A heavy engineering vehicle driving performance evaluation method based on natural driving data

Abstract

The application discloses a heavy engineering vehicle driving performance evaluation method based on natural driving data, first, vehicle motion data under a natural driving environment is collected, and data preprocessing is carried out; then, corresponding statistical features of driving behaviors under different road types are extracted; secondly, driving performance evaluation is carried out based on driving behavior statistical indexes; finally, driving scores are calculated based on driving performance evaluation results, and finally, a driving performance evaluation system is obtained. The application can serve driving performance examination and safety education management of drivers, enhance road traffic safety consciousness of the drivers, and reduce road traffic accidents.

Description

Technical Field

[0001] This invention relates to a method for evaluating the driving performance of heavy-duty engineering vehicles based on natural driving data, belonging to the field of driving performance evaluation technology. Background Technology

[0002] With the acceleration of urbanization in my country, heavy-duty engineering transport vehicles, including concrete mixer trucks, sand trucks, and dump trucks, have become important carriers for transporting materials between construction sites and resource depots. Heavy-duty engineering vehicles have a high center of gravity and large load capacity, making driver behavior difficult to monitor and posing a high risk. Driver performance is a crucial indicator of driving safety, and evaluating the driving performance of heavy-duty engineering vehicle drivers helps fleet management, creating safer and more economical driving conditions for the fleet.

[0003] While significant progress has been made in research on driving behavior, several shortcomings remain: Domestic research on driving performance evaluation is still in its early stages. Existing methods for evaluating the driving performance of engineering vehicle drivers are mostly simple weighted scoring systems. However, due to differences in national conditions, vehicle types, and driver characteristics between China and other countries, comprehensive driving performance evaluation models used abroad cannot be directly applied to heavy engineering vehicles in China. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method for evaluating the driving performance of heavy engineering vehicles based on natural driving data, and to provide a method for evaluating the driving performance of heavy engineering vehicle drivers in combination with the natural driving environment.

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0006] A method for evaluating the driving performance of heavy-duty engineering vehicles based on natural driving data includes the following steps:

[0007] Step 1: Collect vehicle motion data of heavy engineering vehicles in a natural driving environment, and preprocess the vehicle motion data to obtain driving behavior data;

[0008] Step 2: Divide the driving behavior data obtained in Step 1 into driving behavior data under urban road type and driving behavior data under rural road type. Extract statistical features from the driving behavior data under different road types to obtain the driving performance evaluation index for heavy engineering vehicles.

[0009] Step 3: Based on the evaluation indicators obtained in Step 2, perform correlation analysis and principal component analysis on them in sequence, cluster the obtained principal components, and establish a driving performance evaluation model based on the clustering results.

[0010] Step 4: Calculate the driving score based on the driving performance evaluation model established in Step 3, and construct a driving performance evaluation system for heavy engineering vehicles.

[0011] In a preferred embodiment of the present invention, in step 1, the vehicle motion data includes the following variables: number of satellites, UTC time, latitude and longitude, speed, heading angle, altitude, vertical speed, lateral acceleration, and longitudinal acceleration; the collected vehicle motion data are placed in a matrix in chronological order, with one column of the matrix corresponding to one variable, and one row of the matrix containing all the variables obtained from one collection of vehicle motion data; the vehicle motion data collection frequency is 10Hz;

[0012] Vehicle motion data is preprocessed to obtain driving behavior data, as follows:

[0013] 1.1 Invalid values ​​are processed for vehicle motion data. Invalid column variables and invalid sample rows are deleted. Column variables with all empty data are defined as invalid column variables. Sample rows with all empty data, sample rows with fewer than 5 satellites, and sample rows with latitude, longitude, speed, and altitude all being 0 are invalid sample rows.

[0014] 1.2 Based on 1.1, delete the sample rows where the velocity, heading angle, vertical velocity, lateral acceleration, and longitudinal acceleration are all 0. Use the moving average method to fill in the missing data, taking the average of the 10 values ​​before and after the column containing the missing data.

[0015] 1.3 Convert the vehicle motion data collected at a frequency of 10Hz to 1Hz, that is, use the average of the 10 data points collected in the current second as the statistical value for the current second.

[0016] As a preferred embodiment of the present invention, in step 2, the scope of rural roads is delineated on the topographic map, and driving behavior data with latitude and longitude within the delineated scope are marked as driving behavior data under the rural road type, while driving behavior data with latitude and longitude outside the delineated scope are marked as driving behavior data under the urban road type.

[0017] For driving behavior data under each road type, the driving segment is divided into several driving segments with a step size of 1 second and a sliding time window length of 60 seconds. When a driving segment contains driving behavior data under both urban road type and rural road type, the type corresponding to the part with longer driving time is taken as the type of the driving segment.

[0018] Statistical features were extracted from longitudinal and lateral driving behavior data under different road types. Longitudinal driving behavior includes speed and longitudinal acceleration, while lateral driving behavior includes lateral acceleration. This yielded performance evaluation indicators for heavy engineering vehicle drivers. The evaluation indicators include: mean absolute deviation of speed, coefficient of variation of speed, quartile divergence coefficient of speed, maximum value of longitudinal acceleration, minimum value of longitudinal acceleration, mean value of longitudinal acceleration, standard deviation of longitudinal acceleration, mean absolute deviation of longitudinal acceleration, and mean absolute deviation of lateral acceleration.

[0019] As a preferred embodiment of the present invention, the specific process of step 3 is as follows:

[0020] 3.1 Perform Pearson correlation analysis on the heavy engineering vehicle driving performance evaluation indicators obtained in step 2, and conduct KMO test and Bartlett test. Use principal component analysis to reduce the dimensionality of the evaluation indicators.

[0021] 3.2 The principal components obtained by principal component analysis are clustered into three categories using the K-means clustering method. The driving performance of each category is determined as aggressive, normal, or cautious driving performance based on the characteristic parameters of the centroids of each category.

[0022] As a preferred embodiment of the present invention, in step 3.2, the feature parameters of the three category center points are compared, and the category with the highest feature parameter is marked as aggressive driving performance, the category with the lowest feature parameter is marked as cautious driving performance, and the category with the feature parameter at an intermediate level is marked as normal driving performance.

[0023] In a preferred embodiment of the present invention, step 4 uses a 123 digitization to represent three types of driving performance: an aggressive driving performance score of 1, a normal driving performance score of 2, and a cautious driving performance score of 3. The driving gain rate is then calculated.

[0024]

[0025] Among them, S t,i S represents the driving performance score of driver i in driving segment t. t-1,i r represents the driving performance score of driver i in driving segment t-1. i,t This represents the driving gain rate of driver i during driving segment t, and signifies the fluctuation in the driver's driving performance during continuous driving.

[0026] Driving volatility score D is calculated based on driving gain rate. volatility That is, the sample standard deviation of the driver's driving gain rate over a certain period of time:

[0027]

[0028] in, represents the average driving gain rate, and n represents the total number of driving segments evaluated by the driver during a trip.

[0029] A computer device includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the computer program to implement the steps of the heavy-duty engineering vehicle driving performance evaluation method based on natural driving data as described above.

[0030] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the heavy-duty engineering vehicle driving performance evaluation method based on natural driving data as described above.

[0031] Compared with the prior art, the present invention, employing the above technical solution, has the following technical effects:

[0032] 1. This invention is the first to incorporate short-term driving performance fluctuations into driving performance evaluation.

[0033] 2. This invention fully considers the different driving performances of drivers under different road types and proposes a driving performance evaluation method that combines road types.

[0034] 3. Existing research on driving performance scores focuses on judging and evaluating driving performance indicator thresholds, and has not yet conducted a comprehensive assessment of driving behavior. This invention provides a driving performance evaluation method that comprehensively considers driving performance and driving performance fluctuations. Attached Figure Description

[0035] Figure 1 This is a flowchart of the heavy-duty engineering vehicle driving performance evaluation method based on natural driving data according to the present invention;

[0036] Figure 2 The graph shows the results of the correlation analysis of variables, where (a) is the correlation analysis of urban areas and (b) is the correlation analysis of rural areas.

[0037] Figure 3 The following are clustering results: (a) shows the clustering results of urban roads, (b) shows the scatter distribution of urban road clusters, (c) shows the clustering results of rural roads, and (d) shows the scatter distribution of rural road clusters.

[0038] Figure 4 It is a line graph showing the change in driving gain rate over time for a driver during a certain trip;

[0039] Figure 5 It is a histogram of driving scores for different drivers on each trip;

[0040] Figure 6 It is a histogram of driving scores for different drivers on each test day;

[0041] Figure 7 It is a distribution chart of driving scores and average driving performance for different drivers. Detailed Implementation

[0042] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0043] This invention proposes a method for evaluating the driving performance of heavy-duty engineering vehicles based on natural driving data, the process of which is as follows: Figure 1 As shown, the specific steps are as follows:

[0044] (1) Collect vehicle motion data under natural driving conditions and perform data preprocessing to obtain effective data;

[0045] This invention utilizes a high-precision data acquisition device—the VBOX-IISX10 GPS system, a high-performance satellite receiver developed by Racelogic—to collect high-frequency natural driving data from a heavy-duty engineering vehicle under natural driving conditions. The VBOX can record information such as the number of satellites, operating time, geographical location, and vehicle kinematic data, specifically including the number of satellites, UTC time, latitude, longitude, speed, heading angle, altitude, vertical velocity, lateral acceleration, and longitudinal acceleration, with a sampling frequency of 10 Hz. In addition, it is equipped with a 2-megapixel dashcam, featuring a 140° field of view and 1080p video resolution. The dashcam is equipped with high-definition night vision front and rear dual cameras. The in-vehicle camera records the driver's actions during driving, such as mobile phone use and body language; the external camera primarily records the conditions in front of and around the test vehicle during its operation, which is later compared with GPS data to eliminate abnormal driving data. Preliminary offline data processing is performed using VBOX Tools. VBOX may inadvertently collect erroneous and biased data during the data recording process, including invalid field columns and invalid sample rows. In particular, there are invalid record rows at the beginning of each file due to restarting a crashed device, which cannot represent the vehicle's motion status. Therefore, data preprocessing is required.

[0046] In the invalid value handling section, invalid column variables are first removed. The original dataset exported from VBOX contains invalid column variables with empty data records, which need to be removed. Secondly, invalid sample rows are deleted. Data with fewer than 5 satellites or missing data due to obstructions such as trees or buildings cannot characterize driving behavior and has no substantial significance for this invention; these invalid values ​​can be directly deleted. When latitude, longitude, speed, and altitude are 0, it indicates that the equipment is in a downtime state and can be directly deleted.

[0047] Handling missing data requires analysis based on the type of missing value, and specific operations include data deletion and data imputation. When speed, heading angle, vertical speed, lateral acceleration, and longitudinal acceleration are all 0 during driving, it indicates that the vehicle is stationary, possibly during a rest period between trips, and these data need to be deleted. The example data in this invention is time series data. Due to the high similarity of data before and after a 0.1s time interval, a moving average method is used for interpolation, taking the average of the 10 values ​​before and after the valid data to fill in the missing values.

[0048] Given that time series data, due to the correlation and randomness of its preceding and following data, cannot truly predict the magnitude of a value at a particular moment, previous smoothing methods would reduce outliers. However, removing outliers would delete normal risky driving samples from the data. Therefore, we chose to down-convert the data, transforming the 10Hz (0.1s step) data into 1Hz, and taking the average of 10 data points within 1 second as the statistic for the current second, in preparation for the extraction of subsequent driving segments.

[0049] (2) Based on the driving behavior data obtained in step (1), divide it into urban roads and rural roads, extract statistical features from the lateral driving behavior and longitudinal driving behavior data under different road types, and obtain the evaluation index system.

[0050] Driving performance evaluation distinguishes between urban and rural road scenarios. First, on the terrain map, the outskirts of the city and areas surrounding mountains are designated as rural roads. Roads within this designated area are marked as rural roads, while those outside are marked as urban roads. When a driving segment traverses both urban and rural roads, the segment with the longer travel time is labeled as the rural road segment.

[0051] This invention calculates driving performance evaluation indicators from two dimensions: longitudinal driving behavior and lateral driving behavior. These indicators include speed, longitudinal acceleration, and lateral acceleration. Driving segments are acquired using a 1-second step and a 60-second sliding time window to calculate the evaluation indicators. The evaluation indicators include common statistical indicators and driving volatility indicators. Common statistical indicators include the mean and standard deviation; the mean represents the average level of driving behavior, and the standard deviation represents the dispersion of driving behavior. Specific driving volatility indicators include the coefficient of variation, mean absolute deviation, and quartile divergence coefficient, with specific formulas as follows. The evaluation indicators and their definitions are shown in Table 1. When selecting the maximum and minimum values ​​of acceleration, to reflect the degree of acceleration variation, the absolute values ​​of acceleration are considered; that is, the maximum and minimum absolute values ​​of acceleration are taken. The other volatility indicators are calculated under normal conditions.

[0052] Table 1 Evaluation Indicator Variables and Definitions

[0053] Variable name Variable definition v_mad Mean absolute deviation of speed v_cv Coefficient of variation of velocity v_qq Interquartile divergence coefficient of velocity lacc_max Maximum longitudinal acceleration lacc_min Minimum longitudinal acceleration lacc_mean Average longitudinal acceleration lacc_std Standard deviation of longitudinal acceleration lacc_mad Mean absolute deviation of longitudinal acceleration latacc_mad Mean absolute deviation of lateral acceleration

[0054] average value It measures the average level of the data.

[0055]

[0056] Where N is the number of data observations, x i These are data observations.

[0057] Standard deviation S dev It measures the level of dispersion between the observed data and the average data.

[0058]

[0059] Coefficient of variation C v It measures the difference between the standard deviation and the absolute value of the data.

[0060]

[0061] Mean absolute deviation D mean It measures the average distance between data observations and the data mean.

[0062]

[0063] quartile divergence coefficient Q cv It measures the degree of dispersion of the data.

[0064]

[0065] Q3 is the 75th percentile, and Q1 is the 25th percentile.

[0066] The descriptive statistics of the final extracted driving segment evaluation indicators are shown in Table 2, based on 1200 urban samples and 3995 rural samples. The results show that the mean speed on urban roads is greater than that on rural roads, but the standard deviation is smaller, indicating that the average speed of construction vehicles on urban roads is higher than on rural roads, but the dispersion is lower. This may be because the speed limit on urban roads is higher than on rural roads, leading to higher vehicle speeds in many cases. However, traffic interference is greater on urban roads, resulting in vehicles exhibiting stable low-speed following behavior due to the influence of vehicles ahead, or stable low-speed driving between adjacent intersections, thus resulting in a lower standard deviation. Regarding longitudinal acceleration, heavy construction vehicle drivers experience greater longitudinal acceleration on urban roads, consistent with the above conclusions. The driving environment and road conditions (such as traffic signals, access points, and commercial facilities) vary more in cities, and interactions with other road users (pedestrians and cyclists) are more complex. Therefore, driving volatility is higher in cities, and with higher speed limits, vehicles accelerate to higher speed levels, indicating that heavy construction vehicles need to pay more attention to longitudinal safety. Regarding lateral acceleration behavior, the lateral acceleration volatility index values ​​on urban roads are lower than those on rural roads, except for the minimum value. This may be because rural roads are more winding than urban roads, making it harder for drivers to anticipate the road conditions and leading to more frequent lane changes and greater volatility. Therefore, evaluating driving performance separately based on road type is reasonable.

[0067] Table 2. Descriptive Statistics of Evaluation Indicators for Urban and Rural Driving Segments

[0068]

[0069] (3) Based on the evaluation indicators obtained in step (2), perform correlation analysis and principal component analysis on them in sequence, cluster the obtained principal components, and establish a driving performance evaluation model.

[0070] In this invention, multiple clustering indicators are used. Directly incorporating them into K-means clustering would create a high-dimensional space. Calculating Euclidean distance in such a high-dimensional space would lead to the curse of dimensionality, rendering the calculated distance meaningless. Principal component analysis (PCA) can reduce the dimensionality of the data, yielding several uncorrelated linear variables without losing the original data information. First, correlation analysis is performed to gain a preliminary understanding of the correlations between the evaluation indicators from a global perspective. Figure 2(a) and (b) show the correlation analysis results of the evaluation indicators for urban and rural driving segments, respectively. In urban roads, the correlation coefficient between the quartile divergence coefficient and the coefficient of variation of speed is as high as 0.92, and the correlation coefficient between the mean absolute deviation and standard deviation of longitudinal acceleration is also as high as 0.97, indicating a strong correlation between the variables. Therefore, principal component analysis can be preliminarily determined as a suitable approach. Next, applicability statistical tests were performed on the evaluation indicators, and principal component analysis was conducted on the characteristics of the evaluation indicators, including the Kaiser-Meyer-Olkin Measure of Adequacy (KMO) and Bartlett's test. The calculated KMO coefficient for urban roads was 0.727, and Bartlett's test of sphericity was significant (p<0.000). The KMO coefficient for rural roads was 0.736, and Bartlett's test of sphericity was significant (p<0.000), indicating that principal component analysis is suitable for the characteristic variables on both urban and rural roads.

[0071] As shown in Table 3, on urban roads, Principal Component 1 shows a high correlation with indicators related to speed and longitudinal acceleration, explaining 51.7% of the variance; while Principal Component 2 shows a high correlation with the maximum value of longitudinal acceleration and the mean absolute deviation of lateral acceleration, explaining 19.1% of the variance. Together, the two principal components explain 70.8% of the variance. On rural roads, Principal Component 1 shows a high correlation with indicators related to speed and longitudinal acceleration, explaining 54.4% of the variance; Principal Component 2 shows a high correlation with the mean absolute deviation of lateral acceleration, explaining 18.2% of the variance. Together, the two principal components explain 72.6% of the variance.

[0072] Table 3 explains the variance and loading matrices.

[0073]

[0074] After clustering, the software automatically generates clusters numbered 0, 1, and 2, but it does not automatically provide the driving performance type represented by each cluster. Analysis based on the central feature value of each cluster is required. Figure 3 The clustering results for urban and rural roads are presented. Figure 3 (a) shows the clustering results on urban roads, with a silhouette coefficient of 0.40. Cluster number 2 has the fewest clusters. Figure 3 (b) shows a scatter plot with the two principal components as the x and y axes. The purple scatter points are more dispersed, and the value of principal component 2 is the largest. Table 4 shows the center point parameter indicators of each cluster of urban roads. It can be seen that the feature parameters of category 2 are the highest, and this type is marked as aggressive driving performance; the feature parameters of category 1 are at the middle level, and this type is marked as normal driving performance; the feature parameters of category 0 are all at the lowest value, and this type is marked as cautious driving performance. Figure 3(c) shows the clustering results on rural roads, with a silhouette coefficient of 0.40. Cluster number 2 has the fewest clusters. Figure 3 A scatter plot was drawn with the two principal components (d) as the x and y axes. The orange scatter points are more dispersed, and the value of principal component 2 is the largest. As shown in Table 4, the feature parameters of category 2 are the highest, and this type is marked as aggressive driving performance; the feature parameters of category 1 are at an intermediate level, and this type is marked as normal driving performance; the feature parameters of category 0 are all at their lowest values, and this type is marked as cautious driving performance.

[0075] Table 4. Cluster centroid indexes

[0076]

[0077]

[0078] (4) Calculate the driving score based on the driving performance evaluation results in step (3) and construct a heavy engineering vehicle driving performance evaluation system.

[0079] To quantify the risk level of driving performance, this invention uses a 123 digit system to represent three categories of driving performance: 1 for aggressive driving, 2 for normal driving, and 3 for cautious driving. Higher scores indicate higher driving safety. Unlike driving performance, driving score measures the volatility of a driver's performance over a period of time. Higher scores indicate greater volatility in driving performance. In this case, to lower the score and optimize the driving performance evaluation, the driver will consciously adopt more cautious and stable vehicle operation behaviors, and vice versa. Table 5 shows the stratified evaluation indicators for driving performance.

[0080] Table 5 Driving Performance Stratification Evaluation Indicators

[0081] Serial Number Evaluation level Evaluation indicators 1 Minute-level driving performance evaluation Driving performance, driving gain rate 2 Trip-level driving performance evaluation Driving score 3 Daily driving performance evaluation Driving score

[0082] First, calculate the driving gain ratio using the following formula:

[0083]

[0084] Among them, S t,i S is the driving performance score of driver i during time period t. t-1,i This is the driving performance score of driver i during time period t-1. i,tThis is the driver gain rate for each assessment period compared to the previous assessment period, representing the fluctuation of the driver's driving performance during continuous driving. If the value is positive, it indicates that the driver is approaching a safe state, such as from driving performance 2 to driving performance 3, from driving performance 1 to driving performance 3, and from driving performance 1 to driving performance 2; if the value is negative, it indicates that the driver is approaching a dangerous state, such as from driving performance 3 to driving performance 2, from driving performance 3 to driving performance 1, and from driving performance 2 to driving performance 1.

[0085] A driver's journey was randomly selected from a natural driving test. Figure 4 The driving gain rate distribution over time shows that when a driver exhibits aggressive driving behavior at 14:03, 14:07, and 14:22, but the driving gain rate is positive, it indicates that their driving behavior tends towards safety. When a driver exhibits aggressive driving behavior at 14:18 and the driving gain rate is negative, it indicates that they are in a more dangerous driving state compared to the previous minute, which requires attention.

[0086] Driving performance varies between each driver's trips, making it crucial to evaluate overall driving performance across all trips. This invention uses driving scores to compare driving performance levels for each trip. The following formula calculates the driving volatility score based on the driving gain rate, which is the sample standard deviation of the driver's driving gain rate over a given trip.

[0087]

[0088] Where, r i,t This represents the driver's gain rate in each driving segment. represents the average driving gain rate, and n represents the total number of driving segments evaluated by the driver during a trip.

[0089] Figure 5 The histograms show the driving scores of all trips for 11 test drivers. In the natural driving test, the number of trips for each driver varied depending on the task arrangement, but it was still possible to generally judge the driver's driving performance level and the differences in driving between trips. As shown in the graphs, the driving scores of the same driver varied across different trips. A higher driving score indicated a greater degree of driving fluctuation and poorer driving stability, requiring the driver to drive more cautiously.

[0090] Drivers exhibit individual differences due to variations in age, driving experience, and personality. A comprehensive evaluation of a driver's driving performance can assess their risk level at the driver level, and the evaluation results are helpful for performance management and reward / penalty management. Similar to the calculation process for trip-level driving performance scores, this method calculates the driving gain rate based on the driver's historical driving performance, and then uses the driving volatility formula to derive the driving performance score.

[0091] Figure 6 The histograms of driving scores for 11 test drivers on each test day are presented. The natural driving experiment conducted in this invention collected driving behavior data from each driver over three days. While driving performance varied from day to day, the overall driving performance level and differences between trips could still be assessed. The histograms show that the driving scores of the same driver differed across test days; higher scores indicated greater driving volatility. For example, Driver No. 5 scored below 0.5 on the first and second days but above 0.5 on the third day, indicating greater volatility in driving performance on the third day and suggesting an increased risk of driving. For Driver No. 7, the differences in driving scores over the three days were relatively small, but all scores were above 0.5, indicating significant driving volatility and a higher overall risk level, requiring increased intervention and improved road safety awareness.

[0092] Figure 7 The relationship between driving performance and driving performance score is illustrated. The horizontal axis represents the average driving performance label score, reflecting the driver's average driving performance level; scores closer to 3 indicate more cautious driving behavior, while scores closer to 1 indicate more aggressive driving behavior. The vertical axis represents the driver's overall performance score, reflecting the overall driving volatility level. The results show that the lower the average driving performance score for heavy engineering vehicle drivers, the higher their driving score, indicating that when drivers tend to drive aggressively, their vehicle operation will exhibit greater volatility. In the example of this invention, the average driving performance of the 11 tested drivers ranged from 2.1 to 2.4, indicating that their driving behavior was generally between normal and cautious, and their overall driving safety was relatively high. Notably, driver number 7 exhibited higher driving volatility and lower driving performance than any other driver, indicating a higher risk of driving. The research shows that the driving performance score proposed in this invention can effectively measure the time-varying characteristics and overall level of a driver's driving performance.

[0093] Based on the same inventive concept, this application provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the aforementioned method for evaluating the driving performance of heavy engineering vehicles based on natural driving data.

[0094] Based on the same inventive concept, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the aforementioned method for evaluating the driving performance of heavy engineering vehicles based on natural driving data.

[0095] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0096] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0097] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0098] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0099] The above embodiments are merely illustrative of the technical concept of the present invention and should not be construed as limiting the scope of protection of the present invention. Any modifications made to the technical solutions based on the technical concept proposed in this invention shall fall within the scope of protection of this invention.

Claims

1. A method for evaluating the driving performance of heavy-duty engineering vehicles based on natural driving data, characterized in that, Includes the following steps: Step 1: Collect vehicle motion data of heavy engineering vehicles in a natural driving environment, and preprocess the vehicle motion data to obtain driving behavior data; Step 2: Divide the driving behavior data obtained in Step 1 into driving behavior data under urban road type and driving behavior data under rural road type. Extract statistical features from the driving behavior data under different road types to obtain the driving performance evaluation index for heavy engineering vehicles. Step 3: Based on the evaluation indicators obtained in Step 2, perform correlation analysis and principal component analysis on them sequentially. Cluster the obtained principal components and establish a driving performance evaluation model based on the clustering results. The specific process is as follows: Step 3.1: Perform Pearson correlation analysis on the heavy engineering vehicle driving performance evaluation indicators obtained in Step 2, and conduct KMO test and Bartlett test. Use principal component analysis to reduce the dimensionality of the evaluation indicators. Step 3.2: The principal components obtained by principal component analysis are clustered into three categories using K-means clustering. The feature parameters of the centroids of the three categories are compared. The category with the highest feature parameter is labeled as aggressive driving performance, the category with the lowest feature parameter is labeled as cautious driving performance, and the category with the feature parameter in the middle level is labeled as normal driving performance. Step 4: Calculate the driving score based on the driving performance evaluation model established in Step 3, and construct a heavy engineering vehicle driving performance evaluation system; In step 4, the three types of driving performance are represented by a 123 digit system. When the driving performance is aggressive, the driving performance score is 1; when the driving performance is normal, the driving performance score is 2; and when the driving performance is cautious, the driving performance score is 3. The driving gain rate is then calculated. ， in, Indicates driver In driving footage Driving performance rating in the system Indicates driver In driving footage Driving performance rating in the system Indicates driver In driving footage The driving gain rate represents the fluctuation of a driver's driving performance during continuous driving. Driving volatility score calculated based on driving gain rate. That is, the sample standard deviation of the driver's driving gain rate over a certain period of time: ， in, This represents the average value of the driving gain rate. This indicates the total number of assessed driving segments for a driver during a trip.

2. The method for evaluating the driving performance of heavy-duty engineering vehicles based on natural driving data according to claim 1, characterized in that, In step 1, the vehicle motion data includes the following variables: number of satellites, UTC time, latitude and longitude, speed, heading angle, altitude, vertical speed, lateral acceleration, and longitudinal acceleration. The collected vehicle motion data are placed in a matrix in chronological order, with one column of the matrix corresponding to one variable, and one row of the matrix containing all the variables obtained from one collection of vehicle motion data. The vehicle motion data collection frequency is 10Hz. Vehicle motion data is preprocessed to obtain driving behavior data, as follows: 1.

1. Invalid values ​​are processed for vehicle motion data. Invalid column variables and invalid sample rows are deleted. Column variables with all empty data are defined as invalid column variables. Sample rows with all empty data, sample rows with fewer than 5 satellites, and sample rows with latitude, longitude, speed, and altitude all being 0 are invalid sample rows. 1.2, Based on 1.1, delete the sample rows where the velocity, heading angle, vertical velocity, lateral acceleration, and longitudinal acceleration are all 0. Use the moving average method to fill in the missing data, taking the average of the 10 values ​​before and after the column containing the missing data. 1.3 Convert the vehicle motion data collected at a frequency of 10Hz to 1Hz, that is, use the average of the 10 data points collected in the current second as the statistical value for the current second.

3. The method for evaluating the driving performance of heavy-duty engineering vehicles based on natural driving data according to claim 1, characterized in that, In step 2, the scope of rural roads is delineated on the topographic map, and driving behavior data with latitude and longitude within the delineated scope are marked as driving behavior data under the rural road type, while driving behavior data with latitude and longitude outside the delineated scope are marked as driving behavior data under the urban road type. For driving behavior data under each road type, the driving segment is divided into several driving segments with a step size of 1 second and a sliding time window length of 60 seconds. When a driving segment contains driving behavior data under both urban road type and rural road type, the type corresponding to the part with longer driving time is taken as the type of the driving segment. Statistical features were extracted from longitudinal and lateral driving behavior data under different road types. Longitudinal driving behavior includes speed and longitudinal acceleration, while lateral driving behavior includes lateral acceleration. This yielded performance evaluation indicators for heavy engineering vehicle drivers. The evaluation indicators include: mean absolute deviation of speed, coefficient of variation of speed, quartile divergence coefficient of speed, maximum value of longitudinal acceleration, minimum value of longitudinal acceleration, mean value of longitudinal acceleration, standard deviation of longitudinal acceleration, mean absolute deviation of longitudinal acceleration, and mean absolute deviation of lateral acceleration.

4. A computer device comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the heavy-duty engineering vehicle driving performance evaluation method based on natural driving data as described in any one of claims 1 to 3.

5. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the heavy-duty engineering vehicle driving performance evaluation method based on natural driving data as described in any one of claims 1 to 3.

Images