Vehicle fault positioning and rescue scheduling platform and method based on GPS track acquisition

Abstract

The invention discloses a vehicle fault positioning and rescue scheduling platform and method based on GPS track acquisition, and relates to the field of vehicle fault positioning and rescue scheduling, and the method comprises the steps: uploading vehicle GPS and OBD data obtained by a vehicle-mounted terminal to a cloud database through employing a mobile network technology; according to the vehicle GPS and OBD data in the cloud database, the vehicle fault type is predicted and judged, and the position coordinate information of the vehicle fault is obtained; based on an expert scoring method, processing response levels of different vehicle fault types are set; accessing a high-precision digital map API (Application Program Interface), and marking schedulable rescue points and fixed GPS (Global Positioning System) coordinate data in a digital map; and establishing a vehicle fault rescue scheduling model, and carrying out comprehensive evaluation and scheduling on surrounding schedulable rescue points. The method has the advantages of effectively predicting the state and fault of the vehicle, efficiently responding to the processing of the fault of the vehicle, and preferentially scheduling and screening the schedulable rescue points.

Description

Technical Field

[0001] The present invention relates to the field of vehicle fault location and rescue dispatch, and specifically to a vehicle fault location and rescue dispatch platform and method based on GPS trajectory collection. Background Art

[0002] With the development of new energy technologies, new energy vehicles are becoming more and more popular. For emergencies such as thermal runaway of new energy vehicle batteries and collision events, rescue points need to respond within minutes. However, traditional manual inspections or post-accident feedback from vehicle owners have both reflected the low efficiency of vehicle fault location and rescue dispatch, as well as the problem of untimely response. Therefore, by monitoring the GPS trajectory data and OBD data of vehicles, and using AI intelligent algorithms to obtain the vehicle's status information and predict the type of vehicle faults, it will effectively improve the preprocessing response time of rescue points, make rescue arrangements in a timely manner, reduce the degree of accidents caused by vehicle faults, and thus make vehicle fault rescue predictable and intelligent, which is of great significance for the safe driving of vehicles.

[0003] Existing technologies are single and unstable in vehicle fault judgment and prediction. For example, the driving state of a vehicle is evaluated through single GPS data or OBD data, or by combining both GPS and OBD data and obtaining rescue point information through simple matching. Such technologies cannot predict and evaluate the dynamic driving state of a vehicle due to the singularity of the data. Among them, only using OBD data cannot locate the fault occurrence location. For example, the P0A80 battery fault may occur on any section of the road, and only using GPS trajectories cannot determine the type of fault. Secondly, due to the lack of a mechanism for dynamic scheduling and rescue of vehicle faults in existing technologies, vehicle fault rescue is not timely, or the scheduling of rescue vehicles lacks global optimization, which is likely to cause property losses and the risk of secondary accidents. Summary of the Invention

[0004] To solve the above technical problems, a vehicle fault location and rescue dispatch platform and method based on GPS trajectory collection are provided. The present technical solution solves the problems raised in the above background art, namely, the inability to predict and evaluate the dynamic driving state of a vehicle due to the singularity of the data, and secondly, due to the lack of a mechanism for dynamic scheduling and rescue of vehicle faults in existing technologies, vehicle fault rescue is not timely, or the scheduling of rescue vehicles lacks global optimization, which is likely to cause property losses and the risk of secondary accidents.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A vehicle fault location and rescue dispatch method based on GPS trajectory collection, comprising: Using mobile network technology to upload the vehicle GPS and OBD data obtained by an in-vehicle terminal to a cloud database; Predict and judge the vehicle fault type and obtain the location coordinate information of the vehicle fault according to the vehicle GPS and OBD data in the cloud database; According to different vehicle fault types, set the processing response levels of different vehicle fault types based on the expert scoring method; Access the high-precision digital map API, obtain the road network topology and traffic congestion information in real time, and mark the dispatchable rescue points and fixed GPS coordinate data in the digital map; Establish a vehicle fault rescue dispatch model according to the coordinate data determined to be vehicle faults, and comprehensively evaluate and dispatch the surrounding dispatchable rescue points.

[0006] Preferably, the predicting and judging the vehicle fault type and obtaining the location coordinate information of the vehicle fault according to the vehicle GPS and OBD data in the cloud database specifically includes: Normalize the vehicle GPS and OBD data according to the normalization formula; Reduce the dimension and extract the principal component features of the GPS and OBD data in different vehicle fault types according to the principal component analysis method; Extract the time series features of the GPS and OBD data through the LSTM network and perform data encoding respectively; Calculate the attention weights of the GPS and OBD data according to the spatio-temporal attention mechanism formula, and obtain the correlation of the GPS and OBD data; Establish a vehicle fault prediction model, predict the vehicle fault occurrence probability and type according to the real-time received GPS and OBD data, and mark the location coordinate information of the vehicle fault.

[0007] Preferably, the establishing a vehicle fault rescue dispatch model according to the coordinate data determined to be vehicle faults and comprehensively evaluating and dispatching the surrounding dispatchable rescue points specifically includes: Obtain the GPS coordinate information of the vehicle fault according to the prediction result of the vehicle fault prediction model; Obtain the GPS coordinate data of each dispatchable rescue point according to the rescue resource database; Determine the straight-line distance from the dispatchable rescue point to the vehicle fault location based on the distance formula according to the GPS coordinate data of the vehicle fault and the dispatchable rescue point; Establish a search radius expansion formula, and dynamically adjust the search radius of the vehicle fault point according to the straight-line distance value from the dispatchable rescue point to the vehicle fault location; Among them, judge whether the straight-line distance value from the dispatchable rescue point to the vehicle fault location is less than the cumulative value of the search radius expansion. If so, it means that the dispatchable rescue point is a preselected point for dispatchable rescue. If not, it means that the dispatchable rescue point is not a preselected point for dispatchable rescue; Obtain the coordinate data of the schedulable rescue preselected points, and according to the road network topology and traffic congestion information in the rescue resource database, obtain the actual distances and congestion times from each schedulable rescue preselected point to the vehicle breakdown point; Establish a vehicle breakdown rescue scheduling model to comprehensively evaluate and schedule the surrounding schedulable rescue points; According to the machine learning algorithm, obtain the dynamic value of the balanced economic weight in the vehicle breakdown rescue scheduling model; The search radius expansion formula is: , In the formula, is the th search radius, is the reference constant, determined by the regional average density, is the number of schedulable rescue points found in the th search, is the attenuation coefficient, is the pre-set target number of schedulable rescue points; , In the formula, is the minimum value of the comprehensive evaluation of the balanced economy and response time for vehicle breakdown rescue scheduling, is the weight of the balanced economy, is the estimated economic cost of vehicle breakdown rescue scheduling, is the response time of vehicle breakdown rescue scheduling.

[0008] Furthermore, this solution proposes a vehicle breakdown location and rescue scheduling platform based on GPS trajectory collection, which is used to implement the vehicle breakdown location and rescue scheduling method based on GPS trajectory collection as described above, including: A data acquisition module, which is used to upload the vehicle GPS and OBD data obtained by the in-vehicle terminal to the cloud database by using mobile network technology; A prediction and location module, which is used to predict and judge the vehicle breakdown type and obtain the location coordinate information of the vehicle breakdown according to the vehicle GPS and OBD data in the cloud database; A rescue dispatching module, which is used to set the processing response levels for different vehicle fault types based on the expert scoring method according to different vehicle fault types; access the high-precision digital map API to obtain the road network topology and traffic congestion information in real time, and mark the dispatchable rescue points and fixed GPS coordinate data on the digital map; establish a vehicle fault rescue dispatching model based on the coordinate data determined to be vehicle faults, and comprehensively evaluate and dispatch the surrounding dispatchable rescue points.

[0009] Preferably, the rescue dispatching module specifically includes: A processing response level unit, which is used to set the processing response levels for different vehicle fault types based on the expert scoring method according to different vehicle fault types; A coordinate marking unit, which is used to access the high-precision digital map API to obtain the road network topology and traffic congestion information in real time, and mark the dispatchable rescue points and fixed GPS coordinate data on the digital map; A comprehensive dispatching unit, which is used to establish a vehicle fault rescue dispatching model based on the coordinate data determined to be vehicle faults, and comprehensively evaluate and dispatch the surrounding dispatchable rescue points.

[0010] Compared with the prior art, the beneficial effects of the present invention are as follows: This solution proposes a vehicle fault location and rescue dispatching method based on GPS trajectory collection. Through the dual fusion of vehicle GPS and OBD data, a vehicle fault prediction model is established based on the LSTM network and spatio-temporal attention mechanism. According to the real-time received GPS and OBD data, the probability and type of vehicle faults are predicted, and the position coordinate information of vehicle faults is marked, so as to realize the early prediction and high-precision positioning of faults. Secondly, according to the results of vehicle fault prediction, the vehicle fault coordinate information is obtained, and through the search radius expansion formula, the preselected points that meet the dispatchable rescue are screened out. Then, through the vehicle fault rescue dispatching model, the surrounding dispatchable rescue points are comprehensively evaluated and dispatched to obtain the optimal dispatchable rescue points, so as to effectively predict the vehicle state and faults, efficiently respond to the handling of vehicle faults, and select and screen the dispatchable rescue points preferentially. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 It is a flowchart of the vehicle fault location and rescue dispatching method based on GPS trajectory collection of the present invention; Figure 2 It is a flowchart of predicting and judging the vehicle fault type and obtaining the position coordinate information of vehicle faults according to the vehicle GPS and OBD data in the cloud database of the present invention; Figure 3Based on the coordinate data determined to be vehicle failures, a vehicle failure rescue scheduling model is established to comprehensively evaluate and schedule the surrounding schedulable rescue points. Detailed implementation manners

[0012] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments in the following description are only examples, and those skilled in the art can think of other obvious variations.

[0013] Refer to Figure 1 As shown, a vehicle failure positioning and rescue scheduling method based on GPS trajectory collection includes: Using mobile network technology to upload the vehicle GPS and OBD data obtained by the in-vehicle terminal to the cloud database; According to the vehicle GPS and OBD data in the cloud database, predict and judge the vehicle failure type and obtain the position coordinate information of the vehicle failure; According to different vehicle failure types, based on the expert scoring method, set the processing response levels for different vehicle failure types; Access the high-precision digital map API, obtain the road network topology and traffic congestion information in real time, and mark the schedulable rescue points and fixed GPS coordinate data in the digital map; Based on the coordinate data determined to be vehicle failures, establish a vehicle failure rescue scheduling model to comprehensively evaluate and schedule the surrounding schedulable rescue points.

[0014] It can be explained that GPS data can effectively obtain the running trajectory and failure positioning points of the vehicle, while OBD data can feedback the driving state and vehicle failure type of the vehicle. Therefore, by fusing GPS and OBD dual data, the ability to identify vehicle failure types and the ability to locate failure coordinates can be effectively improved. For example, when it is monitored through GPS data that the vehicle trajectory drifts abnormally within 10 minutes, that is, the radius > 15m, and according to the fault code feedback by the OBD interface and the monitored battery temperature > 80°C, then it can be judged as a vehicle battery thermal runaway failure. Or, when GPS shows that the vehicle idles for a long time and OBD monitors that the coolant temperature continues to rise, then it can be judged as a vehicle fan failure or coolant leakage. Secondly, according to the results of vehicle failure prediction, the vehicle failure coordinate information is obtained, the preselected points that meet schedulable rescue are screened through the search radius expansion formula, and through the vehicle failure rescue scheduling model, the surrounding schedulable rescue points are comprehensively evaluated and scheduled to obtain the optimal schedulable rescue points, so as to effectively predict the vehicle state and failure, efficiently respond to the handling of vehicle failures, and select the best schedulable rescue points by preference.

[0015] The process of uploading vehicle GPS and OBD data obtained by a vehicle terminal to a cloud database using mobile network technology specifically includes: According to the edge-cloud collaboration architecture of the cloud data transmission side, the transmission of vehicle operation data is divided into a vehicle terminal data acquisition layer, a data transmission edge layer, and a cloud data processing layer; Using mobile network technology, upload the vehicle GPS and OBD data collected by the vehicle terminal to the cloud database.

[0016] It can be explained that the accuracy of vehicle fault type identification and coordinate positioning depends on the accuracy of the monitoring data. Therefore, when collecting GPS and OBD data, it is necessary to ensure the consistency of the timestamps of the data to avoid data misalignment and reduced accuracy caused by inconsistent timestamps. Therefore, this solution establishes a synchronization mechanism for GPS, local crystal oscillator, and the cloud through a timestamp synchronization mechanism to achieve high-precision data alignment and status awareness. Among them, The vehicle terminal data acquisition layer includes: First, set a high-precision GPS acquisition device to obtain the GPS position data of the vehicle. Second, obtain the OBD interface data of the vehicle through the vehicle OBD interface; The data transmission edge layer specifically includes: Based on the Kalman filter algorithm, filter and denoise the received GPS and OBD data; Synchronize the acquisition frequencies of GPS and OBD data through the timestamp synchronization mechanism; Among them, the timestamp synchronization mechanism refers to obtaining the transmission delay time of vehicle data from the terminal to the cloud through GPS-UTC time, local crystal oscillator time, and cloud reception time; The cloud data processing layer refers to predicting and judging the vehicle fault type and obtaining the vehicle fault location data through GPS and OBD data; The expression for the transmission delay time of vehicle data from the terminal to the cloud is: , In the formula, is the transmission delay time of vehicle data from the terminal to the cloud, is the timestamp of the data received by the cloud, is the timestamp of the local crystal oscillator, is the timestamp of GPS-UTC, is the trust weight of the local crystal oscillator, , are fixed values, , are respectively the signal strength threshold and the duration threshold, which are fixed values. Among them, , , , It can be confirmed through big data analysis or test experiments.

[0017] Refer to Figure 2 As shown, the prediction and judgment of vehicle fault types and the acquisition of the position coordinate information of vehicle faults based on vehicle GPS and OBD data in the cloud database specifically include: Normalize the vehicle GPS and OBD data according to the normalization formula; Reduce the dimension and extract the principal component features of GPS and OBD data for different vehicle fault types according to the principal component analysis method; Extract the temporal features of GPS and OBD data through the LSTM network and perform data encoding respectively; Calculate the attention weights of GPS and OBD data according to the spatio-temporal attention mechanism formula to obtain the correlation of GPS and OBD data; Establish a vehicle fault prediction model, predict the probability and type of vehicle faults according to the real-time received GPS and OBD data, and mark the position coordinate information of vehicle faults.

[0018] It can be explained that when fusing GPS and OBD data, it is necessary to ensure the spatio-temporal alignment of the two groups of data. In this solution, the LSTM network is used to extract the temporal features of GPS and OBD data respectively, and according to the spatio-temporal attention mechanism formula, the attention weights of GPS and OBD data are calculated to obtain the correlation of GPS and OBD data, solving the data asynchrony problem in traditional data fusion methods and improving the joint accurate prediction of vehicle fault types and positions. Among them, the GPS data feature items include: the moving distance, heading angle, altitude, vehicle lateral acceleration and roll rate within a fixed time of the vehicle, and the OBD data feature items include: engine speed, engine load, fuel system status, ignition system status, coolant temperature and circulation efficiency, intake pressure, temperature, exhaust oxygen content, vehicle speed, vehicle longitudinal and lateral acceleration, braking state, steering wheel rotation angle, tire pressure, exhaust gas oxygen content, catalytic converter efficiency, EGR valve opening and flow, battery voltage, air-fuel ratio, throttle opening, engine knock count; The basic formula for extracting the temporal features of GPS and OBD data based on the LSTM network is: The forget gate adopts: , the input gate adopts: , the candidate memory adopts: , the memory update adopts: , the output gate adopts: , the hidden state adopts: , where , , , are weight matrices, is the Sigmoid function, is the element-wise multiplication symbol. Among them, due to the different input feature dimensions of GPS and OBD data, different weight matrices are adopted in this scheme to independently extract the temporal features of GPS and OBD data, thus avoiding feature confusion when fusing data; The formula of the spatio-temporal attention mechanism is: , In the formula, are the attention weights of GPS and OBD data, is the -th GPS feature data at the -th timestamp, is the -th OBD feature data at the -th timestamp, is the vector dimension, is the number of groups of GPS and OBD data.

[0019] The specific steps of setting the processing response levels for different vehicle fault types based on the expert scoring method include: Set the response levels for vehicle fault handling according to different vehicle fault types. Among them, the response levels are divided into four levels: low-level response, medium-level response, high-level response, and extra-high-level response; Set the response time intervals for vehicle fault handling according to the response levels of vehicle fault handling; Based on the historical data of vehicle fault handling responses, obtain the comprehensive value of vehicle fault handling response time and satisfaction feedback; Based on the Pearson correlation coefficient formula, obtain the correlation between the vehicle fault handling response time and the comprehensive value of satisfaction feedback; Based on the expert scoring method, generate a scoring matrix for the vehicle fault handling response levels corresponding to different vehicle fault types, so as to determine the processing response levels for different vehicle fault types.

[0020] It can be explained that for different vehicle fault types, different processing response levels can be set according to the degree of vehicle faults and the degree of traffic impact. Different response time intervals for vehicle fault handling can be set through different processing response levels. The processing response levels for different vehicle fault types can be determined through the expert scoring method. However, a single expert scoring method may lead to inaccurate evaluation results due to the influence of subjective experience. Therefore, in this solution, based on the historical data of vehicle fault handling responses, the comprehensive value of vehicle fault handling response time and satisfaction feedback is obtained, and based on the Pearson correlation coefficient formula, the correlation between the vehicle fault handling response time and the satisfaction feedback comprehensive value is obtained. By introducing the satisfaction feedback comprehensive value, the accuracy and reliability of the evaluation of the processing response levels for different vehicle fault types are improved. Among them, , In the formula, is the correlation value of the comprehensive value of the vehicle fault handling response time and satisfaction feedback for the th vehicle fault type, is the scoring matrix of the vehicle fault handling response level corresponding to the th vehicle fault type.

[0021] The access to the high-precision digital map API, real-time acquisition of the road network topology structure and traffic congestion information, and marking of the dispatchable rescue points and fixed GPS coordinate data on the digital map specifically include: Access the high-precision digital map API, and through the high-precision digital map, real-time acquire the road network topology structure and traffic congestion information; Mark the dispatchable rescue points on the high-precision digital map, and acquire the GPS coordinate data of each dispatchable rescue point; Build a rescue resource database according to the road network topology structure, traffic congestion information and the GPS coordinate data of the dispatchable rescue points; Set an automatic trigger instruction for data call. When it is determined that the vehicle has a fault, an instruction is automatically sent to retrieve the data in the rescue resource database.

[0022] It can be explained that when positioning the vehicle fault point, not only the GPS coordinates of the vehicle are required, but also the road network topology structure and traffic congestion information of the faulty vehicle, as well as the GPS coordinate data of each dispatchable rescue point. Therefore, in this solution, by accessing the high-precision digital map API, such as Amap, Baidu Map, etc., the coordinate information of the vehicle and the dispatchable rescue points is accurately acquired through the high-precision digital map API, so as to understand the road conditions information of the faulty vehicle.

[0023] Refer to Figure 3As shown, establishing a vehicle fault rescue scheduling model based on the coordinate data determined to be vehicle faults and comprehensively evaluating and scheduling the schedulable rescue points in the vicinity specifically includes: Obtaining the GPS coordinate information of the vehicle fault according to the prediction result of the vehicle fault prediction model; Obtaining the GPS coordinate data of each schedulable rescue point according to the rescue resource database; Based on the distance formula and according to the GPS coordinate data of the vehicle fault and the schedulable rescue points, determining the straight-line distance from the schedulable rescue point to the vehicle fault location; Establishing a search radius expansion formula and dynamically adjusting the search radius of the vehicle fault point according to the straight-line distance value from the schedulable rescue point to the vehicle fault location; Among them, it is judged whether the straight-line distance value from the schedulable rescue point to the vehicle fault location is less than the cumulative value of the search radius expansion. If so, it means that the schedulable rescue point is a preselected point for schedulable rescue. If not, it means that the schedulable rescue point is not a preselected point for schedulable rescue; Obtaining the coordinate data that meets the preselected points for schedulable rescue, and obtaining the actual distance and congestion time from each preselected point for schedulable rescue to the vehicle fault point according to the road network topology structure and traffic congestion information in the rescue resource database; Establishing a vehicle fault rescue scheduling model and comprehensively evaluating and scheduling the schedulable rescue points in the vicinity; Obtaining the dynamic value of the balanced economic weight in the vehicle fault rescue scheduling model according to the machine learning algorithm; The search radius expansion formula is: , In the formula, is the -th search radius, is the reference constant determined by the regional average density, is the -th number of schedulable rescue points searched, is the attenuation coefficient, is the preset target number of schedulable rescue points; The expression for establishing the vehicle fault rescue scheduling model is: , In the formula, is the minimum value of the comprehensive evaluation of the balanced economy and response time of the vehicle fault rescue scheduling, is the weight of the balanced economy, is the estimated economic cost of the vehicle fault rescue scheduling, is the response time of the vehicle fault rescue scheduling.

[0024] It can be explained that this solution provides a method for expanding the dynamic search radius. By using the GPS coordinate data of vehicle failures and dispatchable rescue points, the straight-line distance value from the dispatchable rescue point to the vehicle failure location is obtained. By determining whether the straight-line distance value from the dispatchable rescue point to the vehicle failure location is less than the cumulative value of the search radius expansion, it is determined whether the dispatchable rescue point is a preselected point for dispatchable rescue. Finally, through the search radius expansion formula, the range of the search radius expansion is gradually adjusted dynamically according to the number of dispatchable rescue points searched. Secondly, vehicle failure rescue dispatch is mainly determined by dispatch economic costs and processing response times. Therefore, this solution provides a vehicle failure rescue dispatch model. By obtaining the minimum comprehensive evaluation value that balances economy and response time in vehicle failure rescue dispatch, the optimal dispatchable rescue point is determined, thereby intelligently screening out suitable dispatchable rescue points. Among them, in the search radius expansion formula is a reference constant determined by the regional average density. For example, in the city, C = 50, and in the suburbs, C = 100. The area can be divided through the high-precision digital map API, and through the coordinate information, the value can be automatically obtained.

[0025] The specific process of obtaining the dynamic value of the balanced economic weight in the vehicle failure rescue dispatch model according to the machine learning algorithm is as follows: Extract the feature items that affect vehicle failure rescue dispatch according to the influencing factors of vehicle failure rescue dispatch; Among them, the feature items that affect vehicle failure rescue dispatch include: the actual distance from the preselected dispatchable rescue point to the vehicle failure point, congestion time, processing response level, and dispatch cost; Based on the historical data of vehicle failure rescue dispatch, establish a training sample set and a target sample set for the balanced economic weight of vehicle failure rescue dispatch; The balanced economic weight and the feature items that affect vehicle failure rescue dispatch satisfy a linear regression equation, and its formula is: where, is the balanced economic weight value in the vehicle failure rescue dispatch model, is the balance constant term, , , , are the parameter values of the actual distance from the preselected dispatchable rescue point to the vehicle failure point, congestion time, processing response level, and dispatch cost respectively, , , , are the actual distance from the preselected dispatchable rescue point to the vehicle failure point, congestion time, processing response level, and dispatch cost; The least squares method is used for the loss function. , where is the loss degree value between the training value and the target value of the balanced economic weight in the vehicle fault rescue scheduling model, is the number of data groups in the training sample set and the target sample set, is the target value of the balanced economic weight in the vehicle fault rescue scheduling model; The update equation of the parameters is obtained by using the batch gradient descent algorithm, and its expression is: , where is the updated value of the parameter of the th feature item affecting vehicle fault rescue scheduling, is the th feature item of vehicle fault rescue scheduling and the th parameter value, is the learning rate, is the number of batches participating in data update, is the th feature item of vehicle fault rescue scheduling and the th data value.

[0026] By using the machine learning algorithm of traditional linear regression, the dynamic value of the balanced economic weight in the vehicle fault rescue scheduling model is obtained.

[0027] It can be explained that since the actual distances, congestion times, processing response levels, and scheduling costs from the schedulable rescue preselected points corresponding to different vehicle fault types to the vehicle fault points are different, and the influence degrees of these features on the scheduling economic cost and processing response time are also different. For example, the higher the processing response level, the shorter the required processing response time, and then the greater the weight value corresponding to the processing response time. Therefore, to ensure the accuracy of the vehicle fault rescue scheduling model, it is necessary to obtain the dynamic balanced economic weight value in the vehicle fault rescue scheduling model. Therefore, this solution uses the machine learning algorithm of traditional linear regression to obtain the dynamic value of the balanced economic weight in the vehicle fault rescue scheduling model, thereby ensuring the accuracy and reliability of the vehicle fault rescue scheduling model.

[0028] Furthermore, based on the same inventive concept as the above vehicle fault location and rescue scheduling method based on GPS trajectory collection, this solution proposes a vehicle fault location and rescue scheduling platform based on GPS trajectory collection, including: A data acquisition module, which is used to upload the vehicle GPS and OBD data obtained by the in-vehicle terminal to the cloud database by using mobile network technology; A prediction and positioning module, which is used to predict and judge the types of vehicle faults and obtain the position coordinate information of vehicle faults according to the vehicle GPS and OBD data in the cloud database; A rescue dispatching module, which is used to set the handling response levels for different types of vehicle faults based on the expert scoring method according to different types of vehicle faults; access the high-precision digital map API to obtain the road network topology and traffic congestion information in real time, and mark the dispatchable rescue points and fixed GPS coordinate data in the digital map; establish a vehicle fault rescue dispatching model according to the coordinate data determined to be vehicle faults, and comprehensively evaluate and dispatch the surrounding dispatchable rescue points; The rescue dispatching module specifically includes: A handling response level unit, which is used to set the handling response levels for different types of vehicle faults based on the expert scoring method according to different types of vehicle faults; A coordinate marking unit, which is used to access the high-precision digital map API to obtain the road network topology and traffic congestion information in real time, and mark the dispatchable rescue points and fixed GPS coordinate data in the digital map; A comprehensive dispatching unit, which is used to establish a vehicle fault rescue dispatching model according to the coordinate data determined to be vehicle faults, and comprehensively evaluate and dispatch the surrounding dispatchable rescue points.

[0029] In summary, the advantages of the present invention are as follows: effectively predicting the vehicle state and faults, efficiently responding to the handling of vehicle faults, and preferentially dispatching and screening dispatchable rescue points.

[0030] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art of this industry should understand that the present invention is not limited by the above embodiments. What is described in the above embodiments and the specification is only the principle of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A vehicle fault location and rescue dispatching method based on GPS trajectory collection, characterized in that, Including: Using mobile network technology to upload the vehicle GPS and OBD data obtained by the vehicle terminal to the cloud database; Predicting and judging the vehicle fault type and obtaining the location coordinate information of the vehicle fault according to the vehicle GPS and OBD data in the cloud database; Setting the processing response levels for different vehicle fault types based on the expert scoring method according to different vehicle fault types; Accessing the high-precision digital map API to obtain the road network topology and traffic congestion information in real time, and marking the dispatchable rescue points and fixed GPS coordinate data in the digital map; Establishing a vehicle fault rescue dispatch model according to the coordinate data determined as vehicle faults, and comprehensively evaluating and dispatching the surrounding dispatchable rescue points.

2. The vehicle fault location and rescue scheduling method based on GPS trajectory collection according to claim 1, characterized in that, The specific steps of using mobile network technology to upload the vehicle GPS and OBD data obtained by the vehicle terminal to the cloud database include: According to the cloud data transmission end-edge-cloud collaborative architecture, dividing the vehicle operation data transmission into the vehicle terminal data collection layer, the data transmission edge layer, and the cloud data processing layer; Using mobile network technology to upload the vehicle GPS and OBD data collected by the vehicle terminal to the cloud database.

3. The vehicle fault location and rescue scheduling method based on GPS trajectory collection according to claim 2, wherein, The specific steps of predicting and judging the vehicle fault type and obtaining the location coordinate information of the vehicle fault according to the vehicle GPS and OBD data in the cloud database include: Normalizing the vehicle GPS and OBD data according to the normalization formula; Performing dimensionality reduction and principal component feature extraction on the GPS and OBD data in different vehicle fault types according to the principal component analysis method; Extracting the time series features of the GPS and OBD data through the LSTM network and performing data encoding respectively; Calculating the attention weights of the GPS and OBD data according to the spatio-temporal attention mechanism formula to obtain the correlation of the GPS and OBD data; Establishing a vehicle fault prediction model, predicting the vehicle fault occurrence probability and type according to the real-time received GPS and OBD data, and marking the location coordinate information of the vehicle fault.

4. The vehicle fault location and rescue scheduling method based on GPS trajectory collection according to claim 3, wherein, The specific steps of setting the processing response levels for different vehicle fault types based on the expert scoring method according to different vehicle fault types include: Setting the response levels for vehicle fault handling according to different vehicle fault types, where the response levels are divided into four levels: low-level response, medium-level response, high-level response, and extra-high-level response; Setting the response time intervals for vehicle fault handling according to the response levels for vehicle fault handling; Obtaining the comprehensive value of the vehicle fault handling response time and satisfaction feedback based on the historical data of vehicle fault handling responses; Obtaining the correlation between the vehicle fault handling response time and the comprehensive value of satisfaction feedback based on the Pearson correlation coefficient formula; Generating a scoring matrix for the vehicle fault handling response levels corresponding to different vehicle fault types based on the expert scoring method, so as to determine the processing response levels for different vehicle fault types.

5. A vehicle fault location and rescue scheduling method based on GPS trajectory collection according to claim 4, characterized in that, The specific steps of accessing the high-precision digital map API to obtain the road network topology and traffic congestion information in real time, and marking the dispatchable rescue points and fixed GPS coordinate data in the digital map include: Access the high-precision digital map API, and through the high-precision digital map, obtain the road network topology and traffic congestion information in real time; Mark the dispatchable rescue points in the high-precision digital map, and obtain the GPS coordinate data of each dispatchable rescue point; Construct a rescue resource database based on the road network topology, traffic congestion information, and GPS coordinate data of the dispatchable rescue points; Set an automatic trigger instruction for data call. When it is determined that the vehicle has a fault, an instruction is automatically sent to retrieve the data in the rescue resource database.

6. The vehicle fault location and rescue scheduling method based on GPS trajectory collection according to claim 5, wherein The comprehensive evaluation and dispatch of the surrounding dispatchable rescue points by establishing a vehicle fault rescue dispatch model according to the coordinate data determined to be a vehicle fault specifically include: Obtain the GPS coordinate information of the vehicle fault according to the prediction result of the vehicle fault prediction model; Obtain the GPS coordinate data of each dispatchable rescue point according to the rescue resource database; Based on the distance formula, determine the straight-line distance from the dispatchable rescue point to the vehicle fault location according to the vehicle fault and the GPS coordinate data of the dispatchable rescue point; Establish a search radius expansion formula, and dynamically adjust the search radius of the vehicle fault point according to the straight-line distance value from the dispatchable rescue point to the vehicle fault location; Among them, it is judged whether the straight-line distance value from the dispatchable rescue point to the vehicle fault location is less than the cumulative value of the search radius expansion. If so, it means that the dispatchable rescue point is a preselected point for dispatchable rescue. If not, it means that the dispatchable rescue point is not a preselected point for dispatchable rescue; Obtain the coordinate data that meets the preselected points for dispatchable rescue, and according to the road network topology and traffic congestion information in the rescue resource database, obtain the actual distance and congestion time from each preselected point for dispatchable rescue to the vehicle fault point; Establish a vehicle fault rescue dispatch model to comprehensively evaluate and dispatch the surrounding dispatchable rescue points; Obtain the dynamic value of the balanced economic weight in the vehicle fault rescue dispatch model according to the machine learning algorithm; The search radius expansion formula is: ， Wherein, is the th search radius, is a reference constant determined by the regional average density, is the number of schedulable rescue points searched at the th time, is the attenuation coefficient, is the preset number of target schedulable rescue points; The expression for establishing the vehicle fault rescue dispatch model is: ， In the formula, is the minimum value of the comprehensive evaluation of vehicle fault rescue scheduling for balancing economy and response time, is the weight of the balanced economy, is the estimated economic cost of vehicle fault rescue scheduling, is the response time of vehicle fault rescue scheduling.

7. A vehicle fault location and rescue dispatch method based on GPS trajectory collection according to claim 6, characterized in that The obtaining of the dynamic value of the balanced economic weight in the vehicle fault rescue dispatch model according to the machine learning algorithm specifically includes: Extract the characteristic items affecting the vehicle fault rescue dispatch according to the influencing factors of the vehicle fault rescue dispatch; Among them, the characteristic items affecting the vehicle fault rescue dispatch include: the actual distance from the preselected point for dispatchable rescue to the vehicle fault point, the congestion time, the processing response level, and the dispatch cost; Based on the historical data of the vehicle fault rescue dispatch, establish a training sample set and a target sample set for the balanced economic weight of the vehicle fault rescue dispatch; The balanced economic weight and the characteristic items affecting vehicle breakdown rescue scheduling satisfy a linear regression equation, and its formula is: , where is the balanced economic weight value in the vehicle breakdown rescue scheduling model, is the balance constant term, , , , are respectively the parameter values of the actual distance, congestion time, processing response level, and scheduling cost from the schedulable rescue preselected point to the vehicle breakdown point, , , , are the actual distance, congestion time, processing response level, and scheduling cost from the schedulable rescue preselected point to the vehicle breakdown point; The loss function uses the least squares method, , where is the loss degree value of the equilibrium economic weight training value and the equilibrium economic weight target value in the vehicle fault rescue scheduling model, is the number of data groups in the training sample set and the target sample set, is the equilibrium economic weight target value in the vehicle fault rescue scheduling model; The update equation of the parameters is obtained by using the batch gradient descent algorithm, and its expression is: , where is the updated value of the th feature item parameter that affects vehicle fault rescue scheduling, is the th feature item and the th parameter value that affects vehicle fault rescue scheduling, is the learning rate, is the number of batches participating in data update, is the th feature item and the th data value that affects vehicle fault rescue scheduling; Obtain the dynamic value of the balanced economic weight in the vehicle fault rescue dispatch model by using the machine learning algorithm of traditional linear regression.

8. A vehicle fault location and rescue scheduling platform based on GPS trajectory collection, characterized in that, For implementing the vehicle fault location and rescue dispatch method based on GPS trajectory collection as described in any one of claims 1-7, including: A data acquisition module, which is used to upload the vehicle GPS and OBD data obtained by the in-vehicle terminal to the cloud database by using mobile network technology; A prediction and positioning module, which is used to predict and judge the types of vehicle faults and obtain the location coordinate information of vehicle faults according to the vehicle GPS and OBD data in the cloud database; A rescue dispatching module, which is used to set the processing response levels for different vehicle fault types based on the expert scoring method according to different vehicle fault types; access the high-precision digital map API to obtain the road network topology and traffic congestion information in real time, and mark the dispatchable rescue points and fixed GPS coordinate data on the digital map; establish a vehicle fault rescue dispatching model according to the coordinate data determined to be vehicle faults, and conduct comprehensive evaluation and dispatching of the surrounding dispatchable rescue points.

9. The vehicle fault location and rescue dispatch platform based on GPS trajectory collection according to claim 8, characterized in that, The rescue dispatching module specifically includes: A processing response level unit, which is used to set the processing response levels for different vehicle fault types based on the expert scoring method according to different vehicle fault types; A coordinate marking unit, which is used to access the high-precision digital map API to obtain the road network topology and traffic congestion information in real time, and mark the dispatchable rescue points and fixed GPS coordinate data on the digital map; A comprehensive dispatching unit, which is used to establish a vehicle fault rescue dispatching model according to the coordinate data determined to be vehicle faults, and conduct comprehensive evaluation and dispatching of the surrounding dispatchable rescue points.

Images