as the stable starting time T8. Then the stable time is T8–T7, and the total stable time H = (T8 1 -T7 1 )+(T8 2 -T7 2 )+……+(T8 h -T7 h ); Total number of frequent idlings P: The data processing module calculates the time T m between the idling end point T4 m+1 and the next idling starting point T3 p . There is a threshold C in the data processing module. When T p ≤ C, it is marked as one frequent idling. The scoring calculation module counts the total number of frequent idlings P from the original vehicle driving data; Number of hard accelerations and number of hard brakes: There is a threshold B inside the data processing module. When the acceleration increase value △a1 ≥ B within 3 seconds of the acceleration a, it is marked as one hard acceleration; when the acceleration decrease value △a2 ≤ B within 3 seconds of the acceleration a, it is marked as one hard brake; The data processing module counts the number of hard accelerations N and the number of hard brakes M in the original vehicle operation data; Number of smooth drives Z and number of frequent brakes F: The data processing module extracts the number of brakes L from the original vehicle driving data. The scoring calculation module counts the number of brakes L and calculates the time difference T L between adjacent brakes. The data processing module has thresholds D and E. When the time difference T L ≤ D, it is marked as one frequent brake. If the time difference T L ≥ E, it is marked as one smooth drive; The scoring calculation module counts the number of smooth drives Z and the number of frequent brakes F; S3: The data processing module sends the key driving behavior characteristic data to the scoring calculation module; S4: The scoring calculation module calculates the dynamic driving behavior score according to the characteristic data; S5: The scoring calculation module sends the score to the client platform.

[0007] The client platform provides the commercial vehicle driving behavior comprehensive score and the long-term driving behavior trend to the driving behavior scoring dashboard through the visualization interface.

[0008] Further, the step S4 includes the following steps: S41: The scoring calculation module calculates the scores of different dimensions of driving behavior according to the feature data; S42: The scoring calculation module adjusts the weight coefficient of environmental factors on the score in real time through a dynamic environment weight model.

[0009] Further, the step S41 specifically includes the following content: S411: The scoring calculation module scores the vehicle speed control; In the scoring standard, if speeding for one minute, 5 points will be deducted; for each hard acceleration and each hard brake, 3 points will be deducted; if the vehicle travels at a low speed for more than 30 seconds, 4 points will be deducted per minute; Vehicle speed control score calculation formula: Score_V = 100 – { T n *5 + (M + N) * 3 + G * 4}; S412: The scoring calculation module scores the throttle control; In the scoring standard, for each hard acceleration, 5 points will be deducted, and for each minute of smooth time, 10 points will be added; Throttle control score calculation formula: Score_T = 100 – N * 5 + H * 10; S413: The scoring calculation module scores the idle speed control; In the scoring rule, for each minute of idle time T m 6 points will be deducted, and for each frequent idle, 3 points will be deducted; Idle speed control score calculation formula: Score_I = 100 – (T m *6 + P * 3); S414: The scoring calculation module scores the brake control; In the scoring rule, for each hard brake, 5 points will be deducted, and for each frequent brake, 3 points will be deducted; for each smooth driving, 8 points will be rewarded, with a maximum reward of 16 points; Brake control score calculation formula: When Z ≤ 2, Score_B = 100 – (M * 5 + F * 3 – Z * 8), when Z > 2, Score_B = 100 – (M * 5 + F * 3 – 16); S415: The scoring calculation module calculates the sum of the vehicle speed control score, the throttle control score, the idle speed control score, and the brake control score.

[0010] Perform weighted calculations on the four items of vehicle speed control, throttle control, idle speed control, and brake control, where the vehicle speed control accounts for 20%, the throttle control accounts for 40%, the idle speed control accounts for 15%, and the brake control accounts for 25%, to obtain the comprehensive score total-score = (Score_V * 0.20) + (Score_T * 0.40) + (Score_I * 0.15) + (Score_B * 0.25).

[0011] Furthermore, perform weighted calculations on the four items of vehicle speed control, throttle control, idle speed control, and brake control, where the vehicle speed control accounts for 20%, the throttle control accounts for 40%, the idle speed control accounts for 15%, and the brake control accounts for 25%, to obtain the comprehensive score total-score = (Score_V * 0.20) + (Score_T * 0.40) + (Score_I * 0.15) + (Score_B * 0.25).

[0012] Furthermore, the specific steps of step S42 are as follows: The scoring calculation module identifies different driving environments through GPS, high-precision maps, and traffic big data, and generates environmental data; the environmental data includes road characteristics, dynamic environmental characteristics, and time characteristics; S421: The scoring calculation module extracts road characteristics, dynamic environmental characteristics, and time characteristics, combines the three characteristics, and outputs an eight-dimensional vector; S422: The scoring calculation module defines a dynamic environmental weight model, and the dynamic environmental weight model establishes a road attention network and an environmental attention network; S423: The scoring calculation module calculates the road weight and obtains the final score.

[0013] Furthermore, the scoring calculation module splices in the order of road characteristics, dynamic environmental characteristics, and time characteristics, and finally outputs an eight-dimensional feature vector; among them, the road characteristics include one-hot encoding of road types, road curvature, and road slope, the dynamic environmental characteristics include weather severity, normalized visibility, and traffic density, and the time characteristics include hour cycle encoding: converting 24 hours into a sine wave form and holiday flags.

[0014] Furthermore, establish a road attention network: input road characteristics, map three-dimensional characteristics to an eight-dimensional space through linear transformation, filter negative values through the ReLU activation function, and finally output 5 road-related weights; The method of establishing an environmental attention network is the same as that of establishing a road attention network, and the parameters are independent.

[0015] Further, the specific content of step S423: Then calculate the road weights: Take the first 3 dimensions of the eight-dimensional feature vector, and obtain 5 road-related weights through the road attention network, namely road1, road2, road3, road4, and road5. After softmax normalization, the sum of the 5 road-related weights is 1. Calculate the environmental weights: Take the 3-6 dimensions of the eight-dimensional feature vector, and obtain 5 environment-related weights through the environmental attention network, namely env1, env2, env3, env4, and env5. After being compressed to the range of 0-1 by sigmoid and multiplied by 2 to amplify the influence; The scoring calculation module fuses the road weights and environmental weights into a dynamic weight C = road1 * env1 + road2 + env2 + road3 * 1.2 + env3 * 1.5 + 1 - road4 + 1 - env4 + road5 / env5; Finally, the scoring calculation module multiplies the comprehensive score total-score by the dynamic weight C to obtain the dynamic driving behavior score.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: The method proposed by the present invention scores the driving behavior of commercial vehicles from four dimensions: vehicle speed control, throttle control, idle speed control, and brake control, and performs weighted calculation according to the different degrees of influence of each item on the driving behavior of commercial vehicles, which can more accurately reflect the driving behavior of drivers.

[0017] The method proposed by the present invention: Based on the four dimensions of vehicle speed control, throttle control, idle speed control, and brake control, collect environmental data according to the driving environment of commercial vehicles, calculate the dynamic weight C according to the dynamic environment weight model, and update the dynamic driving behavior score in real time. Description of the Drawings

[0018] Figure 1 It is a step flow chart of a method for dynamically adjusting and scoring the driving behavior of a commercial vehicle according to the present invention. Detailed Embodiment

[0019] In order to further understand the purpose, structure, features, and functions of the present invention, the following is a detailed description in conjunction with the embodiments.

[0020] As Figure 1 shown, a method for dynamically adjusting and scoring the driving behavior of a commercial vehicle includes the following steps: S1: The data acquisition module collects the original vehicle operation data according to the acquisition frequency; The data acquisition module includes an OBD device, a CAN bus, and an in-vehicle terminal device; The original vehicle operation data includes real-time vehicle speed V, acceleration a, rotational speed f, throttle opening, gear data, and braking data.

[0021] The data acquisition module uploads the collected original vehicle driving data to the data processing module according to a certain acquisition frequency.

[0022] Collecting vehicle driving data from multiple perspectives, including not only the driving data of commercial vehicles but also the data of the surrounding environment, comprehensively collecting information, ensuring the integrity of the original vehicle operation data, and making the final analysis results more accurate and comprehensive.

[0023] S2: The data acquisition module sends the collected original vehicle operation data to the data processing module; Specifically, it includes the following steps: S21: The data processing module cleans the original vehicle operation data to obtain vehicle operation data; The cleaning includes removing outliers, duplicate data, and incomplete data; S22: The data processing module extracts key driving behavior feature data from the vehicle operation data; The key driving behavior feature data includes the speeding duration T n , the idling duration T m , the total low-speed time G, the total stable time H, the total number of frequent idlings P, the number of hard accelerations N, the number of hard brakings M, the number of smooth driving times Z, and the number of frequent brakings F; Speeding duration T n : The data processing module is provided with a speed threshold V1. The data processing module extracts the real-time vehicle speed V and the corresponding real-time time T from the original vehicle operation data. When V - V1 ≥ 0, the real-time vehicle speed V is speeding; When V - V1 changes from less than 0 to greater than 0, the real-time time T1 at this time is the starting time of the speeding time; when V - V1 changes from greater than 0 to less than 0, the real-time time T2 at this time is the ending time of the speeding time, and the speeding time is T2 - T1; The original vehicle operation data contains multiple sections of speeding time, and the total speeding time T n is the sum of multiple sections of speeding time, T n = (T2 1 - T1 1 ) + (T2 2 - T1 2 ) +... + (T2 n - T1 n ); Idling duration T m: When the real-time vehicle speed V becomes 0, the corresponding real-time time is recorded as the idling start point T3; when the real-time vehicle speed V changes from 0, the corresponding real-time time is marked as the idling end point T4, and the idling duration is recorded as T4 - T3; The original vehicle operation data includes multiple idling times, and the total idling time T m is the sum of multiple overspeed times, T m = (T4 1 - T3 1 ) + (T4 2 - T3 2 ) + …… + (T4 m - T3 m ); Total low-speed time G: The data processing module is provided with a threshold A. When the real-time vehicle speed V ≤ A, the current vehicle speed is marked as low speed. When the real-time vehicle speed V changes from greater than the threshold A to less than the threshold A, the corresponding time is marked as T5. When the real-time vehicle speed V changes from less than the threshold A to greater than the threshold A, the corresponding time is marked as T6. Then the low-speed time is T6 - T5, and the total low-speed time is G = (T6 1 - T5 1 ) + (T6 2 - T5 2 ) + …… + (T6 g - T5 g ); Total stable time H: When A ≤ V ≤ V1, the real-time vehicle speed V is marked as a stable vehicle speed. When the real-time vehicle speed V changes from less than A to greater than A, the corresponding time is marked as the stable start time T7. When the real-time vehicle speed V changes from greater than A to less than A, the corresponding time is marked as the stable start time T8. Then the stable time is T8 – T7, and the total stable time H = (T8 1 - T7 1 ) + (T8 2 - T7 2 ) + …… + (T8 h - T7 h ); Total number of frequent idlings P: The data processing module calculates the time T m between the idling end point T4 m+1 and the next idling start point T3 p . The data processing module is provided with a threshold C. When the T p ≤ C, it is marked as one frequent idling. The scoring calculation module counts the total number of frequent idlings P from the original vehicle driving data; Number of hard accelerations and number of hard brakes: Inside the data processing module, there is a threshold B. When the increase value △a1 of the acceleration a within 3 seconds is ≥ B, it is marked as one hard acceleration; when the decrease value △a2 of the acceleration a within 3 seconds is ≤ B, it is marked as one hard brake; The data processing module counts the number of hard accelerations N and the number of hard brakes M in the original vehicle operation data; Number of smooth driving times Z and number of frequent braking times F: The data processing module extracts the number of braking times L from the original vehicle driving data, and the scoring calculation module counts the number of braking times L and calculates the time difference T between adjacent brakings L , and the data processing module has thresholds D and E. When the time difference T L ≤ D, it is marked as one frequent braking. If the time difference T L ≥ E, it is marked as one smooth driving; The scoring calculation module counts the number of smooth driving times Z and the number of frequent braking times F; The scoring calculation module counts the overspeed duration T n , the idling duration T m , the total low-speed time G, the total stable time H, the total number of frequent idlings P, the number of hard accelerations N, the number of hard brakes M, the number of smooth driving times Z, and the number of frequent braking times F, and counts multi-faceted data to ensure the effectiveness and accuracy of calculating the driving behavior score.

[0024] S3: The data processing module sends the key driving behavior feature data to the scoring calculation module; S4: The scoring calculation module calculates the dynamic driving behavior score according to the feature data; S5: The scoring calculation module sends the score to the client platform.

[0025] The client platform displays the scores of the four items of vehicle speed control, throttle control, idling control, and braking control in the form of a four-dimensional score on the driving behavior dashboard. The scoring calculation module sends the scores of the four items of vehicle speed control, throttle control, idling control, and braking control and the comprehensive commercial vehicle driving behavior score to the client platform in real time, and the client continuously updates the scores of the four items of vehicle speed control, throttle control, idling control, and braking control and the comprehensive commercial vehicle driving behavior score in real time on the driving behavior dashboard; the client platform analyzes the long-term driving behavior trend to obtain a driving behavior analysis report.

[0026] By combining real-time and historical data, it supports real-time driving behavior scoring and provides long-term trend analysis at the same time, providing a driving behavior analysis report for drivers and administrators.

[0027] Further, step S4 includes the following steps: S41: The scoring calculation module calculates the scores of different dimensions of driving behavior based on the feature data; S42: The scoring calculation module adjusts the weight coefficient of environmental factors on the score in real time through a dynamic environment weight model.

[0028] Further, step S41 specifically includes the following: S411: The scoring calculation module scores the vehicle speed control; In the scoring standard, for every minute of speeding, 5 points are deducted; for each hard acceleration and each hard brake, 3 points are deducted; if the vehicle is driving at a low speed for more than 30 seconds, 4 points are deducted per minute; Vehicle speed control score calculation formula: Score_V = 100 – {T n *5 + (M + N) * 3 + G * 4}; The initial score of the vehicle speed control is 100 points; S412: The scoring calculation module scores the throttle control; In the scoring standard, for each hard acceleration, 5 points are deducted, and for each minute of smooth time, 10 points are added; Throttle control score calculation formula: Score_T = 100 – N * 5 + H * 10; The initial score of the throttle control is 100 points; S413: The scoring calculation module scores the idle speed control; In the scoring rule, for each minute of idle speed time T m 6 points are deducted, and for each frequent idle speed, 3 points are deducted; Idle speed control score calculation formula: Score_I = 100 – (T m *6 + P * 3); The initial score of the idle speed control is 100 points; S414: The scoring calculation module scores the brake control; In the scoring rule, for each hard brake, 5 points are deducted, and for each frequent brake, 3 points are deducted; for each smooth driving, 8 points are rewarded, with a maximum reward of 16 points; Brake control score calculation formula: When Z ≤ 2, Score_B = 100 – (M * 5 + F * 3 – Z * 8), when Z > 2, Score_B = 100 – (M * 5 + F * 3 – 16); The initial score of the brake control is 100 points; Currently, most commercial vehicles are automatic cars, and there is no need to consider the impact of gear data on driving behavior. Therefore, single-item scores are calculated separately for the following four aspects: speed control, throttle control, idle control, and brake control. In terms of speed control, considering the different impacts of overspeed time, hard braking times, hard acceleration times, and low-speed time on speed control, the single-item score of speed control is calculated; in terms of throttle control, considering the different impacts of hard acceleration times and smooth time on throttle control, the single-item score of throttle control is calculated; in terms of idle control, considering the different impacts of idle time and frequent idling on idle control, the single-item score of idle control is calculated; in terms of brake control, considering the different impacts of hard braking times, frequent braking, and smooth driving on brake control, the single-item score of brake control is calculated. In this way, the single-item scores of these four aspects are more meaningful and valuable.

[0029] S415: The scoring calculation module calculates the sum of the speed control score, throttle control score, idle control score, and brake control score.

[0030] Further, the specific content of step S415 is as follows: weighted calculation is performed on the four aspects of speed control, throttle control, idle control, and brake control. Among them, the proportion of speed control is 20%, the proportion of throttle control is 40%, the proportion of idle control is 15%, and the proportion of brake control is 25%. The comprehensive score total-score = (Score_V * 0.20) + (Score_T * 0.40) + (Score_I * 0.15) + (Score_B * 0.25) is obtained.

[0031] The determination of the weight distribution is based on: speed directly affects fuel consumption and driving safety, but commercial vehicles need to balance efficiency, so the weight is moderate; throttle use has the greatest impact on fuel consumption, and hard acceleration is the main cause of fuel waste, so the weight is the highest; idle has a certain impact on fuel consumption, but commercial vehicles stop frequently, so the weight is lower; frequent or hard braking increases fuel consumption and affects safety, and is equally important as speed control.

[0032] Through this weighted calculation method, the calculation scoring module can more accurately reflect the driving behavior of the driver.

[0033] Further, the specific steps of step S42 include the following steps: The scoring calculation module identifies different driving environments through GPS, high-precision maps, and traffic big data, and generates environmental data; the environmental data includes road characteristics, dynamic environmental characteristics, and time characteristics; S421: The scoring calculation module extracts road characteristics, dynamic environmental characteristics, and time characteristics, combines the three characteristics, and outputs an eight-dimensional vector; S422: The scoring calculation module defines a dynamic environment weight model, and the dynamic environment weight model establishes a road attention network and an environment attention network; S423: The scoring calculation module calculates the road weight and obtains the final score.

[0034] Furthermore, the scoring calculation module splices in the order of road features, dynamic environment features, and time features, and finally outputs an eight-dimensional feature vector; among them, the road features include one-hot encoding of road types, road curvature, and road slope, the dynamic environment features include weather severity, normalized visibility, and traffic density, and the time features include hour cycle encoding: converting 24 hours into a sine wave form and holiday flag; the eight-dimensional feature vector is output as: (one-hot encoding of road types, road curvature, road slope, weather severity, normalized visibility, traffic density, hour cycle encoding, holiday flag).

[0035] Furthermore, establish a road attention network: input road features, map three-dimensional features to an eight-dimensional space through linear transformation, filter negative values through the ReLU activation function, and finally output 5 road-related weights; The method of establishing the environment attention network is the same as that of establishing the road attention network, and the parameters are independent.

[0036] Furthermore, the specific content of step S423: Then calculate the road weight: take the first 3 dimensions of the eight-dimensional feature vector, obtain 5 road-related weights through the road attention network, namely road1, road2, road3, road4, and road5, and normalize them through softmax so that the sum of the 5 road-related weights is 1; calculate the environment weight: take the 3-6 dimensions of the eight-dimensional feature vector, obtain 5 environment-related weights through the environment attention network, namely env1, env2, env3, env4, and env5, compress them to the range of 0-1 through sigmoid and multiply by 2 to amplify the influence; The scoring calculation module fuses the road weight and the environment weight into a dynamic weight C = road1 * env1 + road2 + env2 + road3 * 1.2 + env3 * 1.5 + 1 - road4 + 1 - env4 + road5 / env5; Finally, the scoring calculation module multiplies the comprehensive score total-score by the dynamic weight C to obtain the dynamic driving behavior score. The scoring calculation module takes into account the commercial vehicle driving environment's influence on the driver's driving behavior. During driving, the scoring calculation module dynamically adjusts the dynamic weight C according to the environment where the commercial vehicle is located, so as to obtain the final score, which more accurately reflects the driver's driving behavior.

[0037] The present invention has been described by the above related embodiments. However, the above embodiments are only examples for implementing the present invention. It must be pointed out that the disclosed embodiments do not limit the scope of the present invention. On the contrary, modifications and refinements made without departing from the spirit and scope of the present invention fall within the scope of patent protection of the present invention.

Claims

1. A method for dynamically adjusting scores based on driving behavior of commercial vehicles, characterized in that: The following steps are involved: S1: The data acquisition module collects the original vehicle operation data according to the acquisition frequency; S2: The data acquisition module sends the collected original vehicle operation data to the data processing module; The specific steps include: S21: The data processing module cleans the original vehicle operation data to obtain vehicle operation data; S22: The data processing module extracts key driving behavior feature data from the vehicle operation data; The key driving behavior characteristic data includes the speeding duration T n , idling time T m , total low speed time G, total stable time H, total number of frequent idling times P, number of sudden acceleration times N, number of sudden braking times M, number of stable driving times Z and number of frequent braking times F; S3: The data processing module sends the key driving behavior feature data to the scoring calculation module; S4: The scoring calculation module calculates the driving behavior score according to the feature data; Using the dynamic environment weight model, the total score is dynamically adjusted according to the changes in the driving environment to obtain the driving behavior score; S5: The score calculation module sends the score to the client platform.

2. The method for dynamically adjusting the scoring based on the driving behavior of a commercial vehicle according to claim 1, characterized in that: The step S4 comprises the following steps: S41: The scoring calculation module calculates scores of different dimensions of driving behavior according to the feature data; S42: The score calculation module adjusts the weight coefficient of the environmental factor to the score in real time through a dynamic environmental weight model.

3. The method for dynamically adjusting the scoring based on the driving behavior of a commercial vehicle according to claim 2, characterized in that: The step S41 specifically includes the following contents: S411: The scoring calculation module scores the vehicle speed control; In the scoring criteria, 5 points will be deducted for every minute of speeding; 3 points will be deducted for each sudden acceleration and each sudden braking; 4 points will be deducted for every minute the vehicle is driven at a low speed; Speed ​​control score calculation formula: Score_V = 100 – { T n *5+(M+N)*3+G*4}; S412: The scoring calculation module scores the throttle control; In the scoring criteria, 5 points are deducted for each sudden acceleration, and 10 points are added for each minute of stable time; Throttle control score calculation formula: Score_T = 100 –N*5+H*10; S413: The scoring calculation module scores the idle speed control; In the scoring rules, every minute of idling time T m 6 points will be deducted, and 3 points will be deducted for each frequent idling. Idle speed control score calculation formula: Score_I = 100 – (T m *6+P*3); S414: The scoring calculation module scores the braking control; In the scoring rules, 5 points will be deducted for each sudden braking, and 3 points will be deducted for each frequent braking; 8 points will be awarded for each smooth driving, with a maximum of 16 points; Braking control score calculation formula: when Z≤2, Score_B =100 – (M*5+F*3 –Z*8); when Z>2, Score_B =100 – (M*5+F*3 –16); S415: The score calculation module calculates the sum of the vehicle speed control score, the throttle control score, the idle speed control score and the brake control score.

4. The method for dynamically adjusting the scoring based on the driving behavior of a commercial vehicle according to claim 3, characterized in that: The specific content of step S415 is as follows: A weighted calculation is performed on the four items of vehicle speed control, throttle control, idle speed control and brake control, among which vehicle speed control accounts for 20%, throttle control accounts for 40%, idle speed control accounts for 15%, and brake control accounts for 25%. The comprehensive score total-score = (Score_V * 0.20) + (Score_T * 0.40) + (Score_I * 0.15) + (Score_B * 0.25) is obtained.

5. The method for dynamically adjusting the scoring based on the driving behavior of a commercial vehicle according to claim 2, characterized in that: The step S42 specifically includes the following steps: S421: The scoring calculation module extracts road features, dynamic environment features, and time features, and combines the three features to output an eight-dimensional vector; S422: The scoring calculation module defines a dynamic environment weight model, and the dynamic environment weight model establishes a road attention network and an environment attention network; S423: The score calculation module calculates the road weight and obtains the final score.

6. The method for dynamically adjusting the scoring based on the driving behavior of a commercial vehicle according to claim 5, characterized in that: The specific content of step S421 is as follows: The scoring calculation module sequentially splices road features, dynamic environment features and time features, and finally outputs an eight-dimensional feature vector; The road features include unique hot encoding of road type, road curvature and road slope; the dynamic environment features include weather severity, normalized visibility and traffic density; the time features include hourly cycle encoding: converting 24 hours into a sine wave form and holiday signs.

7. The method for dynamically adjusting the scoring based on the driving behavior of a commercial vehicle according to claim 5, characterized in that: The specific content of step S422 is as follows: Establish a road attention network: input road features, map the three-dimensional features to eight-dimensional space through linear transformation, filter negative values ​​through ReLU activation function, and finally output 5 road-related weights; The method of establishing the environment attention network is consistent with that of establishing the road attention network, and the parameters are independent.

8. The method for dynamically adjusting the scoring based on the driving behavior of a commercial vehicle according to claim 5, characterized in that: The specific content of step S423 is as follows: Then calculate the road weight: take the first 3 dimensions of the 8-dimensional feature vector, and use the road attention network to obtain 5 road-related weights, namely road1, road2, road3, road4 and road5. After softmax normalization, the sum of the 5 road-related weights is 1; calculate the environment weight: take dimensions 3-6 of the 8-dimensional feature vector, and use the environment attention network to obtain 5 environment-related weights, namely env1, env2, env3, env4 and env5. After sigmoid compression to the range of 0-1, multiply by 2 to amplify the impact; The scoring calculation module combines the road weight and the environment weight into a dynamic weight C=road1*env1+road2+env2+road3*1.2+env3*1.5+1-road4+1-env4+road5 / env5; Finally, the score calculation module multiplies the comprehensive score total-score by the dynamic weight C to obtain the dynamic driving behavior score.

Images