Bus simulated driving data analysis system and method

Abstract

The invention provides a bus simulated driving data analysis system and method, and relates to the field of intelligent data analysis. According to the method, the time queue data of the steering angle of a steering wheel, the time queue data of the force of an accelerator pedal, the time queue data of the force of a brake pedal and the time queue data of gears in the simulated driving process of a driver are subjected to comprehensive time sequence analysis to judge whether the driver drives stably or not; and based on this, a driving suggestion conforming to the driving behavior of the driver is given. In this way, personalized feedback and suggestions can be provided for the driver.

Description

Technical Field

[0001] The present application relates to the field of intelligent data analysis, and more specifically, to a bus simulation driving data analysis system and method. Background Art

[0002] As the main force of urban public transportation, buses carry a large number of passengers, and their driving safety is of vital importance. Through simulated driving, drivers can face various complex road conditions and emergencies in a virtual environment, such as emergency braking, pedestrians crossing, bad weather, etc., and practice repeatedly to improve their coping ability without causing safety risks on real roads. This helps new drivers quickly familiarize themselves with the business, and also allows experienced drivers to continue to strengthen their skills, providing a more solid safety guarantee for actual operations.

[0003] Analyzing bus simulation driving data can reveal bad habits of drivers during operation, such as frequent sudden braking, sudden acceleration, oversteering, etc. Through targeted correction, it can not only reduce vehicle wear and tear, reduce energy consumption, but also improve passenger comfort, avoid passengers in the car from falling and getting injured due to frequent bumps, and improve the quality of bus services. Many traditional bus simulation driving data analysis systems are based on rules or simple statistical models, and each driver has his or her own unique driving style and habits. Such rule-based or simple statistical models often assume that all drivers follow the same pattern, ignoring individual differences, which leads to the inability to provide personalized feedback or suggestions.

[0004] Therefore, an optimized bus simulation driving data analysis solution is needed. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present application provides a bus simulation driving data analysis system and method.

[0006] A bus simulation driving data analysis system, comprising:

[0007] The bus simulation driving data acquisition module is used to obtain the time queue data of the driver's steering wheel steering angle, the time queue data of the accelerator pedal force, the time queue data of the brake pedal force, and the time queue data of the gear position during the simulation driving process;

[0008] A bus simulation driving data timing coding module is used to perform timing coding and sorting on the time queue data of the steering wheel steering angle, the time queue data of the accelerator pedal force, the time queue data of the brake pedal force, and the time queue data of the gear position to obtain a bus driving monitoring parameter timing matrix;

[0009] A bus driving stability assessment feature generation module is used to extract multi-scale time series features from the bus driving monitoring parameter time series matrix to obtain a bus driving stability assessment feature vector;

[0010] The driving suggestion generating module is used for analyzing the bus driving stability evaluation feature vector and generating driving suggestions for the driver.

[0011] A bus simulation driving data analysis method, comprising:

[0012] Obtain the time queue data of the driver's steering wheel angle, the time queue data of the accelerator pedal force, the time queue data of the brake pedal force, and the time queue data of the gear position during the simulated driving process;

[0013] Performing time series coding and sorting on the time series data of the steering wheel steering angle, the time series data of the accelerator pedal force, the time series data of the brake pedal force, and the time series data of the gear position to obtain a bus driving monitoring parameter time series matrix;

[0014] Performing multi-scale time series feature extraction on the bus driving monitoring parameter time series matrix to obtain a bus driving stability evaluation feature vector;

[0015] The bus driving stability evaluation feature vector is analyzed to generate driving suggestions for the driver.

[0016] This application has significant technical effects due to the adoption of the above technical solutions:

[0017] The bus simulation driving data analysis system and method provided by the present application judges whether the driver is driving smoothly by performing comprehensive time series analysis on the time queue data of the steering wheel steering angle, the time queue data of the accelerator pedal force, the time queue data of the brake pedal force, and the time queue data of the gear position during the driver's simulated driving process, and gives driving suggestions that are consistent with the driver's driving behavior based on the time series data. In this way, personalized feedback and suggestions can be provided to the driver. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0019] Figure 1 4 is a system block diagram of a bus simulation driving data analysis system according to an embodiment of the present application.

[0020] Figure 2 It is a block diagram of a bus simulation driving data timing encoding module in a bus simulation driving data analysis system according to an embodiment of the present application.

[0021] Figure 3 It is a block diagram of a bus driving stability assessment feature generation module in a bus simulation driving data analysis system according to an embodiment of the present application.

[0022] Figure 4 This is a block diagram of a driving suggestion generation module in a bus simulation driving data analysis system according to an embodiment of the present application.

[0023] Figure 5 Flow chart of a bus simulation driving data analysis method according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0025] Based on the technical problems raised in the above background technology, the present application provides a bus simulation driving data analysis system. Figure 1 FIG. 1 is a system block diagram of a bus simulation driving data analysis system according to an embodiment of the present application. Figure 1 As shown, in the bus simulation driving data analysis system 100, it includes: a bus simulation driving data acquisition module 110, which is used to obtain the time queue data of the steering wheel steering angle, the time queue data of the accelerator pedal force, the time queue data of the brake pedal force, and the time queue data of the gear position of the driver during the simulation driving process; a bus simulation driving data time series encoding module 120, which is used to time series encode and organize the time queue data of the steering wheel steering angle, the time queue data of the accelerator pedal force, the time queue data of the brake pedal force, and the time queue data of the gear position to obtain a bus driving monitoring parameter time series matrix; a bus driving stability evaluation feature generation module 130, which is used to perform multi-scale time series feature extraction on the bus driving monitoring parameter time series matrix to obtain a bus driving stability evaluation feature vector; a driving suggestion generation module 140, which is used to analyze the bus driving stability evaluation feature vector and generate driving suggestions for the driver.

[0026] In the embodiment of the present application, the bus simulation driving data acquisition module 110 is used to obtain the time queue data of the steering wheel steering angle, the time queue data of the accelerator pedal force, the time queue data of the brake pedal force, and the time queue data of the gear position during the driver's simulation driving process. It should be understood that the steering wheel steering angle is an important indicator reflecting the driver's adjustment of the vehicle's controllability and driving trajectory. Different steering angles will affect the vehicle's driving direction and stability. If the driver frequently and significantly turns the steering wheel, it may mean that his judgment of the road conditions is inaccurate or the operation is not smooth enough, which may easily cause the vehicle to drive unsteadily. Therefore, analyzing the steering wheel steering angle can help the model understand the driver's ability and habits in controlling the vehicle's driving direction, and then give suggestions on how to turn the steering wheel more reasonably and smoothly in the driving advice, such as appropriately adjusting the steering angle in advance, reducing unnecessary large steering, etc., to improve the driving stability and safety. The accelerator pedal force is directly related to the acceleration performance of the vehicle. The strength and frequency of the driver's stepping on the accelerator pedal will affect the speed change of the vehicle. Frequent or excessive stepping on the accelerator pedal will cause the vehicle to accelerate rapidly, which not only increases energy consumption, but also may make passengers feel uncomfortable and even affect driving safety. By analyzing the accelerator pedal force, optimization suggestions can be given according to the driver's operating habits, such as how to control the vehicle acceleration more smoothly, avoid sudden and sharp acceleration, and develop a soft and gradual acceleration habit, so as to improve the passenger's riding experience and the vehicle's energy efficiency. The brake pedal force reflects the driver's braking operation. Frequent or excessive pressure on the brake pedal will cause the vehicle to brake suddenly, increase the wear of vehicle components, and also cause passengers to have a strong sense of leaning forward, which can easily cause passengers to be injured. The analysis of the brake pedal force can let the model know the driver's braking habits. In the driving suggestions, the driver can be reminded to predict the road conditions in advance, avoid emergency braking, and use the method of lightly pressing the brakes in advance and reasonably controlling the braking force to ensure the smoothness of driving and the comfort of passengers. The operation of the gear is crucial to the power output and driving state of the vehicle. Whether the driver can switch gears accurately and timely will affect the power connection and driving smoothness of the vehicle. For example, improper gear shifting timing or driving in an inappropriate gear for a long time will cause the vehicle to be underpowered or consume too much fuel. Analyzing the time queue data of gear positions can provide drivers with the correct gear shifting timing and gear selection suggestions, helping them to operate the vehicle more efficiently and improve driving consistency and fuel economy. Considering that driving behavior is a dynamic process, only obtaining indicator data at a certain moment cannot fully reflect the driver's operating habits and skill level. By obtaining the time queue data of these indicators, the model can understand the driver's operating trend during the entire simulated driving process, such as how the driver adjusts the steering wheel steering angle, accelerator pedal force, brake pedal force and gear position in different sections and different road conditions.For example, whether drivers frequently make inappropriate operations on certain roads or in certain scenarios, these patterns can be identified through time queue data, and targeted suggestions can be provided.

[0027] In the embodiment of the present application, the bus simulation driving data time series encoding module 120 is used to perform time series encoding and sorting on the time series data of the steering wheel steering angle, the time series data of the accelerator pedal force, the time series data of the brake pedal force, and the time series data of the gear position to obtain a bus driving monitoring parameter time series matrix. Specifically, Figure 2 FIG. 4 is a block diagram of a bus simulation driving data timing encoding module in a bus simulation driving data analysis system according to an embodiment of the present application. Figure 2 As shown, the bus simulation driving data timing encoding module 120 includes: a timing data encoding unit 121, which is used to respectively pass the time queue data of the steering wheel steering angle, the time queue data of the accelerator pedal force, the time queue data of the brake pedal force, and the time queue data of the gear through a timing encoder to obtain a steering wheel steering angle timing feature vector, an accelerator pedal force timing feature vector, a brake pedal force timing feature vector, and a gear timing feature vector; a timing feature sorting unit 122, which is used to arrange the steering wheel steering angle timing feature vector, the accelerator pedal force timing feature vector, the brake pedal force timing feature vector, and the gear timing feature vector to obtain the bus driving monitoring parameter timing matrix.

[0028] In an embodiment of the present application, the time series data encoding unit 121 is used to pass the time series data of the steering wheel steering angle, the time series data of the accelerator pedal force, the time series data of the brake pedal force, and the time series data of the gear through a time series encoder to obtain a steering wheel steering angle time series feature vector, a accelerator pedal force time series feature vector, a brake pedal force time series feature vector, and a gear time series feature vector. It should be understood that driving behavior is a time-dependent process. At different time points, the driver's operations are interrelated and will affect subsequent operations. For example, the change in the brake pedal force may be related to the previous accelerator pedal operation. In order to be able to capture the time dependency in different time series data well, and then to be able to dig out the sequence and mutual influence relationship between the operations, so as to more comprehensively understand the driver's driving behavior pattern, in this application, it is necessary to pass the time series data of the steering wheel steering angle, the time series data of the accelerator pedal force, the time series data of the brake pedal force, and the time series data of the gear through a time series encoder for time series processing. Through the time series encoder, the originally complex time series data can be converted into a more representative and easy-to-process feature vector. For example, the timing encoder can extract key features from the time queue data of the steering wheel steering angle, such as the frequency of change of the steering angle, the distribution of the change amplitude, etc. Here, the timing encoder described in the present application is a recurrent neural network model. When processing the time queue data of bus simulation driving, RNN can adapt well to the sequential characteristics of the data. Taking the time queue data of the accelerator pedal force as an example, RNN can process the accelerator pedal force data points one by one in chronological order, and when processing the current data point, it can use the accelerator pedal force information of the previous time step to pass the historical information to the current calculation, thereby effectively capturing the dynamic changes and dependencies in the accelerator pedal force time series.

[0029] In an embodiment of the present application, the timing feature sorting unit 122 is used to arrange the steering wheel steering angle timing feature vector, the accelerator pedal force timing feature vector, the brake pedal force timing feature vector and the gear timing feature vector to obtain the bus driving monitoring parameter timing matrix. It should be understood that bus driving is a complex process involving the coordination of multiple operating parameters. By arranging the steering wheel steering angle timing feature vector, the accelerator pedal force timing feature vector, the brake pedal force timing feature vector and the gear timing feature vector into a matrix, these different but related driving parameter information can be integrated into a data structure to comprehensively represent the driver's operation during the simulated driving process. This makes it easier for subsequent analysis to simultaneously consider the relationship between steering, accelerator, brake and gear operations, thereby more accurately depicting the overall picture of driving behavior.

[0030] In the embodiment of the present application, the bus driving stability evaluation feature generation module 130 is used to extract multi-scale time series features from the bus driving monitoring parameter time series matrix to obtain a bus driving stability evaluation feature vector. Specifically, Figure 3 FIG. 1 is a block diagram of a bus driving stability evaluation feature generation module in a bus driving simulation data analysis system according to an embodiment of the present application. Figure 3 As shown, the bus driving stability assessment feature generation module 130 includes: a first scale analysis unit 131, used to perform a first scale feature association analysis on the bus driving monitoring parameter time series matrix to obtain a first scale bus driving stability assessment feature vector; a second scale analysis unit 132, used to perform a second scale feature association analysis on the bus driving monitoring parameter time series matrix to obtain a second scale bus driving stability assessment feature vector; a second scale analysis unit 133, used to fuse the first scale bus driving stability assessment feature vector and the second scale bus driving stability assessment feature vector to obtain the bus driving stability assessment feature vector.

[0031] In an embodiment of the present application, the first scale analysis unit 131 is used to perform a first scale feature association analysis on the bus driving monitoring parameter time series matrix to obtain a first scale bus driving stability evaluation feature vector. Specifically, in an embodiment of the present application, the first scale analysis unit is used to: pass the bus driving monitoring parameter time series matrix through the first bus driving monitoring parameter association feature capturer to obtain the first scale bus driving stability evaluation feature vector. It should be understood that although the bus driving monitoring parameter time series matrix integrates the information of multiple key driving parameters, this information is still a relatively complex feature combination. It is difficult to directly judge whether the driving is stable from this matrix, and further analysis and processing are required to extract deep-level features related to driving stability. For example, each element in the matrix represents the driving parameter feature at different time points. The relationship between these features and their change patterns in the time series are crucial to evaluating driving stability, and these relationships and patterns are hidden in the matrix structure and require special analysis methods to mine. By analyzing the matrix, the coordinated change information between different driving parameters can be captured. For example, the synergistic relationship between the steering angle of the steering wheel and the vehicle speed (indirectly reflected by the strength of the accelerator pedal and the brake pedal). If the vehicle speed can remain stable when turning, or the vehicle speed is appropriately adjusted according to the size of the steering angle, this indicates that the driving behavior is relatively stable. On the contrary, if the vehicle speed changes suddenly when turning, or the throttle and brake operations are not coordinated with the steering operation, this may mean that the driving is not stable enough. Specifically, the bus driving monitoring parameter time series matrix is ​​input into the first bus driving monitoring parameter associated feature capturer for processing. Here, the first bus driving monitoring parameter associated feature capturer in the present application is a convolutional neural network model using a first-scale two-dimensional convolution kernel. The two-dimensional convolution kernel can perform convolution operations on local areas in the matrix, thereby effectively extracting local features. In the bus driving monitoring parameter time series matrix, the features of the local area may correspond to the relationship between driving parameters within a specific time period. For example, within a short time window, the convolution kernel can capture the pattern of simultaneous changes in steering angle and brake pedal force. This local feature is very useful for identifying instantaneous operation changes during driving. And because the matrix contains time series information of multiple driving parameters, the two-dimensional convolutional neural network can perform feature fusion on the parameter space dimension (different driving parameters and time points represented by the rows and columns of the matrix). It can automatically learn the combined features of different driving parameters in space and time, rather than simply considering each parameter separately. Through convolution operations, it can learn how the three parameters of accelerator pedal force, brake pedal force and steering wheel steering angle cooperate with each other to affect driving stability at a specific time point. This feature extraction method that integrates multiple parameter space dimensions can more comprehensively reflect the essential characteristics of driving behavior.

[0032] In the embodiment of the present application, the second scale analysis unit 132 is used to perform a second scale feature association analysis on the bus driving monitoring parameter time series matrix to obtain a second scale bus driving stability evaluation feature vector. Specifically, in the embodiment of the present application, the second scale analysis unit is used to: pass the bus driving monitoring parameter time series matrix through the second bus driving monitoring parameter association feature capturer to obtain the second scale bus driving stability evaluation feature vector. Here, the second bus driving monitoring parameter association feature capturer in the present application is a convolutional neural network model using a second scale two-dimensional convolution kernel, wherein the first scale is different from the second scale. It should be understood that bus driving behavior is a complex process that contains a variety of information of different frequencies and ranges. Feature extraction at a single scale may not be able to fully capture all key features of driving behavior. For example, some driving stability-related features may appear as local, rapid parameter changes (such as momentary slight steering adjustments and throttle coordination) at a smaller time-operation space scale, while other features may appear as an overall, continuous operation mode (such as maintaining a stable speed and direction on a long straight section) at a larger scale. By using two-dimensional convolution kernels of different scales, it is possible to mine features in driving behavior from different perspectives to more comprehensively evaluate driving stability.

[0033] In the embodiment of the present application, the multi-scale feature fusion unit 133 is used to fuse the first-scale bus driving stability assessment feature vector and the second-scale bus driving stability assessment feature vector to obtain the bus driving stability assessment feature vector. It should be understood that the first-scale bus driving stability assessment feature vector and the second-scale bus driving stability assessment feature vector capture the characteristics of driving behavior from different perspectives. The first scale may focus on details and local operating characteristics, such as frequent operating changes in a short period of time, such as instantaneous small-amplitude steering or slight accelerator pedal adjustment. The second scale pays more attention to macro and overall driving trends, such as the speed stability maintained on a long road section or the continuity of large-amplitude steering. In order to comprehensively consider the performance of driving behavior at the micro and macro levels, so as to obtain a more comprehensive and complete description of driving behavior characteristics, it is necessary to fuse the first-scale bus driving stability assessment feature vector and the second-scale bus driving stability assessment feature vector in the present application. Through fusion, their complementarity can be fully utilized to make up for their respective shortcomings and more accurately grasp the essential characteristics of driving behavior.

[0034] In the embodiment of the present application, the driving suggestion generating module 140 is used to analyze the bus driving stability evaluation feature vector and generate driving suggestions for the driver. Specifically, Figure 4FIG. 1 is a block diagram of a driving suggestion generation module in a bus simulation driving data analysis system according to an embodiment of the present application. Figure 4 As shown, the driving suggestion generation module 140 includes: a bus driving smoothness evaluation feature optimization unit 141, which is used to adapt the target domain feature distribution of the bus driving smoothness evaluation feature vector based on the intrinsic decomposition space to obtain an optimized bus driving smoothness evaluation feature vector; a bus driving smoothness evaluation feature parsing unit 142, which is used to pass the optimized bus driving smoothness evaluation feature vector through a suggestion generator to obtain a generation result, and the generation result is used to represent the driving suggestion to the driver.

[0035] In an embodiment of the present application, the bus driving stability evaluation feature optimization unit 141 is used to adapt the target domain feature distribution of the bus driving stability evaluation feature vector based on the intrinsic decomposition space to obtain an optimized bus driving stability evaluation feature vector. In particular, considering that the difference in the simulated driving scene may cause different change patterns of various parameters in the simulated driving process, that is, the feature distribution in different target domains in the bus driving stability evaluation feature vector may be significantly different. Specifically, if the simulated driving scene focuses on training drivers to deal with emergency situations, such as sudden appearance of pedestrians or vehicles, then in this scene, the time queue data of the brake pedal force will have a sudden high value, and the steering angle of the steering wheel may also change sharply. If the simulated driving scene is to make the driver familiar with the regular daily driving route, the changes in various parameters will be relatively stable. The traditional method lacks an adaptive adjustment mechanism for changes in feature distribution of different target domains. Once a new target domain with a feature distribution different from the training data is encountered, the model cannot automatically adjust its own parameters to adapt to the new situation, which may cause the model to deviate from the evaluation of driving behavior, thereby causing the generated driving advice to be inaccurate. Based on this, in the technical solution of the present application, the bus driving smoothness evaluation feature vector is adapted for the target domain feature distribution based on the intrinsic decomposition space to obtain an optimized bus driving smoothness evaluation feature vector.

[0036] Specifically, in an embodiment of the present application, the bus driving smoothness assessment feature optimization unit is used to: map the bus driving smoothness assessment feature vector to the intrinsic decomposition space to obtain the bus driving smoothness assessment intrinsic decomposition feature vector; extract the maximum eigenvalue and the minimum eigenvalue of the bus driving smoothness assessment intrinsic decomposition feature vector; calculate the difference between the maximum eigenvalue and the minimum eigenvalue as the bus driving smoothness assessment target domain edge anchoring description operator; extract the mean and standard deviation of the bus driving smoothness assessment intrinsic decomposition feature vector, and divide the mean by the standard deviation to obtain the bus driving smoothness assessment optimization direction description operator; based on the bus driving smoothness assessment target domain edge anchoring description operator and the bus driving smoothness assessment optimization direction description operator, match and optimize the bus driving smoothness assessment feature vector to obtain the optimized bus driving smoothness assessment feature vector.

[0037] In the embodiment of the present application, specifically, the bus driving smoothness evaluation feature vector is adapted to the target domain feature distribution based on the intrinsic decomposition space to obtain the optimized bus driving smoothness evaluation feature vector, including: processing the bus driving smoothness evaluation feature vector according to the following formula to obtain the optimized bus driving smoothness evaluation feature vector; wherein the formula is:

[0038]

[0039] Among them, V 1 represents the bus driving stability evaluation feature vector, PCA (V 1 ) indicates that V 1 Mapped to the eigendecomposition space, U 1 is the sequence of eigendecomposition vectors for bus driving stability evaluation, Λ 1 The diagonal matrix, U, is used to evaluate the smoothness of bus driving. 1 T For U 1 The transpose of v 11 、v 12 、v 1m The first, second and mth bus driving steady evaluation eigendecomposition vectors of the sequence of bus driving steady evaluation eigendecomposition vectors, λ 11 , 1m are the eigenvalues ​​of the first and mth positions of the diagonal matrix of bus driving stability evaluation, respectively. V represents the eigendecomposition eigenvector of bus driving stability evaluation, v max and v minThey represent the maximum eigenvalue and the minimum eigenvalue of the intrinsic decomposition eigenvector of the bus driving stability evaluation, α represents the edge anchoring description operator of the bus driving stability evaluation target domain, μ and σ represent the mean and standard deviation of the intrinsic decomposition eigenvector of the bus driving stability evaluation, τ represents the optimization direction description operator of the bus driving stability evaluation, ⊙ and They represent point-by-point addition, point-by-point multiplication, and point-by-point subtraction, respectively. exp represents the natural exponential function with the natural constant e as the base. V 1 ⊙-1 represents the calculation of the inverse of each eigenvalue of the bus driving smoothness evaluation feature vector, and V' represents the optimized bus driving smoothness evaluation feature vector.

[0040] In view of the above technical problems, in the technical solution of the present application, the target domain feature distribution of the bus driving smoothness evaluation feature vector is adapted based on the eigendecomposition space. Specifically, the bus driving smoothness evaluation feature vector is first mapped to the eigendecomposition space to obtain the bus driving smoothness evaluation eigendecomposition feature vector. It should be understood that with the help of eigendecomposition, the original high-dimensional bus driving smoothness evaluation feature vector can be projected into a low-dimensional space composed of an orthogonal basis. The direction in this low-dimensional space is defined by the eigenvector, and its corresponding eigenvalue describes the variance or intensity distribution of the data in different directions. The process of projecting to the eigendecomposition space not only removes the redundant noise of the original bus driving smoothness evaluation feature through dimensionality reduction, but also maps the original vector of complex distribution to a more physically intuitive decomposition dimension, so that the structured information of the bus driving smoothness evaluation feature is clearer.

[0041] Next, the maximum eigenvalue and the minimum eigenvalue of the intrinsic decomposition eigenvector of the bus driving smoothness evaluation are extracted. It should be understood that the maximum eigenvalue and the minimum eigenvalue reflect the data characteristics of the eigenvector in the most significant direction and the least significant direction, respectively. The maximum eigenvalue represents the proportion of data information in the main direction and is the core explanatory factor of the feature distribution, while the minimum eigenvalue is usually related to noise or data errors and represents the weakest change in direction. The maximum and minimum eigenvalues ​​are important representations of the spatial morphology of the driver's simulated driving data. During the analysis and optimization process, they provide a mathematical description for the global distribution characteristics of the bus driving smoothness evaluation features. At the same time, the distribution pattern of the eigenvalues ​​also implies the complexity of the data in the target domain. For example, when the maximum eigenvalue is much larger than other eigenvalues, the feature has an obvious main axis direction in the space; if the eigenvalue distribution is more uniform, there may be higher complexity or diversity.

[0042] Then, the difference between the maximum eigenvalue and the minimum eigenvalue is calculated as the edge anchoring description operator of the bus driving stability evaluation target domain. It should be understood that the eigenvalue difference clarifies the distribution range and span of the bus driving stability evaluation feature in the target domain from a geometric perspective. It corresponds to the difference between the major axis and the minor axis on the intrinsic decomposition dimension, and can be regarded as an anchoring indicator of the feature distribution, reflecting the boundary characteristics that may appear in the target domain. It is worth noting that this operator is not limited to describing the distribution differences of the driver's simulated driving data. It also provides a measurement method for capturing the information singularities that establish boundary performance in the feature space, and supporting the subsequent division and adjustment process of the target domain features. Through deep association with the target domain, the operator can further guide the subsequent optimization steps, so that the optimized feature vector is close to the domain edge attributes and enhances its in-domain adaptability.

[0043] At the same time, the mean and standard deviation of the eigendecomposition eigenvector of the bus driving smoothness evaluation are extracted, and the mean is divided by the standard deviation to obtain the bus driving smoothness evaluation optimization direction description operator. It should be understood that the mean characterizes the concentration trend of the eigenvalue in a specific direction, while the standard deviation describes its discrete characteristics within the overall range. By normalizing the ratio of the mean to the standard deviation, this bus driving smoothness evaluation optimization direction description operator provides a normalization mechanism, so that the feature optimization process can better cope with the imbalance of different data scales while ensuring the robustness of the model. In addition, the ratio of the mean to the standard deviation further provides a stable basis for directional optimization, ensuring that the optimization process will not be excessively disturbed by extreme values ​​or outliers.

[0044] Finally, based on the edge anchoring description operator of the bus driving smooth evaluation target domain and the bus driving smooth evaluation optimization direction description operator, the bus driving smooth evaluation feature vector is matched and optimized to obtain the optimized bus driving smooth evaluation feature vector. It should be understood that matching optimization is a multi-objective optimization strategy that needs to balance the edge attributes and directional attributes of feature distribution at the same time to avoid overfitting or deviation caused by preference for a certain goal. The edge anchoring description operator of the bus driving smooth evaluation target domain captures the significance of edge characteristics to ensure that the distribution of feature vectors can better reflect the global and local distribution patterns of the target domain; the bus driving smooth evaluation optimization direction description operator further provides a direction adjustment reference in this process, thereby ensuring that the optimization can be promoted in a more effective gradient direction. In this way, the optimized bus driving smooth evaluation feature vector can better reflect the driver's real driving characteristics, so that the model can generate more targeted suggestions during analysis.

[0045] In the embodiment of the present application, the bus driving smooth evaluation feature parsing unit 142 is used to pass the optimized bus driving smooth evaluation feature vector through the suggestion generator to obtain a generation result, and the generation result is used to represent the driving suggestion for the driver. It should be understood that the optimized bus driving smooth evaluation feature vector contains the driver's driving operation smoothness information, driving habit information, etc. The information contained in the driving smooth evaluation feature vector of each driver is obtained based on his personal driving behavior and is unique. Through the suggestion generator, targeted driving suggestions can be provided to each driver based on these personalized features. In detail, the suggestion generator is a machine learning model that contains a neural network structure with multiple levels for processing the input feature vector. In the training stage, it uses annotated driving data to learn how to generate reasonable suggestions. These models can automatically discover various patterns in the feature vector and match them with corresponding suggestions. Generate specific improvement measures for the driver's problems in driving operation smoothness. For example, if it is found that the driver is not stable enough when braking, the generated result may include suggestions such as "gradually increase the brake pedal force when braking and avoid suddenly stepping on the brake pedal with force." For steering operations, there may be suggestions such as "Start adjusting the steering wheel angle in advance before turning, and turn the steering wheel at a steady speed to avoid large angle adjustments during the turn."

[0046] In summary, the bus simulation driving data analysis system 100 based on the embodiment of the present application is explained, which determines whether the driver's driving is stable by performing a comprehensive time series analysis on the time queue data of the steering wheel steering angle, the time queue data of the accelerator pedal force, the time queue data of the brake pedal force, and the time queue data of the gear position during the driver's simulated driving process, and based on this, gives driving suggestions that are consistent with the driver's driving behavior. In this way, personalized feedback and suggestions can be provided to the driver.

[0047] Figure 5 FIG. 1 is a flow chart of a method for analyzing bus simulation driving data according to an embodiment of the present application. Figure 5As shown, in the bus simulation driving data analysis method, it includes: S110, obtaining the time queue data of the steering wheel steering angle, the time queue data of the accelerator pedal force, the time queue data of the brake pedal force, and the time queue data of the gear position of the driver during the simulated driving process; S120, performing time series encoding and sorting on the time queue data of the steering wheel steering angle, the time queue data of the accelerator pedal force, the time queue data of the brake pedal force, and the time queue data of the gear position to obtain a bus driving monitoring parameter time series matrix; S130, performing multi-scale time series feature extraction on the bus driving monitoring parameter time series matrix to obtain a bus driving stability evaluation feature vector; S140, analyzing the bus driving stability evaluation feature vector to generate driving suggestions for the driver.

[0048] Here, those skilled in the art can understand that the specific operations of each step in the above bus simulation driving data analysis method have been referred to above. Figures 1 to 4 The bus simulation driving data analysis system has been described in detail, and therefore, its repeated description will be omitted.

[0049] In summary, the bus simulation driving data analysis method based on the embodiment of the present application is explained, which determines whether the driver's driving is stable by performing a comprehensive time series analysis on the time queue data of the steering wheel steering angle, the time queue data of the accelerator pedal force, the time queue data of the brake pedal force, and the time queue data of the gear position during the driver's simulated driving process, and based on this, gives driving suggestions that are consistent with the driver's driving behavior. In this way, personalized feedback and suggestions can be provided to the driver.

Claims

1. A bus simulation driving data analysis system, characterized in that: include: The bus simulation driving data collection module is used to obtain the time queue data of the driver's steering wheel steering angle, the time queue data of the accelerator pedal force, the time queue data of the brake pedal force, and the time queue data of the gear position during the simulation driving process; The bus simulation driving data timing coding module is used to perform timing coding and sorting on the time queue data of the steering wheel steering angle, the time queue data of the accelerator pedal force, the time queue data of the brake pedal force, and the time queue data of the gear position to obtain a bus driving monitoring parameter timing matrix; A bus driving stability assessment feature generation module is used to extract multi-scale time series features from the bus driving monitoring parameter time series matrix to obtain a bus driving stability assessment feature vector; The driving suggestion generating module is used for analyzing the bus driving stability evaluation feature vector and generating driving suggestions for the driver.

2. The bus driving simulation data analysis system according to claim 1, characterized in that: The bus simulation driving data timing coding module comprises: A time series data encoding unit, used for respectively passing the time series data of the steering wheel steering angle, the time series data of the accelerator pedal force, the time series data of the brake pedal force, and the time series data of the gear position through a time series encoder to obtain a steering wheel steering angle time series feature vector, an accelerator pedal force time series feature vector, a brake pedal force time series feature vector, and a gear position time series feature vector; The timing feature arrangement unit is used to arrange the steering wheel steering angle timing feature vector, the accelerator pedal force timing feature vector, the brake pedal force timing feature vector and the gear position timing feature vector to obtain the bus driving monitoring parameter timing matrix.

3. The bus driving simulation data analysis system according to claim 2, characterized in that: The bus driving stability evaluation feature generation module comprises: A first scale analysis unit, configured to perform a first scale feature correlation analysis on the bus driving monitoring parameter time series matrix to obtain a first scale bus driving stability evaluation feature vector; A second scale analysis unit, used for performing a second scale feature correlation analysis on the bus driving monitoring parameter time series matrix to obtain a second scale bus driving stability evaluation feature vector; A multi-scale feature fusion unit is used to fuse the first-scale bus driving smoothness evaluation feature vector and the second-scale bus driving smoothness evaluation feature vector to obtain the bus driving smoothness evaluation feature vector.

4. The bus driving simulation data analysis system according to claim 3, characterized in that: The first scale analysis unit is used to: pass the bus driving monitoring parameter time series matrix through a first bus driving monitoring parameter associated feature capturer to obtain the first scale bus driving stability evaluation feature vector.

5. The bus driving simulation data analysis system according to claim 4, characterized in that: The second scale analysis unit is used to: pass the bus driving monitoring parameter time series matrix through a second bus driving monitoring parameter associated feature capturer to obtain the second scale bus driving stability evaluation feature vector.

6. The bus driving simulation data analysis system according to claim 5, characterized in that: The timing encoder is a recurrent neural network model, the first bus driving monitoring parameter associated feature capturer is a convolutional neural network model using a first scale two-dimensional convolution kernel, and the second bus driving monitoring parameter associated feature capturer is a convolutional neural network model using a second scale two-dimensional convolution kernel, wherein the first scale is different from the second scale.

7. The bus driving simulation data analysis system according to claim 6, characterized in that: The driving advice generating module comprises: A bus driving stability evaluation feature optimization unit, used for performing target domain feature distribution adaptation on the bus driving stability evaluation feature vector based on the intrinsic decomposition space to obtain an optimized bus driving stability evaluation feature vector; The bus driving stability evaluation feature parsing unit is used to pass the optimized bus driving stability evaluation feature vector through a suggestion generator to obtain a generation result, and the generation result is used to represent the driving suggestion to the driver.

8. The bus driving simulation data analysis system according to claim 6, characterized in that: The bus driving stability evaluation characteristic optimization unit is used to: Mapping the bus driving stability evaluation feature vector to an eigendecomposition space to obtain a bus driving stability evaluation eigendecomposition feature vector; Extracting the maximum eigenvalue and the minimum eigenvalue of the eigendecomposition eigenvector of the bus driving stability evaluation; Calculate the difference between the maximum eigenvalue and the minimum eigenvalue as an edge anchoring description operator of a bus driving stability evaluation target domain; Extracting the mean and standard deviation of the intrinsic decomposition eigenvector of the bus driving stability assessment, and dividing the mean by the standard deviation to obtain an optimized direction description operator of the bus driving stability assessment; Based on the bus driving smoothness evaluation target domain edge anchoring description operator and the bus driving smoothness evaluation optimization direction description operator, the bus driving smoothness evaluation feature vector is matched and optimized to obtain the optimized bus driving smoothness evaluation feature vector.

9. A bus simulation driving data analysis method, characterized in that: include: Obtain the time queue data of the driver's steering wheel angle, the time queue data of the accelerator pedal force, the time queue data of the brake pedal force, and the time queue data of the gear position during the simulated driving process; Performing time series coding and sorting on the time series data of the steering wheel steering angle, the time series data of the accelerator pedal force, the time series data of the brake pedal force, and the time series data of the gear position to obtain a bus driving monitoring parameter time series matrix; Performing multi-scale time series feature extraction on the bus driving monitoring parameter time series matrix to obtain a bus driving stability evaluation feature vector; The bus driving stability evaluation feature vector is analyzed to generate driving suggestions for the driver.

10. The bus simulation driving data analysis method according to claim 9, characterized in that: Analyzing the bus driving stability evaluation feature vector and generating driving suggestions for the driver, including: Performing target domain feature distribution adaptation on the bus driving stability evaluation feature vector based on the eigendecomposition space to obtain an optimized bus driving stability evaluation feature vector; The optimized bus driving stability evaluation feature vector is passed through a suggestion generator to obtain a generation result, and the generation result is used to represent driving suggestions for the driver.

Images