A driving safety management method and system based on driving track monitoring

Abstract

The embodiment of the present application relates to the technical field of driving safety management, and specifically discloses a driving safety management method and system based on driving track monitoring. The embodiment of the present application directly or cooperatively collects driving state data of a target engineering vehicle; generates smooth track data, performs abnormality evaluation on the smooth track data; plans an optimal driving path, and performs dynamic driving guidance; performs dynamic risk analysis and early warning on the target engineering vehicle; constructs a driving risk portrait, and performs periodic examination on a target driver of the target engineering vehicle. The track can be cleaned and reconstructed, smooth track data can be generated and abnormality evaluation can be performed, dynamic risk analysis and early warning can be performed on the target engineering vehicle, a driving risk portrait can be constructed, and periodic examination can be performed, so that personalized management of driving safety of engineering transport vehicles can be realized, management precision can be improved, driving data can be effectively utilized, and basic support can be provided for management decision guidance and strategy optimization.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of driving safety management, and particularly relates to a driving safety management method and system based on driving track monitoring. BACKGROUND

[0002] Driving safety management is a management activity that ensures the coordinated operation of people, vehicles, roads and other factors through institutionalized, systematic and technological means to monitor and control the entire process of the driving process of motor vehicles, so as to minimize the risk of traffic accidents and protect the efficiency of transportation and the safety of personnel life and property. The main content includes driver behavior management, vehicle state monitoring, road environment assessment, emergency warning and processing mechanism construction, etc.

[0003] Modern driving safety management widely applies information technology such as GPS positioning, track analysis, video monitoring, intelligent warning system, etc., to realize real-time monitoring and dynamic intervention on key indicators such as vehicle position, speed, running track and fatigue driving.

[0004] In the prior art, for the driving safety management of engineering transport vehicles, only whether the speed is exceeded is judged based on the speed limit rules of the road section, the specific behavior pattern and potential risk behavior of the driver are ignored, the personalized management of the driving safety of engineering transport vehicles cannot be realized, and there is lack of data analysis, archiving and feedback mechanism, the data utilization rate is low, and the management decision and strategy optimization cannot be guided. SUMMARY

[0005] The purpose of the embodiment of the application is to provide a driving safety management method and system based on driving track monitoring, which aims to solve the problems proposed in the background art.

[0006] To achieve the above-mentioned purpose, the technical scheme provided by the embodiment of the application is as follows:

[0007] A driving safety management method based on driving track monitoring, the method specifically comprises the following steps:

[0008] Determine a target engineering vehicle, and directly or cooperatively collect driving state data of the target engineering vehicle;

[0009] Track cleaning and reconstruction are performed on the driving state data to generate smooth track data, and abnormal evaluation is performed on the smooth track data to obtain abnormal evaluation data;

[0010] Real-time road condition data is received, an optimal driving path is planned in combination with the real-time road condition data and the driving state data, and dynamic driving guidance is performed according to the optimal driving path;

[0011] Dynamic risk analysis and early warning of the target engineering vehicle are performed according to the optimal driving path and the driving state data.

[0012] The driving state data is updated and recorded, a driving risk portrait is constructed, and a periodic examination is conducted on the target driver of the target engineering vehicle.

[0013] A driving safety management system based on driving track monitoring, which is applied to the driving safety management method based on driving track monitoring, comprises a driving state acquisition unit, a track abnormality evaluation unit, a dynamic driving guidance unit, a risk early warning prompting unit and a driving periodic examination unit, wherein:

[0014] The driving state acquisition unit is used for determining a target engineering vehicle and directly or cooperatively acquiring driving state data of the target engineering vehicle.

[0015] The track abnormality evaluation unit is used for performing track cleaning and reconstruction on the driving state data, generating smooth track data, performing abnormality evaluation on the smooth track data, and obtaining abnormality evaluation data.

[0016] The dynamic driving guidance unit is used for receiving real-time road condition data, planning an optimal driving path in combination with the real-time road condition data and the driving state data, and performing dynamic driving guidance according to the optimal driving path.

[0017] The risk early warning prompting unit is used for performing dynamic risk analysis and early warning prompting on the target engineering vehicle according to the optimal driving path and the driving state data.

[0018] The driving periodic examination unit is used for updating and recording the driving state data, constructing a driving risk portrait, and conducting a periodic examination on the target driver of the target engineering vehicle.

[0019] Compared with the prior art, the driving safety management system based on driving track monitoring has the following beneficial effects:

[0020] The driving safety management system based on driving track monitoring directly or cooperatively acquires driving state data of a target engineering vehicle, generates smooth track data, performs abnormality evaluation on the smooth track data, plans an optimal driving path, performs dynamic driving guidance, performs dynamic risk analysis and early warning prompting on the target engineering vehicle, constructs a driving risk portrait, and conducts a periodic examination on the target driver of the target engineering vehicle. The track cleaning and reconstruction can be performed, the smooth track data can be generated and abnormality evaluation can be performed, the dynamic risk analysis and early warning prompting can be performed on the target engineering vehicle, the driving risk portrait can be constructed, and the periodic examination can be conducted, so that the personalized management of the driving safety of engineering transport vehicles can be realized, the management precision can be improved, the driving data can be effectively utilized, and a basis support can be provided for management decision guidance and strategy optimization. BRIEF DESCRIPTION OF DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application.

[0022] Figure 1 A flow chart of the method provided by the embodiments of the present application is shown.

[0023] Figure 2 A structural block diagram of the driving cycle examination unit in the system provided by the embodiments of the present application is shown. DETAILED DESCRIPTION

[0024] In order to make the objects, technical solutions and advantages of the present application more clear, the following will further describe the present application in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application.

[0025] It can be understood that modern driving safety management widely applies information technology, such as GPS positioning, trajectory analysis, video monitoring, intelligent early warning system, etc., to realize real-time monitoring and dynamic intervention on key indicators such as vehicle position, speed, running trajectory, fatigue driving, etc. In the prior art, for driving safety management of engineering transport vehicles, only whether overspeeding is judged based on road section speed limit rules, the specific behavior mode and potential risk behavior of the driver are ignored, personalized management of driving safety of engineering transport vehicles cannot be realized, and there is lack of data analysis, archiving, feedback mechanism, data utilization rate is low, and management decision and strategy optimization cannot be guided.

[0026] To solve the above problems, the embodiments of the present application directly or cooperatively collect driving state data of the target engineering vehicle by determining the target engineering vehicle; the driving state data is trajectory cleaned and reconstructed to generate smooth trajectory data, and the smooth trajectory data is abnormally evaluated to obtain abnormal evaluation data; real-time road condition data is received, the optimal driving path is planned in combination with the real-time road condition data and the driving state data, and dynamic driving guidance is performed according to the optimal driving path; the target engineering vehicle is dynamically risk analyzed and early warned according to the optimal driving path and the driving state data; the driving state data is updated and recorded to construct a driving risk portrait, and the target driving personnel of the target engineering vehicle is periodically examined. The trajectory can be cleaned and reconstructed to generate smooth trajectory data and perform abnormal evaluation, the target engineering vehicle is dynamically risk analyzed and early warned, the driving risk portrait is constructed, and the periodic examination is performed, which can realize personalized management of driving safety of engineering transport vehicles, improve management accuracy, and effectively utilize driving data to provide a basis for management decision guidance and strategy optimization.

[0027] Figure 1A flow chart of the method provided by the embodiment of the application is shown.

[0028] Specifically, a driving safety management method based on driving track monitoring comprises the following steps:

[0029] In step S101, a target engineering vehicle is determined, and driving state data of the target engineering vehicle is directly or cooperatively collected.

[0030] In the embodiment of the application, by determining a target engineering vehicle for driving safety management, and acquiring transmission communication quality of the target engineering vehicle in real time, the transmission communication quality is compared with preset standard communication quality. When the transmission communication quality is greater than the standard communication quality, driving state data (including current position, current speed, current acceleration, vehicle heading, and vehicle running state, etc.) of the target engineering vehicle is directly collected. When the transmission communication quality is not greater than the standard communication quality, a cooperative engineering vehicle is selected, and a temporary communication connection between the target engineering vehicle and the cooperative engineering vehicle is constructed. Through the temporary communication connection between the target engineering vehicle and the cooperative engineering vehicle, the driving state data of the target engineering vehicle is cooperatively collected by taking the cooperative engineering vehicle as an intermediate medium.

[0031] Specifically, in the preferred embodiment provided by the application, the determination of the target engineering vehicle and the direct or cooperative collection of the driving state data of the target engineering vehicle specifically comprises the following steps:

[0032] The target engineering vehicle for driving safety management is determined;

[0033] The transmission communication quality of the target engineering vehicle is acquired, and the transmission communication quality is compared with the preset standard communication quality;

[0034] When the transmission communication quality is greater than the standard communication quality, the driving state data of the target engineering vehicle is directly collected;

[0035] When the transmission communication quality is not greater than the standard communication quality, a cooperative engineering vehicle is selected, and a temporary communication connection between the target engineering vehicle and the cooperative engineering vehicle is constructed;

[0036] Through the cooperative engineering vehicle, the driving state data of the target engineering vehicle is cooperatively collected.

[0037] Specifically, in the preferred embodiment provided by the application, when the transmission communication quality is not greater than the standard communication quality, the cooperative engineering vehicle is selected, and the temporary communication connection between the target engineering vehicle and the cooperative engineering vehicle is constructed, which specifically comprises the following steps:

[0038] Three-dimensional axial acceleration data of the target engineering vehicle are acquired in real time, and vibration frequency spectrum original data of all candidate engineering vehicles within a preset range are synchronously collected.

[0039] Performing fast Fourier transform on acceleration data of the target engineering vehicle and the candidate vehicle, extracting energy distribution spectrum in the frequency range of 0.1-100Hz, obtaining the main resonance frequency point of the target engineering vehicle and the energy stability index of each axis of the candidate vehicle;

[0040] Judging the main resonance frequency point of the target engineering vehicle, if the main resonance frequency point of the target engineering vehicle falls into the communication frequency band (1.5-2.5GHz) and the energy exceeds the threshold (such as 80dB), it is marked as "loose device risk", and a risk label is obtained;

[0041] According to the risk label, in the high-risk scenario, a candidate vehicle with a resonance axis orthogonal to the target engineering vehicle and an energy stability index > 90 is selected for cooperation; in the low-risk scenario, the candidate vehicle with the highest signal strength is selected for cooperation; the cooperative engineering vehicle ID and the complementary axis identification are obtained;

[0042] Obtaining the BIM model of the construction site within a preset range around the target engineering vehicle, and identifying the metal structure in the BIM model of the construction site within the preset range around the target engineering vehicle which can be used as a reflection source, obtaining the position coordinates of the metal structure;

[0043] Modulating the driving state data of the target engineering vehicle to the environmental radio frequency carrier, irradiating the metal structure through the directional antenna according to the position coordinates of the metal structure, and obtaining the phase jump of the reflected signal;

[0044] According to the cooperative engineering vehicle ID, the cooperative engineering vehicle is selected, and the coherent detector is deployed on the cooperative engineering vehicle; according to the complementary axis identification, the coherent detector of the complementary cooperative engineering vehicle in the corresponding axis is enabled to demodulate the phase jump and restore the data, so as to establish a bidirectional low-rate link.

[0045] In the embodiment of the application, the engineering vehicle generates low-frequency mechanical vibration (such as engine shaking and road bumping) when working, when the vibration frequency approaches the inherent resonance frequency point (such as the antenna PCB board resonance frequency 1.5-2.5GHz) of the communication device, it will cause the antenna impedance mismatch, the signal reflection enhancement, and the communication error rate will soar. The traditional scheme only monitors the electromagnetic signal strength (RSSI), and cannot identify such "invisible" interference. The application selects the cooperative vehicle with stable vibration in the Y axis (energy stability index > 90) when the target vehicle vibrates violently in the X axis direction (resonance frequency point energy > 80dB), because the mechanical vibration in the orthogonal axis does not interfere with each other, the resonance energy can be avoided. It is equivalent to building a "mechanical vibration isolation zone" for the communication system. And based on the temporary communication principle of environmental backscattering.

[0046] The electromagnetic wave reflection characteristics of the metal structure (such as steel bars) at the construction site are utilized, and the target vehicle data is modulated to the ambient radio frequency signal (such as the 900MHz carrier leaked by the nearby base station) through binary phase shift keying (BPSK).

[0047] The data "0" keeps the phase of the reflected signal at 0° (original phase), and the data "1" offsets the phase of the reflected signal by 180° (inverted phase). Then, the cooperative vehicle uses a coherent detector to demodulate the data by comparing the phase difference between the reflected signal and the reference carrier. Since no active radio frequency signal is transmitted, only the ambient electromagnetic wave is relied on, realizing "zero radiation" communication.

[0048] Since the power consumption of the ambient backscatter communication is only 1 / 1000 of the traditional radio frequency transmission (the measured target vehicle power consumption is <1mW), the battery endurance bottleneck of the engineering vehicle is solved. Moreover, the reflected signal power is lower than the ambient noise floor (-110dBm), which cannot be detected by conventional equipment, and the phase modulation signal needs to be accurately synchronized with the carrier to be decoded, and non-cooperative parties cannot crack it, realizing anti-eavesdropping and anti-tampering, and enhancing the safety performance.

[0049] Further, the driving safety management method based on the driving track monitoring further comprises the following steps:

[0050] In step S102, the driving state data is trajectory cleaned and reconstructed to generate smooth trajectory data, and the smooth trajectory data is abnormally evaluated to obtain abnormal evaluation data.

[0051] In the embodiment of the present application, Kalman filtering and / or Savitzky-Golay algorithm is used to clean the driving state data trajectory to obtain cleaned state data, and then the cleaned state data is trajectory smoothed to reconstruct the smooth trajectory data of the target engineering vehicle, and the smooth trajectory data is abnormally evaluated according to multiple abnormal driving states (abnormal evaluation is a process of evaluating abnormal states of the target engineering vehicle such as sudden acceleration, sudden deceleration, abnormal parking and illegal detour), and abnormal evaluation data is obtained.

[0052] Specifically, in the preferred embodiment provided by the present application, the trajectory cleaning and reconstruction of the driving state data to generate smooth trajectory data and the abnormal evaluation of the smooth trajectory data to obtain abnormal evaluation data specifically comprises the following steps:

[0053] The driving state data is trajectory cleaned by using Kalman filtering and / or Savitzky-Golay algorithm to obtain cleaned state data;

[0054] The cleaned state data is trajectory smoothed and reconstructed to generate smooth trajectory data;

[0055] According to the plurality of abnormal driving states, the smooth trajectory data is abnormally evaluated to obtain abnormal evaluation data.

[0056] In the preferred embodiments provided in the application, the trajectory cleaning of the driving state data by using the Kalman filtering and the Savitzky-Golay algorithm to obtain the cleaned state data specifically includes the following steps:

[0057] The driving state data is subjected to trajectory cleaning by using the Kalman filtering and the Savitzky-Golay algorithm respectively to obtain the Kalman filtering result and the Savitzky-Golay filtering result;

[0058] The acceleration threshold value is confirmed through historical data statistics and vehicle types, and the absolute value of the acceleration at the current time is obtained according to the driving state data;

[0059] The absolute value of the acceleration at the current time is compared with the acceleration threshold value, if the absolute value of the acceleration at the current time is less than the acceleration threshold value, it is determined that the driving state is a stable state, the weight is close to 0, and the Savitzky-Golay result is preferentially used, if the absolute value of the acceleration at the current time is greater than or equal to the acceleration threshold value, it is determined that the driving state is a severe motion state, the weight is close to 1, and the Kalman filtering result is preferentially used;

[0060] According to the current driving state, different weights are given to the Kalman filtering result and the Savitzky-Golay filtering result to obtain the acceleration dynamic weight value at the current time, if it is determined that the driving state is a stable state, the weight is close to 0, and the Savitzky-Golay result is preferentially used, if it is determined that the driving state is a severe motion state, the weight is close to 1, and the Kalman filtering result is preferentially used;

[0061] The acceleration dynamic weight value at the current time is applied to the sigmoid function for smooth transition to obtain the smoothed acceleration dynamic weight value;

[0062] The Kalman filtering result and the Savitzky-Golay filtering result are subjected to linear fusion by using the smoothed acceleration dynamic weight value to obtain the filtering data at the current time;

[0063] The filtering data at each time point is obtained in time sequence to obtain the cleaned state data.

[0064] In the embodiments of the present application, the Kalman filtering result suppresses the trajectory data after mutation noise, but may introduce lag error (for example, position update delay when emergency braking). The Savitzky-Golay filtering result retains the smooth trajectory data with details, but has weak noise suppression ability for sharp changes. The present application uses an acceleration-driven dynamic fusion mechanism to dynamically adaptively switch or fuse according to the acceleration of the current vehicle, combines the advantages of the two algorithms, overcomes the limitations of a single filtering algorithm, avoids the lag defect of Kalman filtering, and makes up for the noise resistance of Savitzky-Golay filtering, so as to automatically switch the optimal strategy when the motion state changes. Moreover, the sigmoid function is used to smooth the switching weight, so as to avoid trajectory jump caused by sudden change of filtering strategy.

[0065] For example: smooth working condition (such as uniform speed driving): the absolute value of acceleration is low, at this time, Savitzky-Golay filtering (SG) is dominant, and the local polynomial fitting characteristic is used to retain the trajectory details (such as slow turning curvature change).

[0066] Severe working condition (such as emergency braking, bumpy road condition): the absolute value of acceleration exceeds the threshold value, Kalman filtering (KF) is dominant, and the motion model prediction and measurement value fusion are used to suppress mutation noise (such as GPS signal jump).

[0067] In the preferred embodiments provided by the present application, the abnormal evaluation of the smooth trajectory data according to a plurality of abnormal driving states to obtain abnormal evaluation data specifically includes the following steps:

[0068] Different abnormal driving states are set, and a corresponding reference threshold value is given for each abnormal driving state;

[0069] The data corresponding to the abnormal driving state in the smooth trajectory data is compared with the reference threshold value, and the comparison result is normalized to obtain the basic score of each abnormal type;

[0070] The acceleration vector of each time point is obtained from the smooth trajectory data, and the trajectory dynamic change rate is calculated according to the acceleration vector of each time point;

[0071] The corresponding sensitive coefficient is set for different abnormal driving states, the sensitive coefficient and the trajectory dynamic change rate are weighted and exponentially calculated to obtain the exponential calculation result of different abnormal driving states;

[0072] The exponential calculation results of different abnormal driving states are normalized to obtain the dynamic weight of different abnormal driving states;

[0073] The base scores of each abnormal type are weighted by dynamic weights of different abnormal driving states to obtain final scores of different abnormal driving states, and the final scores of different abnormal driving states are summed to obtain a comprehensive abnormal score.

[0074] The final scores of different abnormal driving states and the comprehensive abnormal score are packaged and associated with corresponding time stamps, vehicle identifiers and metadata to obtain abnormal evaluation data.

[0075] In the embodiment of the application, the corresponding abnormal driving state data in the smoothed trajectory data is compared with the reference threshold, and the corresponding score is obtained according to the comparison result. The trajectory dynamic change rate is used to quantitatively reflect the trajectory fluctuation intensity (such as sudden braking and bumpy road conditions) in real time. The exponential normalization function is used to dynamically assign the weight of each abnormal type based on the trajectory dynamic change rate. When the trajectory dynamic change rate is high (the trajectory is unstable), the weight of the change-sensitive abnormal type (such as sudden turning) is increased. In addition, a multi-dimensional abnormal fusion evaluation mechanism is designed. The base score of each abnormal type (such as sudden acceleration and sudden turning) is calculated independently, and then the dynamic weight is fused into a comprehensive score to reflect the overall risk level.

[0076] Further, the driving safety management method based on driving trajectory monitoring further comprises the following steps:

[0077] Step S103, receiving real-time road condition data, planning an optimal driving path in combination with the real-time road condition data and the driving state data, and performing dynamic driving guidance according to the optimal driving path.

[0078] In the embodiment of the application, real-time road condition data including traffic maps, construction area restrictions, high-speed limit lines, etc. is received, an optimal driving path is planned in combination with the real-time road condition data and the driving state data, dynamic guidance information is generated according to the optimal driving path, and the dynamic guidance information is broadcasted to guide the target engineering vehicle to dynamically drive.

[0079] Specifically, in the preferred embodiment provided by the application, the receiving real-time road condition data, planning an optimal driving path in combination with the real-time road condition data and the driving state data, and performing dynamic driving guidance according to the optimal driving path specifically comprises the following steps:

[0080] Receiving real-time road condition data;

[0081] Planning an optimal driving path in combination with the real-time road condition data and the driving state data;

[0082] Generating dynamic guidance information according to the optimal driving path;

[0083] Broadcast the dynamic guidance information to guide the target engineering vehicle to travel dynamically.

[0084] In the preferred embodiments provided by the application, the combination of the real-time road condition data and the driving state data to plan the optimal travel path specifically includes the following steps:

[0085] The congestion coefficient is calculated according to the real-time road condition data, and the congestion coefficient is smoothed by using an S-shaped function to obtain a time weight;

[0086] The safety weight is set according to the time weight, and the safety weight and the time weight are set to be complementary;

[0087] The energy consumption weight is obtained by calculating the proportion of the remaining fuel quantity in the total fuel tank according to the driving state data;

[0088] Based on the vehicle position and the target location, a plurality of paths are generated by calling the road network topology data;

[0089] The minimum turning radius and the maximum climbing degree of the vehicle are obtained, and the plurality of paths are filtered according to the minimum turning radius and the maximum climbing degree of the vehicle to obtain a plurality of feasible paths;

[0090] The congestion coefficients of the road sections in the plurality of feasible paths are counted, and the expected travel time is calculated by accumulation to obtain a time cost;

[0091] The number of construction closed road sections in the path is counted to obtain a safety cost;

[0092] The length of the feasible path, the fuel consumption efficiency of the vehicle and the slope data are obtained, and the total fuel consumption is calculated according to the length of the feasible path, the fuel consumption efficiency of the vehicle and the slope data to obtain an energy consumption cost;

[0093] The time cost, the safety cost and the energy consumption cost are respectively weighted and fused by using the time weight, the safety weight and the energy consumption weight to obtain a weighted total cost value of each feasible path, and the feasible path with the minimum weighted total cost value is taken as the optimal travel path.

[0094] In the embodiments of the application, the real-time road condition and the vehicle state and the heterogeneous data are integrated, a balanced solution is found in the conflicting targets of time, safety and energy consumption, and different dimensions of information are quantified into a unified cost value by means of dynamic weight, realizing adaptive mapping of "scene perception-strategy generation", so that the weight is adjusted in real time with the road condition and the fuel state, solving the rigid problem of fixed weight strategy.

[0095] Further, the driving safety management method based on the driving trajectory monitoring further includes the following steps:

[0096] Step S104, performing dynamic risk analysis and early warning prompt on the target engineering vehicle according to the optimal driving path and the driving state data.

[0097] In the embodiments of the present application, over-speed judgment is performed according to the optimal driving path and the driving state data, and when over-speed exists, grade over-speed identification (including severe over-speed, moderate over-speed and slight over-speed, etc.) and early warning are performed, and road condition risk judgment is performed according to the optimal driving path and the driving state data, and when road condition risk exists, risk early warning information is generated, and road condition risk prompt (including reducing speed limit in advance when the vehicle is about to enter a sharp curve or a downhill road section, and risk reminding when approaching a construction area or a mixed area of vehicles and pedestrians, etc.) on the target engineering vehicle is performed by broadcasting the risk early warning information.

[0098] Specifically, in the preferred embodiments provided by the present application, the dynamic risk analysis and early warning prompt on the target engineering vehicle according to the optimal driving path and the driving state data specifically include the following steps:

[0099] Performing over-speed judgment according to the optimal driving path and the driving state data;

[0100] When over-speed exists, performing grade over-speed identification and early warning;

[0101] Performing road condition risk judgment according to the optimal driving path and the driving state data;

[0102] When road condition risk exists, generating risk early warning information;

[0103] Broadcasting the risk early warning information to perform road condition risk prompt on the target engineering vehicle.

[0104] In the preferred embodiments provided by the present application, the road condition risk judgment according to the optimal driving path and the driving state data specifically includes the following steps:

[0105] Obtaining path point curvature and path point slope of the optimal driving path;

[0106] Calculating road bending risk according to the path point curvature;

[0107] Calculating slope risk according to the path point slope;

[0108] Adding the road bending risk and the slope risk as a static risk value of the current path point to obtain a static risk value of each path point;

[0109] Obtaining vehicle real-time speed and load, and calculating a speed modulation factor according to the vehicle real-time speed and a current road section safety speed limit;

[0110] The load and slope are combined to calculate the risk of coasting, and a load-slope coupling factor is obtained;

[0111] The speed modulation factor and the load-slope factor are added to obtain a dynamic modulation coefficient of the current point;

[0112] The static risk value of each path point is fused with the corresponding dynamic modulation coefficient to obtain a comprehensive risk value of each path point, and road condition risk judgment is performed according to the comprehensive risk value of each path point.

[0113] In the embodiment of the application, the path geometry attribute (curvature, slope) is converted into a physical field strength, and the greater the curvature or the steeper the slope, the stronger the field strength, so as to realize static risk field construction. The vehicle state (speed, load) is used as a “disturbance source”, and the field strength is amplified when the speed exceeds the limit, and a nonlinear gain is triggered when the load and the slope are coupled (such as risk exponential growth when full load climbing). The static field and the dynamic disturbance are superimposed to generate a comprehensive risk field to judge the road condition risk.

[0114] Further, the driving safety management method based on driving trajectory monitoring further comprises the following steps:

[0115] Step S105, updating the driving state data, constructing a driving risk portrait, and periodically examining the target driver of the target engineering vehicle.

[0116] In the embodiment of the application, the driving state data is updated and recorded to obtain the driving record data of the target engineering vehicle, and then a machine learning algorithm is used to identify the driving record data, construct a driving risk portrait, determine a driving examination period, and then periodically examine the target driver of the target engineering vehicle according to the driving examination period and the driving risk portrait (generate multi-dimensional risk labels such as sudden acceleration preference, frequent yaw, and night speeding tendency).

[0117] Specifically, in the preferred embodiment provided by the application, the updating of the driving state data, the construction of the driving risk portrait, and the periodic examination of the target driver of the target engineering vehicle specifically comprises the following steps:

[0118] The driving state data is updated and recorded to obtain the driving record data of the target engineering vehicle;

[0119] The driving record data is identified by using a machine learning algorithm to construct a driving risk portrait;

[0120] A driving examination period is determined;

[0121] The target driver of the target engineering vehicle is periodically examined according to the driving examination period and the driving risk portrait.

[0122] Specifically, in the preferred embodiments provided by the present application, the identification of the driving record data by using the machine learning algorithm to construct the driving risk portrait specifically includes the following steps:

[0123] Extracting spatio-temporal trajectory features from driving record data, such as time series feature data of sudden acceleration frequency, sudden turning times, and night driving time length proportion; obtaining vehicle real-time load time series data, such as empty / semi-loaded / full-loaded state; generating an environmental risk heat map using the comprehensive risk value of each path point;

[0124] Extracting behavior fluctuation index from spatio-temporal trajectory features to obtain behavior indicators; identifying load transition events from load data to obtain load state; converting the environmental risk heat map into a regional risk level to obtain a risk area label;

[0125] Constructing a double-channel spatio-temporal convolutional neural network, wherein the first channel is sequentially provided with 3 layers of time series convolution, a spatial attention layer, and a decoding layer; the second channel is sequentially provided with 2 layers of one-dimensional convolution, a transition detection layer, and a decoding layer;

[0126] Inputting the behavior fluctuation index into the first channel for behavior feature analysis to obtain the time positioning and deviation value of the behavior abnormal event;

[0127] Inputting the load state into the second channel for load dynamic analysis to obtain the load mutation time point and state transition intensity;

[0128] When the first channel detects the time stamp of the behavior abnormality, immediately activate the same time stamp load state query of the second channel, multiply the behavior deviation value of the first channel with the load ratio (current load / capacity load) of the second channel to generate a load influence factor;

[0129] Adjusting the load influence factor through cross-channel attention gate to obtain a cross-channel fusion weight;

[0130] Aligning the high-order features output by the 3 layers of time series convolution of the first channel and the high-order features output by the 2 layers of one-dimensional convolution of the second channel in dimension, and then weighting and fusing them using the cross-channel fusion weight to obtain a fused spatio-temporal feature tensor; passing the fused spatio-temporal feature tensor through two fully connected layers to obtain a load-behavior coupling coefficient sequence;

[0131] Creating a grid unit of a preset size centered on the vehicle position, mapping the environmental risk heat map onto the grid unit to obtain a position-bound behavior-environment risk mapping table;

[0132] Adjusting the comprehensive risk value of each path point corresponding to each event in the behavior-environment risk mapping table using the load-behavior coupling coefficient sequence to obtain a weighted risk behavior event list;

[0133] The event classification and aggregation statistics of the weighted risk behavior event list are performed to obtain a driving risk portrait of different category dimensions. Different category dimensions include behavior preference obtained by statistically weighting the frequency of each event type, environment sensitivity obtained by calculating the proportion of events in a high-risk environment, and load coupling obtained by comparing the behavior frequency difference under different load states.

[0134] In the embodiment of the present application, the three-way interaction of "behavior-load-environment" is realized through multi-source information fusion, the single-dimensional risk detection is upgraded to multi-dimensional cooperative judgment (behavior abnormality itself may not be high-risk, but the hidden danger is greatly improved by superimposing high-risk load and high-risk environment). And the time sequence convolution is used to capture long and short period behavior patterns, the spatial attention is focused on the key behavior area, the one-dimensional convolution is specially used to detect the mutation point, and the transition detection layer is used to accurately lock the load change, thereby improving the event detection accuracy. And the behavior abnormality and the load state are aligned at the time point, the physical attribution of the behavior risk is realized (such as "full load sharp turn" risk is much higher than "empty load sharp turn"). A dynamic gating weight mechanism is also designed, which can flexibly adjust the "risk contribution degree" according to the specific scene and time through cross-channel attention gating, thereby avoiding the generalization ability deficiency caused by static weight. The environment risk heat map is used as a spatial context to further weight the behavior event, thereby realizing the dynamic risk adjustment of "the same behavior in different environment risks is different". Finally, the risk portrait is output in multiple dimensions, the behavior preference (such as high incidence of sudden acceleration), the environment sensitivity (such as the proportion of events in a high-risk area), and the load coupling (heavy load behavior change) are subdivided, thereby providing a data basis for the accurate portrait of the driver risk.

[0135] Further, Figure 2 The application architecture diagram of the system provided by the embodiment of the present application is shown.

[0136] In another preferred embodiment provided by the present application, a driving safety management system based on driving trajectory monitoring is applied to the driving safety management method based on driving trajectory monitoring, and the system comprises a driving state acquisition unit 101, a trajectory anomaly evaluation unit 102, a dynamic driving guidance unit 103, a risk warning prompt unit 104, and a driving cycle evaluation unit 105, wherein:

[0137] The driving state acquisition unit is used to determine a target engineering vehicle and directly or cooperatively acquire driving state data of the target engineering vehicle.

[0138] The trajectory anomaly evaluation unit is used to perform trajectory cleaning and reconstruction on the driving state data, generate smooth trajectory data, perform anomaly evaluation on the smooth trajectory data, and obtain anomaly evaluation data.

[0139] The dynamic driving guidance unit is configured to receive real-time road condition data, plan an optimal driving path based on the real-time road condition data and the driving state data, and guide the target engineering vehicle to drive according to the optimal driving path.

[0140] The risk early warning unit is configured to perform dynamic risk analysis and early warning for the target engineering vehicle based on the optimal driving path and the driving state data.

[0141] The driving cycle assessment unit is configured to update the driving state data, build a driving risk profile, and perform cycle assessment on a target driver of the target engineering vehicle.

[0142] It should be understood that although each step in the flowchart of each embodiment of the present application is displayed in sequence according to the arrow, these steps are not necessarily executed in the order indicated by the arrow. Unless otherwise specified herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other orders. Moreover, at least part of the steps in each embodiment can include multiple sub-steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution order of these sub-steps or stages is not necessarily sequential, but can be executed in rotation or alternation with at least part of other steps or sub-steps or stages of other steps.

[0143] It can be understood by those skilled in the art that all or part of the processes in the above-mentioned embodiment methods can be completed by a computer program instructing related hardware, and the program can be stored in a non-volatile computer readable storage medium. When the program is executed, it can include the processes of the above-mentioned embodiments of each method. Any reference to memory, storage, database or other medium used in each embodiment provided by the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0144] Any technical features in the above-described embodiments can be combined in any manner, and for the sake of brevity, not all possible combinations are described, however, as long as there is no contradiction, any technical features of the above-described embodiments should be considered to be within the scope of the present disclosure.

[0145] The above-described embodiments only express several embodiments of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the patent of the present application. It should be pointed out that for ordinary skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of protection of the present application. Therefore, the scope of protection of the patent of the present application should be subject to the appended claims.

[0146] The above-described only the preferred embodiments of the present application, and not to limit the present application, any modification, equivalent replacement and improvement within the spirit and principles of the present application, etc., should be included within the scope of protection of the present application.

Claims

1. A driving safety management method based on driving track monitoring, characterized in that, The method specifically comprises the following steps: determining a target engineering vehicle, directly or cooperatively collecting driving state data of the target engineering vehicle; performing trajectory cleaning and reconstruction on the driving state data to generate smooth trajectory data, and performing abnormality evaluation on the smooth trajectory data to obtain abnormality evaluation data; receiving real-time road condition data, planning an optimal driving path in combination with the real-time road condition data and the driving state data, and performing dynamic driving guidance according to the optimal driving path; performing dynamic risk analysis and early warning on the target engineering vehicle according to the optimal driving path and the driving state data; updating the driving state data, constructing a driving risk portrait, and performing periodic examination on a target driver of the target engineering vehicle; The determination of the target engineering vehicle and the direct or cooperative collection of the driving state data of the target engineering vehicle specifically comprises the following steps: determining a target engineering vehicle for driving safety management; obtaining the transmission communication quality of the target engineering vehicle and comparing the transmission communication quality with a preset standard communication quality; when the transmission communication quality is greater than the standard communication quality, directly collecting the driving state data of the target engineering vehicle; when the transmission communication quality is not greater than the standard communication quality, selecting a cooperative engineering vehicle and constructing a temporary communication connection between the target engineering vehicle and the cooperative engineering vehicle; cooperatively collecting the driving state data of the target engineering vehicle through the cooperative engineering vehicle; The trajectory cleaning and reconstruction on the driving state data to generate smooth trajectory data and the abnormality evaluation on the smooth trajectory data to obtain abnormality evaluation data specifically comprises the following steps: cleaning the trajectory of the driving state data by using Kalman filtering and / or the Savitzky-Golay algorithm to obtain cleaned state data; performing trajectory smoothing reconstruction on the cleaned state data to generate smooth trajectory data; performing abnormality evaluation on the smooth trajectory data according to a plurality of abnormal driving states to obtain abnormality evaluation data; The cleaning of the trajectory of the driving state data by using Kalman filtering and the Savitzky-Golay algorithm to obtain cleaned state data specifically comprises the following steps: cleaning the trajectory of the driving state data by using Kalman filtering and the Savitzky-Golay algorithm to obtain Kalman filtering results and Savitzky-Golay filtering results; confirming an acceleration threshold value through historical data statistics and vehicle types, and obtaining the absolute value of the acceleration at the current time according to the driving state data; comparing the absolute value of the acceleration at the current time with the acceleration threshold value, and determining that the driving state is a stable state if the absolute value of the acceleration at the current time is less than the acceleration threshold value, or determining that the driving state is a violent motion state if the absolute value of the acceleration at the current time is greater than or equal to the acceleration threshold value; assigning different weights to the Kalman filtering results and the Savitzky-Golay filtering results according to the current driving state to obtain the dynamic weight value of the acceleration at the current time; The acceleration dynamic weight value of the current time is applied to a sigmoid function for smooth transition to obtain a smoothed acceleration dynamic weight value; The Kalman filtering result and the Savitzky-Golay filtering result are linearly fused by using the smoothed acceleration dynamic weight value to obtain the filtering data at the current time; The filtering data at each time point is obtained in time sequence to obtain the cleaning state data.

2. The driving safety management method based on driving track monitoring according to claim 1, characterized in that, The abnormal evaluation data is obtained by performing abnormal evaluation on the smoothed trajectory data according to the multiple abnormal driving states, and specifically includes the following steps: Different abnormal driving states are set, and a corresponding reference threshold is given for each abnormal driving state; The data corresponding to the abnormal driving state in the smoothed trajectory data is compared with the reference threshold, and the comparison result is normalized to obtain the basic score of each abnormal type; The acceleration vector at each time point is obtained from the smoothed trajectory data, and the trajectory dynamic change rate is calculated according to the acceleration vector at each time point; The corresponding sensitive coefficient is set for different abnormal driving states, the sensitive coefficient and the trajectory dynamic change rate are weighted and exponentially calculated to obtain the exponential calculation result of different abnormal driving states; The exponential calculation results of different abnormal driving states are normalized to obtain the dynamic weight of different abnormal driving states; The basic scores of each abnormal type are weighted by using the dynamic weight of different abnormal driving states to obtain the final score of different abnormal driving states, and the final scores of different abnormal driving states are summed to obtain the comprehensive abnormal score; The final scores of different abnormal driving states and the comprehensive abnormal score are packaged, and the corresponding time stamp, vehicle identifier and metadata are associated to obtain the abnormal evaluation data.

3. The driving safety management method based on driving track monitoring according to claim 2, characterized in that, The real-time road condition data is received, the optimal driving path is planned in combination with the real-time road condition data and the driving state data, and the dynamic driving guidance is performed according to the optimal driving path, which specifically includes the following steps: The real-time road condition data is received; The optimal driving path is planned in combination with the real-time road condition data and the driving state data; The dynamic guidance information is generated according to the optimal driving path; The dynamic guidance information is broadcasted to guide the dynamic driving of the target engineering vehicle.

4. The driving safety management method based on driving track monitoring according to claim 3, characterized in that, The optimal driving path is planned in combination with the real-time road condition data and the driving state data, which specifically includes the following steps: The congestion coefficient is calculated according to the real-time road condition data, and the congestion coefficient is smoothed by using an S-shaped function to obtain a time weight; The safety weight is set according to the time weight, and the safety weight and the time weight are set as complementary; The proportion of the remaining fuel quantity to the total fuel tank capacity is calculated according to the driving state data to obtain an energy consumption weight; Based on the vehicle position and the target location, the road network topology data is called to generate multiple paths; The minimum turning radius and the maximum climbing degree of the vehicle are obtained, and the multiple paths are filtered according to the minimum turning radius and the maximum climbing degree of the vehicle to obtain multiple feasible paths; The congestion coefficients of the road sections in the multiple feasible paths are counted, and the expected travel time is calculated by accumulation to obtain the time cost; The number of construction closed road sections in the path is counted to obtain the safety cost; Obtain the feasible path length, vehicle fuel consumption efficiency and slope data, and calculate the total fuel consumption according to the feasible path length, vehicle fuel consumption efficiency and slope data to obtain the energy consumption cost; The time cost, safety cost and energy consumption cost are respectively weighted and fused by using time weight, safety weight and energy consumption weight to obtain a weighted total cost value of each feasible path, and the feasible path with the minimum weighted total cost value is taken as the optimal driving path.

5. The driving safety management method based on driving track monitoring according to claim 4, characterized in that, The dynamic risk analysis and early warning prompt of the target engineering vehicle according to the optimal driving path and the driving state data specifically includes the following steps: Perform overspeed judgment according to the optimal driving path and the driving state data; When there is overspeed, perform grade overspeed identification and warning; Perform road condition risk judgment according to the optimal driving path and the driving state data; When there is road condition risk, generate risk warning information; Broadcast the risk warning information to prompt the target engineering vehicle of the road condition risk.

6. The driving safety management method based on driving track monitoring according to claim 5, characterized in that, The road condition risk judgment according to the optimal driving path and the driving state data specifically includes the following steps: Obtain the path point curvature and path point slope of the optimal driving path; According to the path point curvature, the road bending risk is calculated; According to the path point slope, the slope risk is calculated; Add the road bending risk and the slope risk as the static risk value of the current path point to obtain the static risk value of each path point; Obtain the vehicle real-time speed and load, and calculate the speed modulation factor according to the vehicle real-time speed and the current road section safety speed limit; Combine the load and the slope to calculate the coasting risk to obtain the load-slope coupling factor; Add the speed modulation factor and the load-slope factor as the dynamic modulation coefficient of the current point; Fuse the static risk value of each path point and the corresponding dynamic modulation coefficient to obtain the comprehensive risk value of each path point, and perform road condition risk judgment according to the comprehensive risk value of each path point.

7. The driving safety management method based on driving track monitoring according to claim 6, characterized in that, The updating record of the driving state data, the construction of the driving risk portrait, and the periodic examination of the target driving personnel of the target engineering vehicle specifically include the following steps: Update the driving record data of the target engineering vehicle by updating the driving state data; Use a machine learning algorithm to identify the driving record data to construct a driving risk portrait; Determine the driving examination period; According to the driving examination period and the driving risk portrait, the target driving personnel of the target engineering vehicle are periodically examined.

8. A driving safety management system based on driving track monitoring, the system being applied to the driving safety management method based on driving track monitoring according to any one of claims 1 to 7, characterized in that, The system includes a driving state acquisition unit, a trajectory anomaly evaluation unit, a dynamic driving guidance unit, a risk warning prompt unit and a driving periodic examination unit, wherein: The driving state acquisition unit is used to determine the target engineering vehicle, and directly or cooperatively acquire the driving state data of the target engineering vehicle; The trajectory anomaly evaluation unit is used to perform trajectory cleaning and reconstruction on the driving state data to generate smooth trajectory data, and perform anomaly evaluation on the smooth trajectory data to obtain anomaly evaluation data; The dynamic driving guidance unit is configured to receive real-time road condition data, plan an optimal driving path in combination of the real-time road condition data and the driving state data, and perform dynamic driving guidance according to the optimal driving path. The risk early warning prompting unit is configured to perform dynamic risk analysis and early warning prompting on the target engineering vehicle according to the optimal driving path and the driving state data. The driving cycle assessment unit is configured to update and record the driving state data, build a driving risk portrait, and perform cycle assessment on a target driver of the target engineering vehicle.

Images