Precise recognition algorithm for passing characteristics of whole vehicle type based on space-time fitting

Abstract

The invention belongs to the technical field of intelligent traffic, and discloses a space-time fitting-based whole vehicle type traffic feature accurate recognition algorithm, which comprises the steps of obtaining driver behavior data, vehicle dynamic data and road supervision data; preprocessing the driver behavior data, the vehicle dynamic data and the road supervision data to obtain a driver behavior feature data set, a vehicle dynamic feature data set and a road supervision feature data set; according to the vehicle dynamic feature data set and the road supervision feature data set, obtaining a vehicle initial driving track; optimizing the initial driving track of the vehicle through an improved space-time interpolation algorithm to obtain a complete driving track of the vehicle; according to the complete driving track of the vehicle, the vehicle driving speed of each road section is calculated and compared with the limiting speed of the corresponding road section, and whether the vehicle is overspeed is judged; and traffic accidents caused by overspeed, fatigue driving and the like are effectively avoided.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation technology, and more specifically, to an algorithm for accurately identifying traffic characteristics of all vehicle types based on spatiotemporal fitting. Background Art

[0002] The patent with the patent announcement number CN111402595A discloses an intelligent management method and system for highway service areas, including monitoring equipment collecting basic information of passing vehicles at the entrance of the highway service area; generating identification records of passing vehicles in the management system, and using face recognition technology to establish a mapping relationship between the driver's face information and the identification records of passing vehicles, and guiding passing vehicles to designated parking spots based on vehicle type information; establishing a mapping relationship between accompanying personnel and drivers and vehicles; using the database resources of the management system to allocate service information corresponding to the basic information of passing vehicles and push it to the mobile terminal of the driver and / or accompanying personnel; responding to the monitoring equipment at the exit of the highway service area collecting passing vehicles, updating the identification record status of passing vehicles. Through precise intelligent terminals, with the help of information interconnection platforms, big data analysis as a means, and cloud services as an environment, the public is provided with intelligent services for the entire highway process.

[0003] The existing algorithms for accurately identifying traffic characteristics of all vehicle types have the following main problems: The assumption of uniform linear motion was not considered to simplify the trajectory prediction process. Instead, the method was directly based on complex motion models or data-driven methods, which may increase the computational complexity and reduce the prediction accuracy due to inaccurate models or data noise. The trajectory was not corrected in combination with road supervision data, which may cause the trajectory to deviate from the actual road conditions. The dynamic adjustment of the correction coefficient was not considered, which may lead to over-correction or under-correction, and it is impossible to make dynamic adjustments according to the actual environment, thus affecting the accuracy and reliability of the trajectory. Without considering the trajectory point matching process, the vehicle trajectory may be out of line with the actual direction of the road, especially in curved or complex road conditions; the predicted vehicle trajectory may deviate from the actual direction of the road, resulting in inaccurate vehicle path planning, affecting the effectiveness of navigation and autonomous driving systems; lack of effective trajectory point matching and direction correction, the vehicle's motion trajectory may not fully match the actual road conditions, reducing the safety and efficiency of vehicle driving, and may also cause traffic accidents, causing damage to people and property; Without using the spatiotemporal interpolation algorithm, it is impossible to infer the trajectory points of unknown time points based on the known trajectory points; as a result, there are gaps or breakpoints in the trajectory in the time dimension, which cannot fully reflect the trajectory characteristics of the vehicle's continuous driving; without using the Kriging interpolation formula and the weight coefficient adjustment formula, the system cannot dynamically adjust the weight of each trajectory point; affected by noise or data errors, the trajectory may show unreasonable jumps and jitters, reducing the smoothness of the data; when the spatiotemporal variation function and covariance function are not constructed, it is impossible to quantify the correlation between the trajectory points in space and time, resulting in the interpolation results being unable to accurately reflect the actual trajectory changes; inaccurate trajectory data may lead to deviations in the judgment of the vehicle's driving status, thereby increasing potential safety risks.

[0004] In view of this, the present invention proposes an accurate recognition algorithm for traffic characteristics of all vehicle types based on spatiotemporal fitting to solve the above problems. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned purpose, the present invention provides the following technical solution: an accurate recognition algorithm for traffic characteristics of all vehicle types based on time-space fitting, comprising: S1. Obtain driver behavior data, vehicle dynamic data and road supervision data; S2, preprocessing the driver behavior data, vehicle dynamic data and road supervision data to obtain a driver behavior feature data set, a vehicle dynamic feature data set and a road supervision feature data set; S3. Obtaining a preliminary vehicle driving trajectory based on the vehicle dynamic feature data set and the road supervision feature data set; optimizing the preliminary vehicle driving trajectory by improving the spatiotemporal interpolation algorithm to obtain a complete vehicle driving trajectory; S4. Calculate the vehicle speed at each road section based on the complete driving trajectory of the vehicle, and compare it with the speed limit of the corresponding road section to determine whether the vehicle is speeding; S5. Obtain a fatigue driving risk prediction model based on the driver behavior feature data set and the vehicle's complete driving trajectory training, and predict the risk probability of fatigue driving based on the fatigue driving risk prediction model; determine whether the driver is fatigue driving by predicting the risk probability of fatigue driving; S6. If the vehicle exceeds the speed limit, the on-board intelligent control terminal automatically issues a vehicle speeding alarm message to remind the driver to slow down in time; if the driver is judged to be driving fatigued, the on-board intelligent control terminal automatically generates a fatigue warning instruction to remind the driver that he is in a fatigue driving state.

[0006] Furthermore, the vehicle dynamic data includes the vehicle's real-time position, vehicle real-time speed and vehicle driving time; the road supervision data includes road type, vehicle checkpoint passing time and road section speed limit information; the driver behavior data includes driver rest time, driver eye closing time, driver blinking frequency, driver heart rate, number of emergency brakes and line of sight deviation rate.

[0007] Furthermore, the method of preprocessing the vehicle dynamic data, the road supervision data and the driver behavior data to obtain the vehicle dynamic feature data set, the road supervision feature data set and the driver behavior feature data set includes: Perform data cleaning on the acquired vehicle dynamic data, road supervision data and driver behavior data, and use linear interpolation to fill in missing values ​​in the acquired vehicle dynamic data, road supervision data and driver behavior data; The density clustering algorithm is used to detect outliers on the cleaned data and remove the identified outliers; the data after removing the outliers is standardized and normalized to obtain the preprocessed vehicle dynamic feature data set, road supervision feature data set and driver behavior feature data set.

[0008] Furthermore, the method for obtaining a preliminary vehicle driving trajectory according to the vehicle dynamic feature data set and the road supervision feature data set includes: S41. Preset the real-time position coordinate sequence of the vehicle at each time point in the vehicle dynamic feature data set as: ;in, The real-time position coordinate sequence of the vehicle at each time point; For time point The coordinates of the vehicle trajectory points; is the horizontal coordinate of the vehicle trajectory point; is the ordinate of the vehicle trajectory point; is the index of the time point; The real-time speed sequence of the vehicle at each time point in the preset vehicle dynamic feature data set is: ;in, The real-time speed sequence of vehicles at each time point; For time point The real-time speed of the vehicle; The vehicle driving time series in the preset vehicle dynamic feature dataset is: ;in, is the vehicle travel time series; For the The time value of each sampling point; the vehicle dynamic feature data set is denoised by Wiener filtering; S42, preset vehicles at each time point The vehicle moves in a straight line at a uniform speed, and the current time point of the vehicle is deduced through the position update formula based on the vehicle dynamic feature data set. The initial driving trajectory of; the position update formula is: ;in, For the vehicle at time The coordinates of the trajectory points at time ; For at time point The heading angle of the vehicle at time ; is the time step, which indicates the interval between two sampling times; is the cosine value of the vehicle's heading angle; is the sine value of the vehicle's heading angle; S43, combining the road supervision feature data set, and correcting the current time point of the deduced vehicle through the trajectory correction formula The initial driving trajectory of ; the trajectory correction formula is: ;in, is the corrected vehicle at time point The coordinates of the trajectory points at time ; is the shortest distance vector from the vehicle trajectory point to the center line of the road; is the correction factor used to balance the correction strength of the trajectory deviation; The correction coefficient is dynamically adjusted through the correction coefficient adjustment formula; the correction coefficient adjustment formula is: ;in, is the correction coefficient after dynamic adjustment; is the weight factor for adjusting the correction coefficient; is the distance sensitivity coefficient; is the modulus of the shortest distance from the trajectory point to the center line of the road; The maximum speed allowed for vehicles on the road; is the smoothing factor; S44, matching each trajectory point to the nearest road center point using a trajectory point matching formula; the trajectory point matching formula is: ;in, For time point Time and vehicle trajectory points The nearest road center point to match; is the set of road centerline points; The center point of the road The coordinates of The vehicle trajectory point With road center The Euclidean distance between For the collection Choose a point in The value of is the smallest; S45. Calculate the cosine similarity of the vehicle speed vector and the road direction vector by using the cosine similarity formula, and correct the moving direction of the vehicle so that the moving direction of the vehicle is consistent with the road direction; the cosine similarity formula is: ;in, is the cosine similarity between the vehicle speed vector and the road direction vector, is the angle between the vehicle's moving direction and the road direction; for The vehicle velocity vector at position; for The road direction vector; for The modulus of the vehicle velocity vector at the time position; is the modulus of the road direction vector; like 1, then the vehicle's direction of movement is determined to be consistent with the road direction; if 1, it is determined that the vehicle's direction of movement deviates from the road direction and needs to be corrected; if 1, then it is determined that the vehicle's moving direction is opposite to the road direction and needs to be corrected; S46. Correcting the speed vector of the vehicle so that its moving direction is consistent with the road direction, thereby obtaining a final corrected preliminary driving trajectory of the vehicle.

[0009] Furthermore, the method of optimizing the preliminary driving trajectory of the vehicle by improving the spatiotemporal interpolation algorithm to obtain the complete driving trajectory of the vehicle includes: S51, the final corrected preliminary driving trajectory of the vehicle is expressed as: ;in, It is a set of all points of the final corrected preliminary driving trajectory of the vehicle; For time point The corrected trajectory point sequence, including coordinate information and sampling time; is the total number of time points of trajectory sampling, that is, the total number of trajectory points; is the sampling time point of the vehicle's initial driving trajectory points, ranging from arrive ; S52. Construct a spatiotemporal variogram to measure the correlation of trajectory points in space and time dimensions; the spatiotemporal variogram is: ;in, The spatiotemporal variogram is a measure of the correlation between trajectory points in spatial and temporal dimensions; is the spatial distance between trajectory points; From the current time To the next time point time difference; is the observation value of the trajectory point; For the next time point Observation values ​​of trajectory points; is the expected value of the trajectory point; S53, describing the spatial and temporal correlation between trajectory points by means of a covariance function; the covariance function is: ;in, is the covariance between trajectory points; is the variance of the trajectory point observations; is the spatial range parameter, controlling the spatial distance The degree of influence on the covariance; is the time range parameter, controlling the time difference The degree of influence on the covariance; S54. For a known time point Corrected trajectory point sequence , using the Kriging interpolation formula to calculate the time point Corresponding trajectory points; the Kriging interpolation formula is: ;in, For time point The corresponding trajectory points; is the observed value of the known trajectory point; is the weight coefficient for each known trajectory point; For the The position of each trajectory point is known. Contains spatial location and timestamp ; is the index of the known trajectory point; S55, dynamically adjust the weight coefficient of each known trajectory point through the weight coefficient adjustment formula; expand and smoothly optimize the trajectory points in the time dimension by gradually updating the trajectory points, and generate a arrive The optimized trajectory points are calculated and the time dimension is gradually expanded to obtain the complete driving trajectory of the vehicle: ;in, For time point Optimized trajectory point sequence; For time point The corresponding optimized horizontal axis; For time point The corresponding optimized ordinate; is the total number of trajectory points after optimization, and .

[0010] Furthermore, the method of dynamically adjusting the weight coefficient of each known trajectory point by means of a weight coefficient adjustment formula includes: Construct a weight coefficient adjustment formula to adjust the weight coefficient of each known trajectory point; the weight coefficient adjustment formula is: ;in, is the weight coefficient of each known trajectory point after adjustment; is the spatial distance from the trajectory point The associated distance decay function; Indicates normalization of all known trajectory points; the distance decay function expression is: ;in, is the minimum distance constant; is the spatial distance between trajectory points The square of .

[0011] Furthermore, the method for calculating the vehicle speed at each road section according to the complete vehicle driving trajectory includes: By dividing the vehicle's complete driving trajectory into road sections, calculate the average speed of all vehicle trajectory points in each road section, and thus obtain the vehicle speed of the road section; If the vehicle speed of each road section is greater than the speed limit of the corresponding road section, the vehicle is judged to be speeding; if the vehicle speed of each road section is less than or equal to the speed limit of the corresponding road section, the vehicle is judged not to be speeding.

[0012] Furthermore, the training method of the fatigue driving risk prediction model includes: The dataset is divided into training set, validation set and test set to train the model, evaluate the model performance and verify the generalization ability of the model; the sample set is a subset of the dataset, each sample set includes the historical driver behavior feature dataset and the complete driving trajectory of the vehicle and the corresponding risk probability of fatigue driving; GBDT is selected as the specific implementation of the gradient boosting tree model to handle the regression task; Initialize GBDT parameters and define the objective function; GBDT parameters include the maximum number of leaf nodes in each tree, learning rate, number of trees and maximum depth of trees; use the historical driver behavior feature data set and the complete vehicle driving trajectory as input data, and the corresponding fatigue driving risk probability as output label to train the fatigue driving risk prediction model; the fatigue driving risk prediction model is a gradient boosting tree model; Use mean square error as the loss function to measure the difference between the model's predicted value and the actual value; in each iteration, GBDT builds a new decision tree to fit the negative gradient of the loss function of the previous step, uses the negative gradient of the objective function as the learning target of the new tree, and adjusts the contribution of the new tree to the final result through the learning rate to minimize the loss function; Use the validation set to evaluate the performance of the model, tune the model, and adjust the model parameters based on its performance feedback until the model performance no longer improves or reaches the preset stopping condition; use the trained fatigue driving risk prediction model to predict the current driver behavior feature dataset and the vehicle's complete driving trajectory to obtain the risk probability of fatigue driving.

[0013] Furthermore, the method for determining whether the driver is driving fatigued by predicting the risk probability of driving fatigued includes: If the predicted risk probability of fatigue driving is less than the preset risk probability threshold of fatigue driving, it is determined that the driver is not driving fatigued; If the predicted risk probability of fatigue driving is greater than or equal to a preset risk probability threshold of fatigue driving, the driver is judged to be fatigue driving.

[0014] The accurate recognition system of traffic characteristics of all vehicle types based on time-space fitting includes: A data acquisition module, used to acquire vehicle dynamic data, road supervision data and driver behavior data; A data processing module is used to pre-process the vehicle dynamic data, the road supervision data and the driver behavior data to obtain a vehicle dynamic feature data set, a road supervision feature data set and a driver behavior feature data set; The vehicle trajectory acquisition module is used to obtain the preliminary driving trajectory of the vehicle based on the vehicle dynamic feature data set and the road supervision feature data set; the preliminary driving trajectory of the vehicle is optimized by improving the spatiotemporal interpolation algorithm to obtain the complete driving trajectory of the vehicle; The vehicle speeding assessment module is used to calculate the vehicle speed of each road section based on the complete driving trajectory of the vehicle, and compare it with the speed limit of the corresponding road section to determine whether the vehicle is speeding; The fatigue driving risk assessment module is used to obtain a fatigue driving risk prediction model based on the driver behavior feature data set and the vehicle's complete driving trajectory training, and predict the risk probability of fatigue driving based on the fatigue driving risk prediction model; and judge whether the driver is fatigue driving by predicting the risk probability of fatigue driving; Feedback reminder module: if the vehicle exceeds the speed limit, the on-board intelligent control terminal will automatically issue a vehicle speeding alarm message to remind the driver to slow down in time; if the driver is judged to be driving fatigued, the on-board intelligent control terminal will automatically generate a fatigue warning instruction to remind the driver that he is in a fatigue driving state; each module is connected by wired and / or wireless means.

[0015] The technical effects and advantages of the accurate recognition algorithm of traffic characteristics of all vehicle types based on time-space fitting of the present invention are as follows: By presupposing that the vehicle moves in a straight line at a uniform speed at each time point and deducing the vehicle trajectory through the position update formula, the trajectory prediction process is simplified to a certain extent while maintaining a high degree of accuracy. In combination with the road supervision feature data set, the initial driving trajectory is corrected through the trajectory correction formula, further improving the accuracy of the trajectory. A correction coefficient is introduced into the trajectory correction formula to balance the correction intensity of the trajectory deviation. Through the correction coefficient adjustment formula, the correction coefficient is dynamically adjusted according to the actual situation, making the trajectory correction more reasonable and reliable. Matching each trajectory point to the nearest road center point through the trajectory point matching formula helps to combine the vehicle trajectory with the actual road direction, enhancing the reliability and practicality of the trajectory; calculating the cosine similarity of the vehicle speed vector and the road direction vector through the cosine similarity formula, correcting the vehicle's movement direction; helping to ensure that the vehicle's movement direction is consistent with the road direction, improving the safety and efficiency of vehicle driving; when the vehicle's movement direction deviates from or is opposite to the road direction, it can be corrected in time to avoid potential driving risks; Through the spatiotemporal interpolation algorithm, the trajectory points of unknown time points can be inferred based on the known trajectory point sequence, thereby expanding and optimizing the integrity of the trajectory in the time dimension; it is of great significance for understanding the continuous driving behavior of vehicles and analyzing traffic flow characteristics; through the Kriging interpolation formula and the weight coefficient adjustment formula, this method can calculate and dynamically adjust the weight coefficient of each known trajectory point, thereby generating a smoother and continuous trajectory; it helps to reduce trajectory jumping and jittering caused by data missing or errors, and improve the accuracy and readability of the trajectory; constructing the spatiotemporal variation function and covariance function, this method can measure the correlation of trajectory points in the spatial and temporal dimensions, and use this correlation information to optimize the interpolation results; it helps to more accurately reflect the spatial position and time changes of the vehicle during driving, and improve the accuracy and reliability of the trajectory. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a schematic diagram of the flow of the algorithm for accurate recognition of traffic characteristics of all vehicle types based on time-space fitting of the present invention; Figure 2 It is a schematic diagram of the structure of the algorithm system for accurately identifying traffic characteristics of all vehicle types based on time-space fitting of the present invention. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] Example 1 See also Figure 1 As shown, the accurate recognition algorithm of traffic characteristics of all vehicle types based on time-space fitting in this embodiment includes: S1. Obtain driver behavior data, vehicle dynamic data and road supervision data; S2, preprocessing the driver behavior data, vehicle dynamic data and road supervision data to obtain a driver behavior feature data set, a vehicle dynamic feature data set and a road supervision feature data set; S3. Obtaining a preliminary vehicle driving trajectory based on the vehicle dynamic feature data set and the road supervision feature data set; optimizing the preliminary vehicle driving trajectory by improving the spatiotemporal interpolation algorithm to obtain a complete vehicle driving trajectory; S4. Calculate the vehicle speed at each road section based on the complete driving trajectory of the vehicle, and compare it with the speed limit of the corresponding road section to determine whether the vehicle is speeding; S5. Obtain a fatigue driving risk prediction model based on the driver behavior feature data set and the vehicle's complete driving trajectory training, and predict the risk probability of fatigue driving based on the fatigue driving risk prediction model; determine whether the driver is fatigue driving by predicting the risk probability of fatigue driving; S6. If the vehicle exceeds the speed limit, the on-board intelligent control terminal automatically issues a vehicle speeding alarm message to remind the driver to slow down in time; if the driver is judged to be driving fatigued, the on-board intelligent control terminal automatically generates a fatigue warning instruction to remind the driver that he is in a fatigue driving state.

[0019] Vehicle dynamic data includes the vehicle's real-time position, real-time speed and driving time; road supervision data includes road type, vehicle checkpoint passing time and road section speed limit information; driver behavior data includes driver rest time, driver eye closing time, driver blinking frequency, driver heart rate, number of emergency brakes and line of sight deviation rate.

[0020] The method of preprocessing the vehicle dynamic data, the road supervision data and the driver behavior data to obtain the vehicle dynamic feature data set, the road supervision feature data set and the driver behavior feature data set includes: Perform data cleaning on the acquired vehicle dynamic data, road supervision data and driver behavior data, and use linear interpolation to fill in missing values ​​in the acquired vehicle dynamic data, road supervision data and driver behavior data; The density clustering algorithm is used to detect outliers on the cleaned data and remove the identified outliers; the data after removing the outliers is standardized and normalized to obtain the preprocessed vehicle dynamic feature data set, road supervision feature data set and driver behavior feature data set.

[0021] The method for obtaining the preliminary driving trajectory of the vehicle according to the vehicle dynamic feature data set and the road supervision feature data set includes: S41. Preset the real-time position coordinate sequence of the vehicle at each time point in the vehicle dynamic feature data set as: ;in, The real-time position coordinate sequence of the vehicle at each time point; For time point The coordinates of the vehicle trajectory points; is the horizontal coordinate of the vehicle trajectory point; is the ordinate of the vehicle trajectory point; is the index of the time point; The real-time speed sequence of the vehicle at each time point in the preset vehicle dynamic feature data set is: ;in, The real-time speed sequence of vehicles at each time point; For time point The real-time speed of the vehicle; The vehicle driving time series in the preset vehicle dynamic feature dataset is: ;in, is the vehicle travel time series; For the The time value of each sampling point; the vehicle dynamic feature data set is denoised by Wiener filtering; S42, preset vehicles at each time point The vehicle moves in a straight line at a uniform speed, and the current time point of the vehicle is deduced through the position update formula based on the vehicle dynamic feature data set. The initial driving trajectory of; the position update formula is: ;in, For the vehicle at time The coordinates of the trajectory points at time ; For at time point The heading angle of the vehicle at time ; is the time step, which indicates the interval between two sampling times; is the cosine value of the vehicle's heading angle; is the sine value of the vehicle's heading angle; S43, combining the road supervision feature data set, and correcting the current time point of the deduced vehicle through the trajectory correction formula The initial driving trajectory of ; the trajectory correction formula is: ;in, is the corrected vehicle at time point The coordinates of the trajectory points at time ; is the shortest distance vector from the vehicle trajectory point to the center line of the road; is the correction factor used to balance the correction strength of the trajectory deviation; The correction coefficient is dynamically adjusted through the correction coefficient adjustment formula; the correction coefficient adjustment formula is: ;in, is the correction coefficient after dynamic adjustment; is the weight factor for adjusting the correction coefficient; is the distance sensitivity coefficient, which controls the influence of the distance from the trajectory point to the center line of the road on the correction coefficient; is the modulus of the shortest distance from the trajectory point to the center line of the road; The maximum speed allowed for vehicles on the road; is a smoothing factor, between 0 and 1, used to control the influence of the original correction coefficient on the dynamically adjusted correction coefficient; S44, matching each trajectory point to the nearest road center point using a trajectory point matching formula; the trajectory point matching formula is: ;in, For time point Time and vehicle trajectory points The nearest road center point to match; is the set of road centerline points; The center point of the road The coordinates of The vehicle trajectory point With road center The Euclidean distance between For the collection Choose a point in The value of is the smallest; S45. Calculate the cosine similarity of the vehicle speed vector and the road direction vector by using the cosine similarity formula, and correct the moving direction of the vehicle so that the moving direction of the vehicle is consistent with the road direction; the cosine similarity formula is: ;in, is the cosine similarity between the vehicle speed vector and the road direction vector, is the angle between the vehicle's moving direction and the road direction; for The vehicle velocity vector at position; for The road direction vector; for The modulus of the vehicle velocity vector at the time position; is the modulus of the road direction vector; like 1, then the vehicle's direction of movement is determined to be consistent with the road direction; if 1, it is determined that the vehicle's direction of movement deviates from the road direction and needs to be corrected; if 1, then it is determined that the vehicle's moving direction is opposite to the road direction and needs to be corrected; S46. Correcting the speed vector of the vehicle so that its moving direction is consistent with the road direction, thereby obtaining a final corrected preliminary driving trajectory of the vehicle.

[0022] For example, suppose the vehicle is at time = 0, the coordinates of the position trajectory point are =(0,0), the vehicle's speed is 20km / h, and its direction is , and the time step is 1 second; convert the speed unit to meters per second: ; Calculate the next position, since the direction of the vehicle is ,so , ; Calculate the trajectory point coordinates of the next position of the vehicle through the position update formula: ; Combined with the road supervision feature dataset, the current time point of the deduced vehicle is corrected through the trajectory correction formula The initial driving trajectory of the vehicle is: ; Correction coefficient is 0.1. According to the trajectory correction formula, the vehicle is at the time point after correction. The coordinates of the trajectory point at time are: ; The weight factor of the adjustment correction coefficient is 1, the distance sensitivity coefficient is 0.5, and the modulus of the shortest distance from the trajectory point to the center line of the road is : The maximum speed allowed for vehicles on the road for ; The original correction coefficient is 0.035, and the smoothing factor is 0.4, so the correction coefficient after dynamic adjustment is 0.05; If we only rely on the information of the current trajectory point, the correction coefficient may be affected by the instantaneous error, causing the trajectory correction to be too drastic or too slow. By introducing the correction coefficient of the previous moment , and use a smoothing factor To balance the weight between the current correction coefficient and the historical correction coefficient, the correction process is smoother and the influence of instantaneous errors is avoided. It effectively avoids possible mutations in the trajectory correction process and makes the correction more in line with the needs of actual driving scenarios.

[0023] Assume that the set of road centerline points is: , and ; Match each trajectory point to the nearest road center point through the trajectory point matching formula; After correction, the vehicle is at time point The coordinates of the trajectory point at arrive The distance is: ; After correction, the vehicle is at time point The coordinates of the trajectory point at arrive The distance is: ; After correction, the vehicle is at time point The coordinates of the trajectory point at arrive The distance is: ; By using the trajectory point matching formula, the shortest distance is , so Matches ; Ensure that the direction of the vehicle trajectory is consistent with the actual driving direction of the road to avoid directional errors.

[0024] Assume that the direction of the road centerline is from Extending forward, at a point in time When , the vehicle velocity vector is ; The road direction vector can be simply estimated by the difference between two adjacent points on the road, that is, ; The modulus of the vehicle velocity vector at the current position for ; Module length of road direction vector for ; The cosine similarity of the vehicle speed vector and the road direction vector is calculated by the cosine similarity formula, and the vehicle's movement direction is corrected to make the vehicle's movement direction consistent with the road direction; then the cosine similarity of the vehicle speed vector and the road direction vector is ; because 1, it is determined that the vehicle's moving direction is opposite to the road direction. At this time, it is necessary to recalibrate the vehicle's trajectory direction to make it consistent with the road direction; by correcting the vehicle's velocity vector, its moving direction is consistent with the road direction.

[0025] The method of optimizing the preliminary driving trajectory of the vehicle by improving the spatiotemporal interpolation algorithm to obtain the complete driving trajectory of the vehicle includes: S51, the final corrected preliminary driving trajectory of the vehicle is expressed as: ;in, It is a set of all points of the final corrected preliminary driving trajectory of the vehicle; For time point The corrected trajectory point sequence, including coordinate information and sampling time; is the total number of time points of trajectory sampling, that is, the total number of trajectory points; is the sampling time point of the vehicle's initial driving trajectory points, ranging from arrive ; S52. Construct a spatiotemporal variogram to measure the correlation of trajectory points in space and time dimensions; the spatiotemporal variogram is: ;in, The spatiotemporal variogram is a measure of the correlation between trajectory points in spatial and temporal dimensions; is the spatial distance between trajectory points; From the current time To the next time point time difference; is the observation value of the trajectory point, indicating that and time point The trajectory points on For the next time point Observation values ​​of trajectory points; is the expected value of the trajectory point; S53, describing the spatial and temporal correlation between trajectory points by means of a covariance function; the covariance function is: ;in, is the covariance between trajectory points; is the variance of the trajectory point observations; is the spatial range parameter, controlling the spatial distance The degree of influence on the covariance; is the time range parameter, controlling the time difference The degree of influence on the covariance; S54, Time point Corrected trajectory point sequence A known time point Corrected trajectory point sequence , for a known time point Corrected trajectory point sequence , using the Kriging interpolation formula to calculate the time point Corresponding trajectory points; the Kriging interpolation formula is: ;in, For time point The corresponding trajectory points; is the observed value of the known trajectory point; is the weight coefficient for each known trajectory point; For the The position of each trajectory point is known. Contains spatial location and timestamp ; is the index of the known trajectory point; S55, dynamically adjust the weight coefficient of each known trajectory point through the weight coefficient adjustment formula; expand and smoothly optimize the trajectory points in the time dimension by gradually updating the trajectory points, and generate a arrive The optimized trajectory points are calculated and the time dimension is gradually expanded to obtain the complete driving trajectory of the vehicle: ;in, For time point Optimized trajectory point sequence; For time point The corresponding optimized horizontal axis; For time point The corresponding optimized ordinate; is the total number of trajectory points after optimization, and .

[0026] The method of dynamically adjusting the weight coefficient of each known trajectory point by means of a weight coefficient adjustment formula includes: Construct a weight coefficient adjustment formula to adjust the weight coefficient of each known trajectory point; the weight coefficient adjustment formula is: ;in, is the weight coefficient of each known trajectory point after adjustment; is the spatial distance from the trajectory point The associated distance decay function; It means that all known trajectory points are normalized so that the sum of the weights of all trajectory points is 1; the distance decay function expression is: ;in, is the minimum distance constant, used to avoid the denominator being zero; is the spatial distance between trajectory points The square of .

[0027] For example, suppose there are three trajectory points , and , and their spatial distances from the interpolation points are , and , the minimum distance constant is 0.1; calculate the distance attenuation function value of each trajectory point: ; ; ; Calculate the sum of all decay function values: ; Normalize the weight of each trajectory point: ; ; ; This means that during the interpolation process, trajectory points closer to the interpolation point (such as ) will receive a larger weight, while trajectory points that are farther away (such as and ) gets a smaller weight.

[0028] According to the complete driving trajectory of the vehicle, the method of calculating the driving speed of the vehicle in each road section includes: By dividing the vehicle's complete driving trajectory into road sections, calculate the average speed of all vehicle trajectory points in each road section, and thus obtain the vehicle speed of the road section; If the vehicle speed of each road section is greater than the speed limit of the corresponding road section, the vehicle is judged to be speeding; if the vehicle speed of each road section is less than or equal to the speed limit of the corresponding road section, the vehicle is judged not to be speeding.

[0029] The training method of the fatigue driving risk prediction model includes: The dataset is divided into training set, validation set and test set to train the model, evaluate the model performance and verify the generalization ability of the model; the sample set is a subset of the dataset, each sample set includes the historical driver behavior feature dataset and the complete driving trajectory of the vehicle and the corresponding risk probability of fatigue driving; GBDT is selected as the specific implementation of the gradient boosting tree model to handle the regression task; Initialize GBDT parameters and define the objective function; GBDT parameters include the maximum number of leaf nodes in each tree, learning rate, number of trees and maximum depth of trees; use the historical driver behavior feature data set and the complete vehicle driving trajectory as input data, and the corresponding fatigue driving risk probability as output label to train the fatigue driving risk prediction model; the fatigue driving risk prediction model is a gradient boosting tree model; The mean square error is used as the loss function to measure the difference between the model's predicted value and the actual value; the mean square error loss function is: ;in, is the number of samples in the data set; For the data set The actual value of samples; For the data set The predicted value of samples; is the index of the sample; In each iteration, GBDT builds a new decision tree to fit the negative gradient of the loss function of the previous step, uses the negative gradient of the objective function as the learning target of the new tree, and adjusts the contribution of the new tree to the final result through the learning rate to minimize the loss function. Use the validation set to evaluate the performance of the model, tune the model, and adjust the model parameters based on its performance feedback until the model performance no longer improves or reaches the preset stopping condition; use the trained fatigue driving risk prediction model to predict the current driver behavior feature dataset and the vehicle's complete driving trajectory to obtain the risk probability of fatigue driving.

[0030] The method of judging whether the driver is driving fatigued by predicting the risk probability of driving fatigued includes: If the predicted risk probability of fatigue driving is less than the preset risk probability threshold of fatigue driving, it is determined that the driver is not driving fatigued; If the predicted risk probability of fatigue driving is greater than or equal to a preset risk probability threshold of fatigue driving, the driver is judged to be fatigue driving.

[0031] The preset fatigue driving risk probability threshold is set by the staff, and the risk probabilities of different fatigue driving are collected through the vehicle-mounted intelligent control terminal, and the average value of the risk probabilities of multiple fatigue driving is taken as the preset fatigue driving risk probability threshold; In this embodiment, by presetting the vehicle to move in a straight line at a uniform speed at each time point and deducing the vehicle trajectory through the position update formula, the trajectory prediction process is simplified to a certain extent while maintaining a high degree of accuracy; in combination with the road supervision feature data set, the initial driving trajectory is corrected through the trajectory correction formula, further improving the accuracy of the trajectory; a correction coefficient is introduced into the trajectory correction formula to balance the correction intensity of the trajectory deviation; through the correction coefficient adjustment formula, the correction coefficient is dynamically adjusted according to the actual situation, making the trajectory correction more reasonable and reliable; Matching each trajectory point to the nearest road center point through the trajectory point matching formula helps to combine the vehicle trajectory with the actual road direction, enhancing the reliability and practicality of the trajectory; calculating the cosine similarity of the vehicle speed vector and the road direction vector through the cosine similarity formula, correcting the vehicle's movement direction; helping to ensure that the vehicle's movement direction is consistent with the road direction, improving the safety and efficiency of vehicle driving; when the vehicle's movement direction deviates from or is opposite to the road direction, it can be corrected in time to avoid potential driving risks; Through the spatiotemporal interpolation algorithm, the trajectory points of unknown time points can be inferred based on the known trajectory point sequence, thereby expanding and optimizing the integrity of the trajectory in the time dimension; it is of great significance for understanding the continuous driving behavior of vehicles and analyzing traffic flow characteristics; through the Kriging interpolation formula and the weight coefficient adjustment formula, this method can calculate and dynamically adjust the weight coefficient of each known trajectory point, thereby generating a smoother and continuous trajectory; it helps to reduce trajectory jumping and jittering caused by data missing or errors, and improve the accuracy and readability of the trajectory; constructing the spatiotemporal variation function and covariance function, this method can measure the correlation of trajectory points in the spatial and temporal dimensions, and use this correlation information to optimize the interpolation results; it helps to more accurately reflect the spatial position and time changes of the vehicle during driving, and improve the accuracy and reliability of the trajectory.

[0032] Example 2 See also Figure 2 As shown, the accurate recognition system for traffic characteristics of all vehicle types based on time-space fitting in this embodiment includes: A data acquisition module, used to acquire vehicle dynamic data, road supervision data and driver behavior data; A data processing module is used to pre-process the vehicle dynamic data, the road supervision data and the driver behavior data to obtain a vehicle dynamic feature data set, a road supervision feature data set and a driver behavior feature data set; The vehicle trajectory acquisition module is used to obtain the preliminary driving trajectory of the vehicle based on the vehicle dynamic feature data set and the road supervision feature data set; the preliminary driving trajectory of the vehicle is optimized by improving the spatiotemporal interpolation algorithm to obtain the complete driving trajectory of the vehicle; The vehicle speeding assessment module is used to calculate the vehicle speed of each road section based on the complete driving trajectory of the vehicle, and compare it with the speed limit of the corresponding road section to determine whether the vehicle is speeding; The fatigue driving risk assessment module is used to obtain a fatigue driving risk prediction model based on the driver behavior feature data set and the vehicle's complete driving trajectory training, and predict the risk probability of fatigue driving based on the fatigue driving risk prediction model; and judge whether the driver is fatigue driving by predicting the risk probability of fatigue driving; Feedback reminder module: if the vehicle exceeds the speed limit, the on-board intelligent control terminal will automatically issue a vehicle speeding alarm message to remind the driver to slow down in time; if the driver is judged to be driving fatigued, the on-board intelligent control terminal will automatically generate a fatigue warning instruction to remind the driver that he is in a fatigue driving state; each module is connected by wired and / or wireless means.

[0033] Since the electronic device introduced in this embodiment is an electronic device used to implement the algorithm for accurately identifying the traffic characteristics of all vehicle models based on time-space fitting in the embodiment of this application, based on the algorithm for accurately identifying the traffic characteristics of all vehicle models based on time-space fitting introduced in the embodiment of this application, the technical personnel of this field can understand the specific implementation of the electronic device of this embodiment and its various variations, so how the electronic device implements the method in the embodiment of this application is not introduced in detail here. As long as the technical personnel of this field implement the electronic device used in the algorithm for accurately identifying the traffic characteristics of all vehicle models based on time-space fitting in the embodiment of this application, it belongs to the scope of protection of this application.

[0034] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0035] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technical users in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. An accurate recognition algorithm for traffic characteristics of all vehicle types based on spatiotemporal fitting, characterized by: include: S1. Obtain driver behavior data, vehicle dynamic data and road supervision data; S2, preprocessing the driver behavior data, vehicle dynamic data and road supervision data to obtain a driver behavior feature data set, a vehicle dynamic feature data set and a road supervision feature data set; S3. Obtaining a preliminary vehicle driving trajectory based on the vehicle dynamic feature data set and the road supervision feature data set; optimizing the preliminary vehicle driving trajectory by improving the spatiotemporal interpolation algorithm to obtain a complete vehicle driving trajectory; S4. Calculate the vehicle speed at each road section based on the complete driving trajectory of the vehicle, and compare it with the speed limit of the corresponding road section to determine whether the vehicle is speeding; S5. Obtain a fatigue driving risk prediction model based on the driver behavior feature data set and the vehicle's complete driving trajectory training, and predict the risk probability of fatigue driving based on the fatigue driving risk prediction model; Determine whether the driver is driving fatigued by predicting the risk probability of driving fatigued; S6. If the vehicle exceeds the speed limit, the on-board intelligent control terminal will automatically send out a speeding alarm message to remind the driver to slow down in time; If the driver is judged to be driving fatigued, the on-board intelligent control terminal will automatically generate a fatigue warning instruction to remind the driver that he is in a fatigue driving state.

2. The accurate recognition algorithm for traffic characteristics of all vehicle types based on time-space fitting according to claim 1 is characterized in that: The vehicle dynamic data includes the vehicle's real-time position, vehicle real-time speed and vehicle driving time; the road supervision data includes road type, vehicle checkpoint passing time and road section speed limit information; the driver behavior data includes driver rest time, driver eye closing time, driver blinking frequency, driver heart rate, number of emergency brakes and line of sight deviation rate.

3. The accurate recognition algorithm for traffic characteristics of all vehicle types based on time-space fitting according to claim 2 is characterized in that: The method of preprocessing the vehicle dynamic data, the road supervision data and the driver behavior data to obtain the vehicle dynamic feature data set, the road supervision feature data set and the driver behavior feature data set comprises: Perform data cleaning on the acquired vehicle dynamic data, road supervision data and driver behavior data, and use linear interpolation to fill in missing values ​​in the acquired vehicle dynamic data, road supervision data and driver behavior data; The density clustering algorithm is used to detect outliers on the cleaned data and remove the identified outliers; the data after removing the outliers is standardized and normalized to obtain the preprocessed vehicle dynamic feature data set, road supervision feature data set and driver behavior feature data set.

4. The accurate recognition algorithm for traffic characteristics of all vehicle types based on time-space fitting according to claim 3 is characterized in that: The method for obtaining a preliminary vehicle driving trajectory according to a vehicle dynamic feature data set and a road supervision feature data set comprises: S41. Preset the real-time position coordinate sequence of the vehicle at each time point in the vehicle dynamic feature data set as: ;in, The real-time position coordinate sequence of the vehicle at each time point; For time point The coordinates of the vehicle trajectory points; is the horizontal coordinate of the vehicle trajectory point; is the ordinate of the vehicle trajectory point; is the index of the time point; The real-time speed sequence of the vehicle at each time point in the preset vehicle dynamic feature data set is: ;in, The real-time speed sequence of vehicles at each time point; For time point The real-time speed of the vehicle; The vehicle driving time series in the preset vehicle dynamic feature dataset is: ;in, is the vehicle travel time series; For the The time value of each sampling point; the vehicle dynamic feature data set is denoised by Wiener filtering; S42, preset vehicles at each time point The vehicle moves in a straight line at a uniform speed, and the current time point of the vehicle is deduced through the position update formula based on the vehicle dynamic feature data set. The initial driving trajectory of; the position update formula is: ;in, For the vehicle at time The coordinates of the trajectory points at time ; For at time point The heading angle of the vehicle at time ; is the time step, which indicates the interval between two sampling times; is the cosine value of the vehicle's heading angle; is the sine value of the vehicle's heading angle; S43, combining the road supervision feature data set, and correcting the current time point of the deduced vehicle through the trajectory correction formula The initial driving trajectory of ; the trajectory correction formula is: ;in, is the corrected vehicle at time point The coordinates of the trajectory points at time ; is the shortest distance vector from the vehicle trajectory point to the center line of the road; is the correction factor used to balance the correction strength of the trajectory deviation; The correction coefficient is dynamically adjusted through the correction coefficient adjustment formula; the correction coefficient adjustment formula is: ;in, is the correction coefficient after dynamic adjustment; is the weight factor for adjusting the correction coefficient; is the distance sensitivity coefficient; is the modulus of the shortest distance from the trajectory point to the center line of the road; The maximum speed allowed for vehicles on the road; is the smoothing factor; S44, matching each trajectory point to the nearest road center point using a trajectory point matching formula; the trajectory point matching formula is: ;in, For time point Time and vehicle trajectory points The nearest road center point to match; is the set of road centerline points; The center point of the road The coordinates of The vehicle trajectory point With road center The Euclidean distance between For the collection Choose a point in The value of is the smallest; S45. Calculate the cosine similarity of the vehicle speed vector and the road direction vector by using the cosine similarity formula, and correct the moving direction of the vehicle so that the moving direction of the vehicle is consistent with the road direction; the cosine similarity formula is: ;in, is the cosine similarity between the vehicle speed vector and the road direction vector, is the angle between the vehicle's moving direction and the road direction; for The vehicle velocity vector at position; for The road direction vector; for The modulus of the vehicle velocity vector at the time position; is the modulus of the road direction vector; like 1, then the vehicle's direction of movement is determined to be consistent with the road direction; if 1, it is determined that the vehicle's direction of movement deviates from the road direction and needs to be corrected; if 1, then it is determined that the vehicle's moving direction is opposite to the road direction and needs to be corrected; S46. Correcting the speed vector of the vehicle so that its moving direction is consistent with the road direction, thereby obtaining a final corrected preliminary driving trajectory of the vehicle.

5. The accurate recognition algorithm for traffic characteristics of all vehicle types based on time-space fitting according to claim 4 is characterized in that: The method of optimizing the preliminary driving trajectory of the vehicle by improving the spatiotemporal interpolation algorithm to obtain the complete driving trajectory of the vehicle includes: S51, the final corrected preliminary driving trajectory of the vehicle is expressed as: ;in, It is a set of all points of the final corrected preliminary driving trajectory of the vehicle; For time point The corrected trajectory point sequence, including coordinate information and sampling time; is the total number of time points of trajectory sampling, that is, the total number of trajectory points; is the sampling time point of the vehicle's initial driving trajectory points, ranging from arrive ; S52. Construct a spatiotemporal variogram to measure the correlation of trajectory points in space and time dimensions; the spatiotemporal variogram is: ;in, The spatiotemporal variogram is a measure of the correlation between trajectory points in spatial and temporal dimensions; is the spatial distance between trajectory points; From the current time To the next time point time difference; is the observation value of the trajectory point; For the next time point Observation values ​​of trajectory points; is the expected value of the trajectory point; S53, describing the spatial and temporal correlation between trajectory points by means of a covariance function; the covariance function is: ;in, is the covariance between trajectory points; is the variance of the trajectory point observations; is the spatial range parameter, controlling the spatial distance The degree of influence on the covariance; is the time range parameter, controlling the time difference The degree of influence on the covariance; S54. For a known time point Corrected trajectory point sequence , using the Kriging interpolation formula to calculate the time point Corresponding trajectory points; the Kriging interpolation formula is: ;in, For time point The corresponding trajectory points; is the observed value of the known trajectory point; is the weight coefficient for each known trajectory point; For the The position of each trajectory point is known. Contains spatial location and timestamp ; is the index of the known trajectory point; S55, dynamically adjust the weight coefficient of each known trajectory point through the weight coefficient adjustment formula; expand and smoothly optimize the trajectory points in the time dimension by gradually updating the trajectory points, and generate a arrive The optimized trajectory points are calculated and the time dimension is gradually expanded to obtain the complete driving trajectory of the vehicle: ;in, For time point Optimized trajectory point sequence; For time point The corresponding optimized horizontal axis; For time point The corresponding optimized ordinate; is the total number of trajectory points after optimization, and .

6. The accurate recognition algorithm for traffic characteristics of all vehicle types based on time-space fitting according to claim 5 is characterized in that: The method for dynamically adjusting the weight coefficient of each known trajectory point by means of a weight coefficient adjustment formula comprises: Construct a weight coefficient adjustment formula to adjust the weight coefficient of each known trajectory point; The weight coefficient adjustment formula is: ;in, is the weight coefficient of each known trajectory point after adjustment; is the spatial distance from the trajectory point The associated distance decay function; Indicates normalization of all known trajectory points; the distance decay function expression is: ;in, is the minimum distance constant; is the spatial distance between trajectory points The square of .

7. The accurate recognition algorithm for traffic characteristics of all vehicle types based on time-space fitting according to claim 6 is characterized in that: The method for calculating the vehicle speed of each road section according to the complete vehicle driving trajectory includes: By dividing the vehicle's complete driving trajectory into road sections, calculate the average speed of all vehicle trajectory points in each road section, and thus obtain the vehicle speed of the road section; If the vehicle speed of each road section is greater than the speed limit of the corresponding road section, the vehicle is judged to be speeding; if the vehicle speed of each road section is less than or equal to the speed limit of the corresponding road section, the vehicle is judged not to be speeding.

8. The accurate recognition algorithm for traffic characteristics of all vehicle types based on time-space fitting according to claim 7 is characterized in that: The training method of the fatigue driving risk prediction model includes: The dataset is divided into training set, validation set and test set to train the model, evaluate the model performance and verify the generalization ability of the model; the sample set is a subset of the dataset, each sample set includes the historical driver behavior feature dataset and the complete driving trajectory of the vehicle and the corresponding risk probability of fatigue driving; GBDT is selected as the specific implementation of the gradient boosting tree model to handle the regression task; Initialize GBDT parameters and define the objective function; GBDT parameters include the maximum number of leaf nodes in each tree, learning rate, number of trees and maximum depth of trees; use the historical driver behavior feature data set and the complete vehicle driving trajectory as input data, and the corresponding fatigue driving risk probability as output label to train the fatigue driving risk prediction model; the fatigue driving risk prediction model is a gradient boosting tree model; Use mean square error as the loss function to measure the difference between the model's predicted value and the actual value; in each iteration, GBDT builds a new decision tree to fit the negative gradient of the loss function of the previous step, uses the negative gradient of the objective function as the learning target of the new tree, and adjusts the contribution of the new tree to the final result through the learning rate to minimize the loss function; Use the validation set to evaluate the performance of the model, tune the model, and adjust the model parameters based on its performance feedback until the model performance no longer improves or reaches the preset stopping condition; use the trained fatigue driving risk prediction model to predict the current driver behavior feature dataset and the vehicle's complete driving trajectory to obtain the risk probability of fatigue driving.

9. The accurate recognition algorithm for traffic characteristics of all vehicle types based on time-space fitting according to claim 8 is characterized in that: The method for determining whether a driver is driving fatigued by predicting the risk probability of driving fatigued comprises: If the predicted risk probability of fatigue driving is less than the preset risk probability threshold of fatigue driving, it is determined that the driver is not driving fatigued; If the predicted risk probability of fatigue driving is greater than or equal to a preset risk probability threshold of fatigue driving, the driver is judged to be fatigue driving.

10. A system for accurately identifying traffic characteristics of all vehicle types based on time-space fitting, used to implement the accurate identification algorithm for traffic characteristics of all vehicle types based on time-space fitting as described in any one of claims 1 to 9, characterized in that: include: A data acquisition module, used to acquire vehicle dynamic data, road supervision data and driver behavior data; A data processing module is used to pre-process the vehicle dynamic data, the road supervision data and the driver behavior data to obtain a vehicle dynamic feature data set, a road supervision feature data set and a driver behavior feature data set; The vehicle trajectory acquisition module is used to obtain the preliminary driving trajectory of the vehicle based on the vehicle dynamic feature data set and the road supervision feature data set; the preliminary driving trajectory of the vehicle is optimized by improving the spatiotemporal interpolation algorithm to obtain the complete driving trajectory of the vehicle; The vehicle speeding assessment module is used to calculate the vehicle speed of each road section based on the complete driving trajectory of the vehicle, and compare it with the speed limit of the corresponding road section to determine whether the vehicle is speeding; A fatigue driving risk assessment module is used to obtain a fatigue driving risk prediction model based on the driver behavior feature data set and the vehicle's complete driving trajectory training, and predict the risk probability of fatigue driving based on the fatigue driving risk prediction model; Determine whether the driver is driving fatigued by predicting the risk probability of driving fatigued; Feedback reminder module: if the vehicle exceeds the speed limit, the on-board intelligent control terminal will automatically send out a speeding alarm to remind the driver to slow down in time; If the driver is judged to be driving fatigued, the on-board intelligent control terminal automatically generates a fatigue warning instruction to remind the driver that he is in a fatigue driving state; each module is connected by wired and / or wireless means.

Images