Driving style evaluation method based on real vehicle data

Abstract

The invention belongs to the technical field of real vehicle test data processing of an electromechanical composite transmission system of a special vehicle and driving behavior evaluation based on data processing, and particularly relates to a driving style evaluation method based on real vehicle data. Comprising the steps of real vehicle test data conversion, acceleration frequency statistics and intensity calculation data processing, steering frequency statistics and intensity calculation data processing and driving style comprehensive evaluation. According to the method, key behavior characteristics such as acceleration times, acceleration intensity and steering modes can be extracted from acceleration and steering test data, and the driving habits of a driver are quantified by analyzing the data trend and slope. The method not only improves the accuracy of driving style analysis, but also can provide data support for performance optimization, driving behavior monitoring, driver training and safety improvement of special vehicles. Through deep analysis of real vehicle data, potential driving problems can be better identified, and a vehicle control strategy is optimized, so that the use efficiency and safety of the special vehicle are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of real vehicle test data processing of special vehicle electromechanical composite transmission system and driving behavior evaluation based on data processing, and specifically relates to a driving style evaluation method based on real vehicle data, which is suitable for special vehicle electromechanical composite system data processing and driving style evaluation. Background Art

[0002] With the widespread use of special vehicles in various complex environments, how to improve the driving efficiency, operational safety and control performance of these vehicles has become an urgent problem to be solved. Special vehicles usually have unique powertrain systems and usage scenarios, and their driving behaviors are quite different from those of ordinary vehicles. Therefore, evaluating the driving style of the driver during the operation of special vehicles can provide an important reference for optimizing vehicle performance, improving safety and operational stability.

[0003] Existing driving style assessment methods mainly rely on subjective reports from drivers or simulated driving experiments, and lack accurate analysis of actual driving data. These methods cannot reflect the driver's operating behavior in real time and comprehensively, especially in the complex and changeable working environment of special vehicles, and often cannot effectively capture the driver's actual driving habits and behavior patterns. Therefore, driving style assessment methods based on real vehicle data have become an effective way to solve this problem.

[0004] At present, some driving style assessment methods based on vehicle sensor data have been applied to the field of conventional vehicles, but for the electromechanical hybrid transmission system of special vehicles, its particularity makes it difficult for existing technologies to be directly applied. The driving style assessment of special vehicles not only needs to consider basic behaviors such as acceleration, braking, and steering, but also needs to comprehensively consider the impact of the electromechanical hybrid transmission system on the vehicle's handling characteristics. Therefore, how to accurately assess the driver's driving style based on the actual vehicle data of special vehicles has become a technical problem that needs to be solved urgently.

[0005] In addition, existing driving style assessment technologies usually focus on the analysis of a single behavior, such as acceleration or braking, but rarely comprehensively analyze the interaction between the driver's acceleration, steering and other multiple behaviors, resulting in limitations and poor accuracy of the assessment results. To address this problem, the present invention proposes a comprehensive assessment method based on real vehicle data, which deeply mines the driver's actual driving style from the data level through trend analysis and slope calculation of acceleration, steering and other behaviors, and can provide more accurate and comprehensive assessment results. Summary of the invention

[0006] 1. Technical issues to be resolved

[0007] The technical problem to be solved by the present invention is: how to effectively process and analyze a large amount of complex test data during the actual vehicle test of the electromechanical hybrid transmission system of a special vehicle, especially in the process of driving style evaluation, the challenges of trend analysis of test data, behavior pattern recognition and data interaction relationship extraction. The present invention solves the problems of insufficient accuracy of special vehicle driving style evaluation, high data processing complexity, and incomplete driving behavior analysis in existing methods by designing a driving style evaluation method based on actual vehicle data and adopting technical means of data trend analysis, slope calculation and multi-dimensional behavior feature comprehensive evaluation.

[0008] This method accurately captures the driver's operating mode and habits through detailed trend analysis and slope calculation of acceleration and steering data, and comprehensively evaluates the driver's key driving characteristics such as acceleration method and steering behavior based on real vehicle data. At the same time, this method can identify potential behavioral problems of drivers through in-depth analysis of data, providing strong support for performance optimization, driving behavior monitoring and safety improvement of special vehicles. By establishing a complete data processing process and behavior evaluation method, it can improve the ability to identify driving behavior under complex conditions and optimize the vehicle's control strategy.

[0009] The technical solution of the present invention establishes a full-process data processing and driving style assessment tool chain for special vehicle electromechanical hybrid transmission systems. Through in-depth mining and analysis of actual vehicle test data, accurate assessment of driving style can be achieved, which can effectively improve the accuracy and efficiency of driver behavior monitoring and safety management.

[0010] (II) Technical solution

[0011] In order to solve the above technical problems, the present invention provides a driving style evaluation method based on real vehicle data, the method comprising the following steps:

[0012] Step 1: Real vehicle test data conversion;

[0013] Step 2: Acceleration times statistics and strength calculation data processing;

[0014] Step 3: Data processing of turning number statistics and intensity calculation;

[0015] Step 4: Comprehensive assessment of driving style.

[0016] Among them, in the step 1, the operating data of the mechatronic transmission system under different working conditions are collected from the actual vehicle test process; the format of the original data collected from the actual vehicle is usually diverse and not uniform, so format conversion is required to ensure that all data have a consistent format. The conversion process includes timestamp synchronization, multi-source data fusion, and data regeneration to obtain the accelerator pedal position curve and the steering wheel steering angle curve.

[0017] Wherein, in said step 1, the operation data of the electromechanical hybrid transmission system under different working conditions are collected from the actual vehicle test process; the data include: an accelerator pedal position signal and a steering wheel angle signal.

[0018] Wherein, in said step 2, acceleration number statistics and intensity calculation data processing are performed;

[0019] Because the driver's acceleration behavior is completed by stepping on the accelerator pedal, the depth and speed of the driver's pedaling will be reflected in the collected acceleration curve with differences in fluctuation amplitude and speed. By analyzing the acceleration curve, the number of accelerations and the acceleration intensity are analyzed; for the accelerator pedal position curve obtained in step 1, a sliding window combined with a threshold analysis method is used to analyze the general trend of the accelerator pedal position curve and determine the upward and downward trends; according to the judgment result, the next trend under the general trend is searched downward based on the timestamp, and the sliding window method is also used to find the position of the trend change, find the maximum or minimum value of the trend change position, record the position point of the maximum or minimum value, calculate the time difference between the initial point and the maximum or minimum point, and calculate the size of each acceleration slope in combination with the value of the maximum or minimum point.

[0020] Among them, in step 2, the process of determining the general trend of data is as follows:

[0021] First, define the size of the sliding window for detecting local trends. The default sliding window size is 15. Define the window size for calculating local trends. The default is 15. Define the size of the threshold for representing data fluctuations for trend analysis. The default value of the threshold is 8. Define the minimum threshold for the slope, which represents the slope threshold for the descending process. The default value is 0. Define the maximum threshold for the slope, which represents the slope threshold for the ascending process. The default value is 0.

[0022] Initialize the number of recorded rising and falling processes, the slope of each rising and falling process, the data length, and the starting index of the data;

[0023] Enter the loop process, the judgment condition is that the sliding window will not exceed the range of the data, and then extract a value with a length of the sliding window size from the data, and the index starts from the initial position to the end position of the data;

[0024] Check the fluctuation in the current window, that is, whether the difference between the maximum and minimum values ​​exceeds the set threshold. If it exceeds the threshold, it means that there is significant fluctuation in the window, and further trend analysis is needed to determine the upward or downward trend.

[0025] If the average value of the data difference in the window is greater than the set maximum slope threshold, it is considered to be an upward trend. The extreme value function is called to find the maximum value point in the data, that is, an upward trend. The upward trend count is increased by 1, indicating that an upward process is detected; the slope of the upward process is added to the array, and the index is updated to the next index of the maximum value. This is used as a new starting point to continue the subsequent analysis;

[0026] If the average value of the data difference in the window is less than the set minimum slope threshold, it is considered to be a downward trend. The extreme value function is called to find the minimum point in the data, that is, the downward trend. The downward trend count is increased by 1, indicating that a downward process is detected; the slope of the downward process is added to the array, and the index is updated to the next index of the minimum value. This is used as a new starting point to continue the subsequent analysis;

[0027] If the fluctuation of the window is not large or the fluctuation of the window does not exceed the threshold, or it is not clear whether it is an upward or downward trend, the program increases the index by 1 and continues to analyze the next window.

[0028] Wherein, in step 2, the method for finding the extreme value is:

[0029] First, initialize the extreme value index found in the data trend judgment, and find the data window starting from the index position with the inspection window size as the size, which is used to compare with the current window when looking for trends;

[0030] Loop search, starting from the initial index position and using the initial index position data window as the window size, until the end of the data is searched;

[0031] If an upward trend is found in the data trend judgment, check whether the data in the current window is in an upward trend, calculate the difference between adjacent data points, and if the mean of the difference is greater than 0, it means that the data in the window is in an upward trend;

[0032] Check the next window in a loop. If the average value of the difference between adjacent data in the window is greater than 0, it means that the trend is still rising. The window data is stored in the storage array. If the storage array stores more than 2 windows of window data, keep the most recent 2 windows. If the current window trend is falling and the difference between the maximum and minimum values ​​in the storage array is less than 12, ignore the current falling trend and continue to search for the rising trend.

[0033] Find the maximum value of the current window and the previous window, record the maximum value index, and calculate the average rate of change from the starting index to the extreme value found, which is the rising slope;

[0034] If a downward trend is found in the data trend judgment, check whether the data in the current window is in a downward trend, and calculate the difference between adjacent data points. If the mean of the difference is less than 0, it means that the data in the window is in a downward trend;

[0035] Check the next window in a loop. If the average value of the difference between adjacent data in the window is less than 0, it means that the trend is still downward. Store the window data in the storage array. If the storage array stores more than 2 windows of window data, keep the most recent 2 windows. If the current window trend is upward and the difference between the maximum and minimum values ​​in the storage array is less than 12, ignore the current upward trend and continue to search for the downward trend.

[0036] Find the minimum value of the current window and the previous window, record the minimum value index, and calculate the average rate of change from the starting index to the extreme value found, which is the descending slope;

[0037] If no trend change point is found in the loop, continue searching forward. If no trend end point is found in the entire data sequence, return the maximum or minimum value of the initial window, and return the corresponding extreme value and average slope.

[0038] Wherein, in said step 3, data processing of turning times statistics and intensity calculation is performed;

[0039] The driver's steering behavior is completed by turning the steering wheel, and the difference between left-turn and right-turn steering wheel reflected in the collected steering wheel angle is the difference between positive and negative; the depth and speed of the driver's steering wheel turning are reflected in the collected steering curve, which will have differences in fluctuation amplitude and speed. By analyzing the steering wheel steering angle curve, the number of left and right turns and the steering intensity are analyzed;

[0040] Because the steering wheel steering angle curve contains the center steering curve content, it is necessary to remove the center steering data from the steering wheel steering angle curve data. After removal, the sliding window analysis method is used again to analyze and find the steering position point. Considering the driving behavior characteristics, the starting zero window position is found, and the timestamp of the position is recorded. Search downward along the timestamp to find a sliding window with non-zero data. After finding it, continue to search downward along the timestamp until the next all-zero window is found, and record the position point of the window; calculate the time difference between the initial position point and the position point, find the extreme values ​​of the initial position and the position point, and thus calculate the slope size of the left turn and the right turn.

[0041] Among them, in step 3, the method of eliminating the center steering data is:

[0042] First, drive through the center turn data analysis, and obtain the center turn position point data. First, initialize the variable values, including the number of left turns, the number of right turns, and the array used to store the steering slope;

[0043] Loop through each location where the center turn occurs, use the sliding window to find the area with a value of 0 and clear the value to zero, use the sliding window to check whether the window is completely zero. If it is zero, skip and find the next non-zero area. After finding the non-zero area, continue to find the next zero window area, that is, find the end point of the non-zero area, clear the values ​​corresponding to the data points from the initial point to the end point, thereby eliminating the center turn data as a new array;

[0044] In step 3, the method for calculating the number of turns and the intensity is:

[0045] The new array is traversed using a sliding window method. The size of each sliding window is the default value. If the first element of the window is 0 and the subsequent elements are non-zero, it indicates the beginning of the turn. The starting position is recorded, and then the sliding window is continued to search downward in a loop until a window of all zeros is found, indicating the end of the turn.

[0046] Calculate the number of turns and the turn slope. For each turn window, first check whether there are negative turns and positive turns. If the number of negative turn elements is greater than 5, check whether the minimum value of the window is less than the set left turn threshold. If so, calculate the slope of the left turn. If the number of positive turn elements is greater than 5, check whether the maximum value of the window is greater than the set right turn threshold. If so, calculate the slope of the right turn.

[0047] Wherein, in said step 4, a comprehensive evaluation of driving style is performed;

[0048] Based on the calculation results of steps 2 and 3, a weighted model is constructed to consider the relative importance of each indicator; each factor including the number of pedaling times, pedaling slope, number of steering times, and steering slope is converted into a style score, and then a weighted sum is performed.

[0049] In step 4, a standardized scoring system is established for each indicator, usually ranging from 0 to 1; each indicator is scored as follows:

[0050] Pedal frequency score: The higher the frequency of pedaling, the higher the score;

[0051] Pedaling slope score: the greater the slope, the higher the score;

[0052] Turn number score: The more turns you make, the higher the score;

[0053] Turning slope score: The greater the turning slope, the higher the score;

[0054] Assume that each factor has a weight, indicating its influence on the driving style score; the driver's overall driving style score S is:

[0055] S=w1·F pedal +w2·G pedal +w3·F steering +w4·G steering

[0056] Among them, F pedal and G pedal They are the pedaling times score and the pedaling slope score; F steering and G steering are the turning number score and turning slope score respectively; w1, w2, w3, w4 are the weights of each factor;

[0057] By adjusting the weight coefficients w1, w2, w3, and w4, you can adjust the degree of attention paid to acceleration, braking, and steering respectively; if you pay attention to the smoothness of driving, increase the weights of the pedal slope and the steering slope; if you pay attention to the impatience of driving, increase the weights of the number of pedaling times and the number of steering times.

[0058] (III) Beneficial effects

[0059] Compared with the prior art, the technical solution of the present invention includes real vehicle test data conversion, acceleration number statistics and intensity calculation data processing method, turn number statistics and intensity calculation data processing method, and multiple driving style comprehensive evaluation method. This solution is aimed at how to effectively process and analyze a large amount of complex test data during the real vehicle test of the electromechanical composite transmission system of special vehicles, especially in the process of driving style evaluation, the challenges of trend analysis of test data, behavior pattern recognition and data interaction relationship extraction. By designing a driving style evaluation method based on real vehicle data, using technical means of data trend analysis, slope calculation and multi-dimensional behavior feature comprehensive evaluation, the problems of insufficient accuracy of special vehicle driving style evaluation, high data processing complexity, and incomplete driving behavior analysis in existing methods are solved.

[0060] The beneficial effects of the present invention are: it can extract key behavioral features from acceleration and steering test data, and quantify the driver's driving habits by analyzing data trends and slopes. This evaluation method not only improves the accuracy of driving style analysis, but also provides data support for performance optimization, driving behavior monitoring, driver training and safety improvement of special vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 It is a test data diagram of acceleration and steering (including center steering) in the present invention.

[0062] Figure 2It is the logic diagram of the major trend analysis in the present invention.

[0063] Figure 3 It is the logic diagram for finding the extreme value in the present invention.

[0064] Figure 4 It is a logic diagram of excluding central steering data in the steering data analysis of the present invention.

[0065] Figure 5 It is the data logic diagram of the redirection data in the present invention.

[0066] Figure 6 It is a flow chart of the technical solution of the present invention. DETAILED DESCRIPTION

[0067] In order to make the purpose, content, and advantages of the present invention more clear, the specific implementation methods of the present invention are further described in detail below in conjunction with the accompanying drawings and examples.

[0068] In order to solve the above technical problems, the present invention provides a driving style evaluation method based on real vehicle data, the method comprising the following steps:

[0069] Step 1: Real vehicle test data conversion;

[0070] Step 2: Acceleration times statistics and strength calculation data processing;

[0071] Step 3: Data processing of turning number statistics and intensity calculation;

[0072] Step 4: Comprehensive assessment of driving style.

[0073] Among them, in the step 1, the operating data of the mechatronic transmission system under different working conditions are collected from the actual vehicle test process; the format of the original data collected from the actual vehicle is usually diverse and not uniform, so format conversion is required to ensure that all data have a consistent format. The conversion process includes timestamp synchronization, multi-source data fusion, and data regeneration to obtain the accelerator pedal position curve and the steering wheel steering angle curve.

[0074] Wherein, in said step 1, the operation data of the electromechanical hybrid transmission system under different working conditions are collected from the actual vehicle test process; the data include: an accelerator pedal position signal and a steering wheel angle signal.

[0075] Wherein, in said step 2, acceleration number statistics and intensity calculation data processing are performed;

[0076] Because the driver's acceleration behavior is completed by stepping on the accelerator pedal, the depth and speed of the driver's pedaling will be reflected in the collected acceleration curve with differences in fluctuation amplitude and speed. By analyzing the acceleration curve, the number of accelerations and the acceleration intensity are analyzed; for the accelerator pedal position curve obtained in step 1, a sliding window combined with a threshold analysis method is used to analyze the general trend of the accelerator pedal position curve and determine the upward and downward trends; according to the judgment result, the next trend under the general trend is searched downward based on the timestamp, and the sliding window method is also used to find the position of the trend change, find the maximum or minimum value of the trend change position, record the position point of the maximum or minimum value, calculate the time difference between the initial point and the maximum or minimum point, and calculate the size of each acceleration slope in combination with the value of the maximum or minimum point.

[0077] Among them, in step 2, the process of determining the general trend of the data is as follows:

[0078] First, define the size of the sliding window for detecting local trends. The default sliding window size is 15. Define the window size for calculating local trends. The default is 15. Define the size of the threshold for representing data fluctuations for trend analysis. The default value of the threshold is 8. Define the minimum threshold for the slope, which represents the slope threshold for the descending process. The default value is 0. Define the maximum threshold for the slope, which represents the slope threshold for the ascending process. The default value is 0.

[0079] Initialize the number of recorded rising and falling processes, the slope of each rising and falling process, the data length, and the starting index of the data;

[0080] Enter the loop process, the judgment condition is that the sliding window will not exceed the range of the data, and then extract a value with a length of the sliding window size from the data, and the index starts from the initial position to the end position of the data;

[0081] Check the fluctuation in the current window, that is, whether the difference between the maximum and minimum values ​​exceeds the set threshold. If it exceeds the threshold, it means that there is significant fluctuation in the window, and further trend analysis is needed to determine the upward or downward trend.

[0082] If the average value of the data difference in the window is greater than the set maximum slope threshold, it is considered to be an upward trend. The extreme value function is called to find the maximum value point in the data, that is, an upward trend. The upward trend count is increased by 1, indicating that an upward process is detected; the slope of the upward process is added to the array, and the index is updated to the next index of the maximum value. This is used as a new starting point to continue the subsequent analysis;

[0083] If the average value of the data difference in the window is less than the set minimum slope threshold, it is considered to be a downward trend. The extreme value function is called to find the minimum point in the data, that is, the downward trend. The downward trend count is increased by 1, indicating that a downward process is detected; the slope of the downward process is added to the array, and the index is updated to the next index of the minimum value. This is used as a new starting point to continue the subsequent analysis;

[0084] If the fluctuation of the window is not large or the fluctuation of the window does not exceed the threshold, or it is not clear whether it is an upward or downward trend, the program increases the index by 1 and continues to analyze the next window.

[0085] Wherein, in step 2, the method for finding the extreme value is:

[0086] First, initialize the extreme value index found in the data trend judgment, and find the data window starting from the index position with the inspection window size as the size, which is used to compare with the current window when looking for trends;

[0087] Loop search, starting from the initial index position and using the initial index position data window as the window size, until the end of the data is searched;

[0088] If an upward trend is found in the data trend judgment, check whether the data in the current window is in an upward trend, calculate the difference between adjacent data points, and if the mean of the difference is greater than 0, it means that the data in the window is in an upward trend;

[0089] Check the next window in a loop. If the average value of the difference between adjacent data in the window is greater than 0, it means that the trend is still rising. The window data is stored in the storage array. If the storage array stores more than 2 windows of window data, keep the most recent 2 windows. If the current window trend is falling and the difference between the maximum and minimum values ​​in the storage array is less than 12, ignore the current falling trend and continue to search for the rising trend.

[0090] Find the maximum value of the current window and the previous window, record the maximum value index, and calculate the average rate of change from the starting index to the extreme value found, which is the rising slope;

[0091] If a downward trend is found in the data trend judgment, check whether the data in the current window is in a downward trend, and calculate the difference between adjacent data points. If the mean of the difference is less than 0, it means that the data in the window is in a downward trend;

[0092] Check the next window in a loop. If the average value of the difference between adjacent data in the window is less than 0, it means that the trend is still downward. Store the window data in the storage array. If the storage array stores more than 2 windows of window data, keep the most recent 2 windows. If the current window trend is upward and the difference between the maximum and minimum values ​​in the storage array is less than 12, ignore the current upward trend and continue to search for the downward trend.

[0093] Find the minimum value of the current window and the previous window, record the minimum value index, and calculate the average rate of change from the starting index to the extreme value found, which is the descending slope;

[0094] If no trend change point is found in the loop, continue searching forward. If no trend end point is found in the entire data sequence, return the maximum or minimum value of the initial window, and return the corresponding extreme value and average slope.

[0095] Wherein, in said step 3, data processing of turning times statistics and intensity calculation is performed;

[0096] The driver's steering behavior is completed by turning the steering wheel, and the difference between left-turn and right-turn steering wheel reflected in the collected steering wheel angle is the difference between positive and negative; the depth and speed of the driver's steering wheel turning are reflected in the collected steering curve, which will have differences in fluctuation amplitude and speed. By analyzing the steering wheel steering angle curve, the number of left and right turns and the steering intensity are analyzed;

[0097] Because the steering wheel steering angle curve contains the center steering curve content, it is necessary to remove the center steering data from the steering wheel steering angle curve data. After removal, the sliding window analysis method is used again to analyze and find the steering position point. Considering the driving behavior characteristics, the starting zero window position is found, and the timestamp of the position is recorded. Search downward along the timestamp to find a sliding window with non-zero data. After finding it, continue to search downward along the timestamp until the next all-zero window is found, and record the position point of the window; calculate the time difference between the initial position point and the position point, find the extreme values ​​of the initial position and the position point, and thus calculate the slope size of the left turn and the right turn.

[0098] Among them, in step 3, the method of eliminating the center steering data is:

[0099] First, drive through the center turn data analysis, and obtain the center turn position point data. First, initialize the variable values, including the number of left turns, the number of right turns, and the array used to store the steering slope;

[0100] Loop through each location where the center turn occurs, use the sliding window to find the area with a value of 0 and clear the value to zero, use the sliding window to check whether the window is completely zero. If it is zero, skip and find the next non-zero area. After finding the non-zero area, continue to find the next zero window area, that is, find the end point of the non-zero area, clear the values ​​corresponding to the data points from the initial point to the end point, thereby eliminating the center turn data as a new array;

[0101] In step 3, the method for calculating the number of turns and the intensity is:

[0102] The new array is traversed using a sliding window method. The size of each sliding window is the default value. If the first element of the window is 0 and the subsequent elements are non-zero, it indicates the beginning of the turn. The starting position is recorded, and then the sliding window is continued to search downward in a loop until a window of all zeros is found, indicating the end of the turn.

[0103] Calculate the number of turns and the turn slope. For each turn window, first check whether there are negative turns and positive turns. If the number of negative turn elements is greater than 5, check whether the minimum value of the window is less than the set left turn threshold. If so, calculate the slope of the left turn. If the number of positive turn elements is greater than 5, check whether the maximum value of the window is greater than the set right turn threshold. If so, calculate the slope of the right turn.

[0104] Wherein, in said step 4, a comprehensive evaluation of driving style is performed;

[0105] Based on the calculation results of steps 2 and 3, a weighted model is constructed to consider the relative importance of each indicator; each factor including the number of pedaling times, pedaling slope, number of steering times, and steering slope is converted into a style score, and then a weighted sum is performed.

[0106] In step 4, a standardized scoring system is established for each indicator, usually ranging from 0 to 1; each indicator is scored as follows:

[0107] Pedal frequency score: The higher the frequency of pedaling, the higher the score;

[0108] Pedaling slope score: the greater the slope, the higher the score;

[0109] Turn number score: The more turns you make, the higher the score;

[0110] Turning slope score: The greater the turning slope, the higher the score;

[0111] Assume that each factor has a weight, indicating its influence on the driving style score; the driver's overall driving style score S is:

[0112] S=w1·F pedal +w2·G pedal +w3·F steering +w4·G steering

[0113] Among them, F pedal and G pedal They are the pedaling times score and the pedaling slope score; F steering and G steering are the turning number score and turning slope score respectively; w1, w2, w3, w4 are the weights of each factor;

[0114] By adjusting the weight coefficients w1, w2, w3, and w4, you can adjust the degree of attention paid to acceleration, braking, and steering respectively; if you pay attention to the smoothness of driving, increase the weights of the pedal slope and the steering slope; if you pay attention to the impatience of driving, increase the weights of the number of pedaling times and the number of steering times.

[0115] Example 1

[0116] In order to solve the challenges of how to effectively process and analyze a large amount of complex test data during the actual vehicle test of the electromechanical hybrid transmission system of existing special vehicles, especially in the process of driving style evaluation, the trend analysis of test data, behavior pattern recognition and data interaction relationship extraction, etc., the present invention provides a driving style evaluation method based on actual vehicle data, which includes the following contents:

[0117] (1) Actual vehicle test data conversion

[0118] 1) Data collection

[0119] First, the operating data of the electromechanical hybrid transmission system under different working conditions are collected during the actual vehicle test. The data includes: accelerator pedal position signal EMT_Sa, steering wheel angle signal EMT_Ss.

[0120] These data are collected through various sensors and data acquisition systems, and transmitted through standard communication protocols (such as CAN or FlexRay bus).

[0121] 2) Data format conversion

[0122] The raw data formats collected from real vehicles are usually diverse and not uniform, so format conversion is required to ensure that all data has a consistent format. The main data processing methods are as follows:

[0123] Timestamp synchronization: The sampling frequencies of various sensors are different. First, all data need to be timestamped to ensure the time synchronization of data from different sensors.

[0124] (2) Acceleration statistics and strength calculation data processing method

[0125] 1) Methods for determining the big trend of data

[0126] First, define the size of the sliding window for detecting local trends. The default sliding window size is 15. Define the window size for calculating local trends. The default is 15. Define the size of the data fluctuation threshold for trend analysis. The default value of the threshold is 8. Define the minimum threshold of the slope, which represents the slope threshold of the descending process. The default value is 0. Define the maximum threshold of the slope, which represents the slope threshold of the ascending process. The default value is 0.

[0127] Initialize the number of times the rising and falling processes are recorded, the slope of each rising and falling process, the data length, and the starting index of the data.

[0128] Entering the loop process, the judgment condition is that the sliding window will not exceed the range of the data, and then extracting a value with a length equal to the sliding window size from the data, and the index starts from the initial position to the end position of the data.

[0129] Check whether the fluctuation in the current window (i.e., the difference between the maximum and minimum values) exceeds the set threshold. If it exceeds the threshold, it means that there is significant fluctuation in the window, and further trend analysis is needed to determine the upward or downward trend.

[0130] If the average value of the data difference in the window is greater than the set maximum slope threshold, it is considered to be an upward trend, and the extreme value function is called to find the maximum value point (upward trend) in the data. The upward trend count is increased by 1, indicating that an upward process is detected. The slope of the upward process is added to the array, and the index is updated to the next index of the maximum value. This is used as a new starting point to continue the subsequent analysis.

[0131] If the average value of the data difference in the window is less than the set minimum slope threshold, it is considered to be a downward trend, and the extreme value function is called to find the minimum point (downward trend) in the data. The downward trend count is increased by 1, indicating that a downward process is detected. The slope of the downward process is added to the array, and the index is updated to the next index of the minimum value. This is used as a new starting point to continue the subsequent analysis.

[0132] If the fluctuation of the window is not large or the fluctuation of the window does not exceed the threshold, or it is not clear whether it is an upward or downward trend, the program increases the index by 1 and continues to analyze the next window.

[0133] 2) Extreme value search method

[0134] First, the extreme value index found in the judgment of the data trend is initialized, and the data window is found from the index position with the inspection window size as the size, which is used to compare with the current window when looking for trends.

[0135] The search is looped, starting from the initial index position and taking the initial index position data window as the window size, until the end of the data is found.

[0136] If an upward trend is found in the data trend judgment, check whether the data in the current window is in an upward trend, calculate the difference between adjacent data points, and if the mean of the difference is greater than 0, it means that the data in the window is in an upward trend.

[0137] Check the next window in a loop. If the average value of the difference between adjacent data in the window is greater than 0, it means that the trend is still rising. The window data is stored in the storage array. If the storage array stores more than 2 windows of window data, keep the most recent 2 windows. If the current window trend is falling and the difference between the maximum and minimum values ​​in the storage array is less than 12, ignore the current falling trend and continue to search for the rising trend.

[0138] Find the maximum value of the current window and the previous window, record the maximum value index, and calculate the average rate of change from the start index to the extreme value found, which is the rising slope.

[0139] If a downward trend is found in the data trend judgment, check whether the data in the current window is in a downward trend, calculate the difference between adjacent data points, and if the mean of the difference is less than 0, it means that the data in the window is in a downward trend.

[0140] Check the next window in a loop. If the average value of the difference between adjacent data in the window is less than 0, it means that the trend is still falling. Store the window data in the storage array. If the storage array stores more than 2 windows of window data, keep the most recent 2 windows. If the current window trend is rising and the difference between the maximum and minimum values ​​in the storage array is less than 12, ignore the current rising trend and continue to search for a falling trend.

[0141] Find the minimum value of the current window and the previous window, record the minimum index, and calculate the average rate of change from the starting index to the extreme value found, which is the descending slope.

[0142] If no trend change point is found in the loop, continue searching forward. If no trend end point is found in the entire data sequence, return the maximum or minimum value of the initial window, and return the corresponding extreme value and average slope.

[0143] (3) Data processing method for turning number statistics and intensity calculation

[0144] 1) Eliminate center turn data method

[0145] First, drive through the center turn data analysis and obtain the center turn position point data. First, initialize the variable values, including the number of left turns, the number of right turns, and the array for storing the steering slope.

[0146] Loop through each location where a center turn occurs, use the sliding window to find an area with a value of 0 and clear the value to zero, and use the sliding window to check whether the window is completely zero. If it is zero, skip and find the next non-zero area. After finding the non-zero area, continue to find the next zero window area, that is, find the end point of the non-zero area, and clear the values ​​corresponding to the data points from the initial point to the end point, thereby eliminating the center turn data as a new array.

[0147] 2) Turning frequency statistics and intensity calculation method

[0148] The new array is traversed using the sliding window method. The size of each sliding window is the default value. If the first element of the window is 0 and the subsequent elements are non-zero, it indicates the beginning of the turn. The starting position is recorded, and then the sliding window method is continued to search downward in a loop until a window of all zeros is found, indicating the end of the turn.

[0149] Calculate the number of turns and the turn slope. For each turn window, first check whether there are negative turns and positive turns. If the number of negative turn elements is greater than 5, check whether the minimum value of the window is less than the set left turn threshold. If so, calculate the slope of the left turn. aSteerSlope = (-min_val) * turn sampling rate / minimum index. If the number of positive turn elements is greater than 5, check whether the maximum value of the window is greater than the set right turn threshold. If so, calculate the slope of the right turn. bSteerSlope = (max_val) * turn sampling rate / maximum index.

[0150] (4) Comprehensive evaluation method of driving style

[0151] 1) Pedal and steering behavior

[0152] The number of pedal strokes refers to the number of times the driver operates the accelerator, brake or clutch pedals per unit time. This indicator mainly reflects the driver's control frequency of acceleration and braking.

[0153] High pedaling frequency: usually means that the driver accelerates or brakes more frequently and may tend to accelerate or brake suddenly, which is a more aggressive driving style.

[0154] Low pedaling frequency: means that the driver usually accelerates and brakes smoothly and has a gentle driving style.

[0155] Pedal slope indicates the rate at which the pedal changes when the driver operates the pedal. It describes how smoothly the driver's action is when accelerating or decelerating.

[0156] High slope (sudden acceleration / sudden braking): Indicates that the driver accelerates or brakes violently and has an impatient driving style.

[0157] Low slope: Indicates that the driver's movements when accelerating or decelerating are relatively smooth and his driving style is relatively gentle.

[0158] The number of steering turns refers to the frequency of the driver's steering wheel operations within a certain period of time. This usually reflects the driver's reaction to road conditions and the frequency of adjustments.

[0159] High steering frequency: indicates that the driver frequently adjusts the steering wheel while driving, which may be due to excessive control or complex road conditions, and the driving style tends to be impatient or overly cautious.

[0160] Low steering frequency: indicates that the driver makes fewer steering wheel adjustments while driving, and the driving style may be smoother or more stable.

[0161] Steering slope refers to the rate or speed at which the steering wheel turns when the driver is operating the steering wheel, and is usually related to the smoothness or aggressiveness of driving.

[0162] High steering slope: Indicates that the driver's steering movements are more violent, which may be a reaction to a sharp turn or emergency situation, and the driving style is more aggressive.

[0163] Low steering slope: Indicates that the driver's steering action is smooth, he is not impatient when turning, and his driving style is relatively smooth.

[0164] 2) Driving style assessment model

[0165] A simple weighted model is built to consider the relative importance of each indicator. Each factor (number of pedal strokes, pedal stroke slope, number of turns, and turn slope) can be converted into a style score and then weighted summed.

[0166] Establish a standardized scoring system for each indicator, usually ranging from 0 to 1. Suppose each indicator is scored as follows:

[0167] Pedal frequency score: The higher the frequency of pedaling, the higher the score.

[0168] Pedaling Slope Rating: The greater the slope, the higher the score.

[0169] Turn number score: The more turns you make, the higher the score.

[0170] Turning Slope Score: The greater the turning slope, the higher the score.

[0171] Assuming that each factor has a weight, indicating its influence on the driving style score, the driver's overall driving style score (S) can be expressed as:

[0172] S=w1·F pedal +w2·G pedal +w3·F steering +w4·G steering

[0173] Among them, F pedal and G pedal They are the pedaling times score and the pedaling slope score. steering and G steering They are the turn number score and the turn slope score. w1, w2, w3, and w4 are the weights of each factor. These weights can be adjusted according to the actual application scenario.

[0174] By adjusting the weight coefficients w1, w2, w3, and w4, you can adjust the degree of attention paid to acceleration, braking, and steering. If you pay attention to driving stability, you can increase the weights of the pedal slope and the steering slope. If you pay attention to driving impatience, you can increase the weights of the number of pedaling and the number of steering.

[0175] In summary, the present invention belongs to the technical field of real vehicle test data processing of special vehicle electromechanical composite transmission system and driving behavior evaluation based on data processing, and specifically relates to a driving style evaluation method based on real vehicle data. The method is oriented to the test data collected from the real vehicle of the electromechanical composite transmission system of the special vehicle, analyzes the internal correlation and interaction relationship of the data, processes the acceleration test data, and finds the rising trend or falling trend of the data through trend analysis, captures the rising number and falling number of the data, calculates the rising slope and the falling slope, and captures the left and right steering times through steering data analysis, calculates the left and right steering slopes, and then evaluates the driver's driving style through comprehensive evaluation of the acceleration and steering times and slopes. The method solves how to accurately evaluate the driver's driving style based on the test data collected from the real vehicle of the electromechanical composite transmission system of the special vehicle, and comprehensively evaluates the driver's driving behavior characteristics, such as acceleration methods, steering habits, etc. This method can extract key driving mode information from real vehicle data, and provide a scientific basis for performance optimization, driving behavior monitoring and safety improvement of special vehicles.

[0176] The method is divided into real vehicle test data conversion, acceleration number statistics and intensity calculation data processing method, turning number statistics and intensity calculation data processing method, and driving style comprehensive evaluation method. The beneficial effect of the present invention is to accurately evaluate the driver's driving style based on the real vehicle data of the electromechanical composite transmission system of the special vehicle. The method can extract key behavioral features from the test data of acceleration and turning, such as acceleration times, acceleration intensity, steering mode, etc., and quantify the driver's driving habits by analyzing the data trend and slope. This evaluation method not only improves the accuracy of driving style analysis, but also provides data support for performance optimization, driving behavior monitoring, driver training and safety improvement of special vehicles. Through in-depth analysis of real vehicle data, potential driving problems can be better identified and vehicle control strategies can be optimized, thereby improving the use efficiency and safety of special vehicles.

[0177] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A driving style evaluation method based on real vehicle data, characterized in that: The method comprises the following steps: Step 1: Real vehicle test data conversion; Step 2: Acceleration times statistics and strength calculation data processing; Step 3: Data processing of turning number statistics and intensity calculation; Step 4: Comprehensive assessment of driving style.

2. The driving style evaluation method based on real vehicle data according to claim 1, characterized in that: In step 1, the operating data of the mechatronic transmission system under different working conditions are collected from the actual vehicle test process; the format of the original data collected from the actual vehicle is usually diverse and not uniform, so format conversion is required to ensure that all data have a consistent format. The conversion process includes timestamp synchronization, multi-source data fusion, and data regeneration to obtain an accelerator pedal position curve and a steering wheel steering angle curve.

3. The driving style evaluation method based on real vehicle data according to claim 2, characterized in that: In the step 1, the operation data of the electromechanical hybrid transmission system under different working conditions are collected during the actual vehicle test; the data include: an accelerator pedal position signal and a steering wheel angle signal.

4. The driving style evaluation method based on real vehicle data according to claim 2, characterized in that: In the step 2, acceleration number statistics and strength calculation data processing are performed; Because the driver's acceleration behavior is completed by stepping on the accelerator pedal, the depth and speed of the driver's pedaling will be reflected in the collected acceleration curve, and there will be differences in fluctuation amplitude and speed. By analyzing the acceleration curve, the number of accelerations and the acceleration intensity are analyzed; for the accelerator pedal position curve obtained in step 1, a sliding window combined with a threshold analysis method is used to analyze the general trend of the accelerator pedal position curve and determine the upward and downward trends; According to the judgment result, the next trend under the big trend is searched downward based on the timestamp. The sliding window method is also used to find the position of the trend change, find the maximum or minimum value of the trend change position, record the position point of the maximum or minimum value, calculate the time difference between the initial point and the maximum or minimum point, and calculate the size of each acceleration slope based on the value of the maximum or minimum point.

5. The driving style evaluation method based on real vehicle data according to claim 4, characterized in that: In step 2, the process of determining the general trend of the data is as follows: First, define the size of the sliding window for detecting local trends. The default sliding window size is 15. Define the window size for calculating local trends. The default is 15. Define the size of the threshold for representing data fluctuations for trend analysis. The default value of the threshold is 8. Define the minimum threshold for the slope, which represents the slope threshold for the descending process. The default value is 0. Define the maximum threshold for the slope, which represents the slope threshold for the ascending process. The default value is 0. Initialize the number of recorded rising and falling processes, the slope of each rising and falling process, the data length, and the starting index of the data; Enter the loop process, the judgment condition is that the sliding window will not exceed the range of the data, and then extract a value with a length of the sliding window size from the data, and the index starts from the initial position to the end position of the data; Check the fluctuation in the current window, that is, whether the difference between the maximum and minimum values ​​exceeds the set threshold. If it exceeds the threshold, it means that there is significant fluctuation in the window, and further trend analysis is needed to determine the upward or downward trend. If the average value of the data difference in the window is greater than the set maximum slope threshold, it is considered to be an upward trend. The extreme value function is called to find the maximum value point in the data, that is, an upward trend. The upward trend count is increased by 1, indicating that an upward process is detected; the slope of the upward process is added to the array, and the index is updated to the next index of the maximum value. This is used as a new starting point to continue the subsequent analysis; If the average value of the data difference in the window is less than the set minimum slope threshold, it is considered to be a downward trend. The extreme value function is called to find the minimum point in the data, that is, the downward trend. The downward trend count is increased by 1, indicating that a downward process is detected. Add the slope of the descending process to the array, and update the next index of the minimum value, taking this as the new starting point to continue the subsequent analysis; If the fluctuation of the window is not large or the fluctuation of the window does not exceed the threshold, or it is not clear whether it is an upward or downward trend, the program increases the index by 1 and continues to analyze the next window.

6. The driving style evaluation method based on real vehicle data according to claim 5, characterized in that: In step 2, the method for finding the extreme value is: First, initialize the extreme value index found in the data trend judgment, and find the data window starting from the index position with the inspection window size as the size, which is used to compare with the current window when looking for trends; Loop search, starting from the initial index position and using the initial index position data window as the window size, until the end of the data is searched; If an upward trend is found in the data trend judgment, check whether the data in the current window is in an upward trend, and calculate the difference between adjacent data points. If the mean of the difference is greater than 0, it means that the data in the window is in an upward trend; Check the next window in a loop. If the average value of the difference between adjacent data in the window is greater than 0, it means that the trend is still rising. The window data is stored in the storage array. If the storage array stores more than 2 windows of window data, keep the most recent 2 windows. If the current window trend is downward and the difference between the maximum and minimum values ​​in the storage array is less than 12, ignore the current downward trend and continue to search for an upward trend; Find the maximum value of the current window and the previous window, record the maximum value index, and calculate the average rate of change from the starting index to the extreme value found, which is the rising slope; If a downward trend is found in the data trend judgment, check whether the data in the current window is in a downward trend, and calculate the difference between adjacent data points. If the mean of the difference is less than 0, it means that the data in the window is in a downward trend; Check the next window in a loop. If the average value of the difference between adjacent data in the window is less than 0, it means that the trend is still downward. The window data is stored in the storage array. If the storage array stores more than 2 windows of window data, keep the most recent 2 windows. If the current window trend is rising and the difference between the maximum and minimum values ​​in the storage array is less than 12, ignore the current rising trend and continue to search for the falling trend; Find the minimum value of the current window and the previous window, record the minimum value index, and calculate the average rate of change from the starting index to the extreme value found, which is the descending slope; If no trend change point is found in the loop, continue searching forward. If no trend end point is found in the entire data sequence, return the maximum or minimum value of the initial window, and return the corresponding extreme value and average slope.

7. The driving style evaluation method based on real vehicle data according to claim 4, characterized in that: In step 3, the number of turns is counted and the intensity is calculated. The driver's steering behavior is completed by turning the steering wheel, and the difference between left-turn and right-turn steering wheel reflected in the collected steering wheel angle is the difference between positive and negative; the depth and speed of the driver's steering wheel turning are reflected in the collected steering curve, which will have differences in fluctuation amplitude and speed. By analyzing the steering wheel steering angle curve, the number of left and right turns and the steering intensity are analyzed; Since the steering wheel steering angle curve contains the center steering curve content, it is necessary to remove the center steering data from the steering wheel steering angle curve data. After removal, the sliding window analysis method is used again to analyze and find the steering position point. Considering the driving behavior characteristics, the starting zero-point window position is found, and the timestamp of the position is recorded. Search downward along the timestamp to find a sliding window with non-zero data. After finding it, continue to search downward along the timestamp until the next all-zero window is found, and record the position point of the window; Calculate the time difference between the initial position and this position, find the extreme values ​​of the initial position and this position, and thus calculate the slope of the left turn and the right turn.

8. The driving style evaluation method based on real vehicle data according to claim 7, characterized in that: In step 3, the method for removing the center steering data is: First, drive through the center turn data analysis, and obtain the center turn position point data. First, initialize the variable values, including the number of left turns, the number of right turns, and the array used to store the steering slope; Loop through each location where the center turn occurs, use the sliding window to find the area with a value of 0 and clear the value to zero, use the sliding window to check whether the window is completely zero. If it is zero, skip and find the next non-zero area. After finding the non-zero area, continue to find the next zero window area, that is, find the end point of the non-zero area, clear the values ​​corresponding to the data points from the initial point to the end point, thereby eliminating the center turn data as a new array; In step 3, the method for calculating the number of turns and the intensity is: The new array is traversed using a sliding window method. The size of each sliding window is the default value. If the first element of the window is 0 and the subsequent elements are non-zero, it indicates the beginning of the turn. The starting position is recorded, and then the sliding window is continued to search downward in a loop until a window of all zeros is found, indicating the end of the turn. Calculate the number of turns and the turn slope. For each turn window, first check whether there is a negative turn or a positive turn; If the number of negative turning elements is greater than 5, check whether the minimum value of the window is less than the set left turning threshold. If so, calculate the slope of the left turn; if the number of positive turning elements is greater than 5, check whether the maximum value of the window is greater than the set right turning threshold. If so, calculate the slope of the right turn.

9. The driving style evaluation method based on real vehicle data according to claim 8, characterized in that: In the step 4, a comprehensive evaluation of driving style is performed; Based on the calculation results of steps 2 and 3, a weighted model is constructed to consider the relative importance of each indicator; each factor including the number of pedaling times, pedaling slope, number of steering times, and steering slope is converted into a style score, and then a weighted sum is performed.

10. The driving style evaluation method based on real vehicle data according to claim 9, characterized in that: In step 4, a standardized scoring system is established for each indicator, usually ranging from 0 to 1; each indicator is scored as follows: Pedal frequency score: The higher the frequency of pedaling, the higher the score; Pedaling slope score: the greater the slope, the higher the score; Turn number score: The more turns you make, the higher the score; Turning slope score: The greater the turning slope, the higher the score; Assume that each factor has a weight, indicating its influence on the driving style score; the driver's overall driving style score S is: S=w1·F pedal +w2·G pedal +w3·F steering +w4·G steering Among them, F pedal and G pedal They are the pedaling times score and the pedaling slope score; F steering and G steering are the turning number score and turning slope score respectively; w1, w2, w3, w4 are the weights of each factor; By adjusting the weight coefficients w1, w2, w3, and w4, you can adjust the degree of attention paid to acceleration, braking, and steering respectively; if you pay attention to the smoothness of driving, increase the weights of the pedal slope and the steering slope; if you pay attention to the impatience of driving, increase the weights of the number of pedaling times and the number of steering times.

Images