A waybill enhancement method based on dangerous goods transport vehicle trajectory data

Abstract

The present invention relates to the field of dangerous goods transportation supervision technology, and in particular to a waybill enhancement method based on trajectory data of dangerous goods transportation vehicles, the method comprising obtaining high-dimensional trajectory data of the transportation vehicle, time series trajectory data without missing values; identifying a valid stop point list of the transportation vehicle; performing behavioral feature analysis and duration analysis on the valid stop point list, correlating and comparing the classified stop information of the same transportation vehicle with the pre-filled waybill information, and generating an enhanced waybill based on the consistency between the classified stop information of the time series trajectory data and the pre-filled waybill information. The present invention significantly improves the monitoring and management capabilities of the dangerous goods transportation process by generating accurate and effective enhanced waybills, ensuring the timely identification and handling of abnormal situations, and the enhanced waybills also help to quickly trace the cause of the accident.

Description

Technical Field

[0001] The present invention relates to the technical field of dangerous goods transportation management, and in particular to a waybill enhancement method based on trajectory data of dangerous goods transportation vehicles. Background Art

[0002] At present, one of the core challenges facing the field of dangerous goods transportation supervision is the accuracy and completeness of waybill information. In the existing technology, the management department relies on the waybill information pre-filled by the transportation company. Although it is simple, it has obvious defects such as inaccurate information and missing details. The pre-filled waybill often cannot accurately reflect the actual parking location, travel chain details and dynamic behavior of the vehicle, resulting in frequent blind spots in supervision.

[0003] Due to changes in the external environment and interference from human factors, there is often an information mismatch between the pre-filled waybills and the actual transportation situation, which seriously restricts the management department's accurate grasp of the actual vehicle transportation situation.

[0004] At present, the location and time information in the pre-filled waybill is not accurate enough, and fuzzy descriptions are often used instead of precise coordinates. The time information is too broad and lacks the accuracy to be refined to specific time periods or timestamps, making it difficult to support detailed analysis of transportation time and efficiency. Although there are attempts to make up for data missing through data interpolation and generative adversarial networks in the existing technology, these methods often have difficulty in effectively capturing long-distance relationships between data when processing high-dimensional data, and are prone to generating duplicate samples, affecting the quality and effectiveness of data completion.

[0005] What is particularly critical is that the existing technology lacks a special method to directly use the real trajectory data of transport vehicles to enhance the waybill information. Even if the trajectory data contains rich real-time and accurate information, this information cannot be effectively integrated into the waybill management system, limiting the accuracy of the waybill information and the improvement of supervision efficiency. In this way, the low accuracy of pre-filled waybills, insufficient information precision and the limitations of data interpolation technology in processing high-dimensional data have further aggravated the problems of blind spots in supervision and information mismatch, and seriously restricted the accurate grasp and efficient supervision of the dangerous goods transportation process. Summary of the invention

[0006] The present invention aims to solve the technical problems of weak enhanced analysis and low supervision efficiency of waybill information of dangerous goods in the prior art, and provides a waybill enhancement method based on trajectory data of dangerous goods transport vehicles.

[0007] In view of the above technical problems, an embodiment of the present invention provides a waybill enhancement method based on dangerous goods transport vehicle trajectory data, comprising:

[0008] Obtain high-dimensional trajectory data of transport vehicles, perform interpolation processing on the high-dimensional trajectory data, and output time series trajectory data without missing values;

[0009] According to the KSTV-DBSCAN algorithm with threshold rules and adaptive mechanism, cluster analysis is performed on the time series trajectory data without missing values ​​to identify the valid stop list of transport vehicles;

[0010] Performing behavior feature analysis and duration analysis on the valid stop point list to obtain classified stop information of transport vehicles at the valid stop points, wherein the classified stop information includes the stop position information of all valid stop points;

[0011] The classified stop information of the same transport vehicle is correlated and compared with the pre-filled waybill information, and an enhanced waybill is generated based on the consistency between the classified stop information of the time series trajectory data and the pre-filled waybill information.

[0012] In the present invention, the high-dimensional trajectory data of dangerous goods transport vehicles is interpolated through a deep learning network, which effectively solves the problem of missing trajectory points caused by equipment failure or signal transmission problems. The deep learning network uses an integrated multi-head self-attention mechanism to focus on different aspects of the data through multiple attention heads working in parallel, effectively capturing the dependencies between long-distance features in the data, which is difficult to accurately achieve with traditional sequence models such as GRU; the parallel processing method not only makes up for the shortcomings of GRU in capturing long-distance dependencies, but also enhances the ability to understand contextual information through rich feature representation, generating more complete and accurate time series data.

[0013] The adaptive mechanism of the present invention can accurately identify dense and sparse areas in the data by dynamically adjusting the values ​​of the domain radius eps and the minimum number of points MinPTs within the domain radius, thereby generating a clustering result that is more in line with the natural distribution of the data. Since different data sets have different feature distributions and density changes, fixed hyperparameter settings are often difficult to apply to all situations. The adaptive mechanism can automatically adjust parameters according to the characteristics of different data sets, enhancing the flexibility and robustness of the KSTV-DBSCAN algorithm when processing different types of data. The adaptive mechanism greatly reduces the need for manual intervention by automatically finding the optimal parameter combination, and can also be used to achieve multi-scale clustering, that is, analyzing the clustering structure of data sets at different scales.

[0014] By analyzing the trajectory data, the present invention records in detail the specific stop information and travel chain in each transportation task, greatly enriching the content of the waybill. This not only makes the transportation process more transparent, but also enhances the monitoring and management capabilities of the transportation process. By comparing the trajectory data with the pre-filled waybill information, the mismatch between the two is effectively eliminated, unreal waybills are removed, and incorrectly filled waybills are corrected. This process improves the authenticity of the waybills. By using the timestamp information in the trajectory data, the transportation time is refined to the second level, which significantly improves the accuracy of the time information in the waybill. The refined information provides managers with more detailed data support, which helps to grasp the transportation progress and optimize the transportation route and time schedule. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 Schematic diagram of a flow chart of a waybill enhancement method in one embodiment of the present invention;

[0016] Figure 2 Schematic diagram of the process of step S10 of the waybill enhancement method in one embodiment of the present invention;

[0017] Figure 3 Schematic diagram of the process of step S103 of the waybill enhancement method in one embodiment of the present invention;

[0018] Figure 4 is a schematic diagram of an interpolation algorithm in one embodiment of the present invention;

[0019] Figure 5 Schematic diagram of the process of step S20 of the waybill enhancement method in one embodiment of the present invention;

[0020] Figure 6 A clustering principle diagram of the DBSCAN algorithm in one embodiment of the present invention;

[0021] Figure 7 is a schematic diagram of a waybill enhancement system in one embodiment of the present invention;

[0022] Figure 8 It is a schematic diagram of the structure of a controller in one embodiment of the present invention. DETAILED DESCRIPTION

[0023] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] In one embodiment, if Figure 1As shown, a waybill enhancement method based on dangerous goods transport vehicle trajectory data includes the following steps S10-S40:

[0025] S10, obtaining high-dimensional trajectory data of transport vehicles, and interpolating the high-dimensional trajectory data to output time series trajectory data without missing values; understandably, by obtaining high-dimensional trajectory data of dangerous goods transport vehicles, raw materials can be provided for subsequent analysis and processing. Since some trajectory points are missing in the high-dimensional trajectory data, interpolation processing can fill these missing values, and the interpolated time series trajectory data is continuous and complete, which can improve the accuracy and reliability of the analysis.

[0026] S20. According to the KSTV-DBSCAN algorithm that introduces threshold rules and adaptive mechanisms, cluster analysis is performed on the time series trajectory data without missing values ​​to identify a list of valid stop points for transport vehicles. It can be understood that the KSTV-DBSCAN algorithm that introduces threshold rules and adaptive mechanisms is used to perform spatial clustering on continuous and complete time series trajectory data to automatically identify and screen out valid stop points for transport vehicles during the transportation process, thereby improving the accuracy and applicability of the clustering results.

[0027] S30, analyze the behavior characteristics and duration of the valid stop point list to obtain classified stop information of the transport vehicle at the valid stop points, wherein the classified stop information includes the stop position information of all valid stop points; understandably, by analyzing the behavior characteristics of the valid stop points, the behavior patterns of the transport vehicle at different stop points can be understood, such as the stop position, activity type, speed change, etc. By analyzing the duration of the stop points, the stay time of the transport vehicle at each stop point can be obtained.

[0028] After behavioral feature analysis and duration analysis, the classified stop information of transport vehicles at effective stop points can be obtained, which is of great significance for enhancing waybill information and improving the level of transport supervision.

[0029] S40, associating and comparing the classified stop information of the same transport vehicle with the pre-filled waybill information, and generating an enhanced waybill based on the consistency between the classified stop information of the time series trajectory data and the pre-filled waybill information. Understandably, associating and comparing the classified stop information with the pre-filled waybill information can identify errors and omissions in the pre-filled waybill, help eliminate the phenomenon of mismatch between the pre-filled waybill and the actual waybill information, and improve the authenticity and accuracy of the waybill information.

[0030] In the present invention, the high-dimensional trajectory data of dangerous goods transport vehicles is interpolated and improved to ensure the integrity and accuracy of the data. Based on the improved high-dimensional trajectory data, the KSTV-DBSCAN algorithm that introduces threshold rules and adaptive mechanisms is used to automatically extract the actual stop points in the transportation process and identify the valid stop points. By performing behavioral feature analysis and duration analysis on the valid stop points, the stop location information of the valid stop points is obtained. Furthermore, the stop location information is intelligently matched and verified with the pre-filled waybill information, errors are automatically corrected and missing records are supplemented, and an accurate and efficient enhanced waybill is generated. The enhanced waybill not only improves the transparency and controllability of the dangerous goods transportation process, but also provides strong technical support for regulatory authorities, thereby ensuring the safety and compliance of dangerous goods transportation.

[0031] In one embodiment, if Figure 2 As shown, the step S10, that is, obtaining high-dimensional trajectory data of the transport vehicle, and interpolating the high-dimensional trajectory data to output time series trajectory data without missing values, includes the following steps S101-S105:

[0032] S101, set a preset time interval to obtain high-dimensional trajectory data of the transport vehicle, the high-dimensional trajectory data includes the location ID of the transport vehicle, the second-level transport timestamp, longitude and latitude, driving speed and driving direction angle; it is understandable that the preset time interval can be set according to demand, the value range of the preset time interval can be 25s-35s, and the adjustability of the preset time interval can adapt to the needs of different monitoring densities in different transport stages. By setting a preset time interval to obtain high-dimensional trajectory data of the transport vehicle, detailed and continuous monitoring of the dangerous goods transportation process can also be achieved. Among them, the location Φ of the transport vehicle provides a unique identifier for the location of the transport vehicle at different time points, which is convenient for tracking and management; the second-level transport timestamp provides the specific time point of the transport vehicle location information, in seconds, to ensure the accuracy of the time information; the longitude and latitude record the specific geographical location of the transport vehicle; the driving speed reflects the moving speed of the transport vehicle at a specific time point, which is an important parameter for analyzing transportation efficiency and safety; the driving direction angle indicates the driving direction of the transport vehicle, which helps to understand the transportation path and behavior pattern.

[0033] S102. Use Kalman filtering to denoise the high-dimensional trajectory data. It can be understood that the Kalman filter uses a mathematical model to predict and smooth data, which can remove errors and redundant information in the high-dimensional trajectory data, clean and denoise the high-dimensional trajectory data, effectively reduce the impact of noise, and ensure the quality and reliability of the data.

[0034] S103, using a deep learning network to interpolate the high-dimensional trajectory data after denoising, and output time series trajectory data without missing values;

[0035] In one embodiment, if Figures 3 to 4 As shown in the figure, the step S103 includes the following steps S1031-S1035:

[0036] S1031, inputting the denoised high-dimensional trajectory data into a generator, wherein the generator includes a first-level FC fully connected layer, a first-level gated time unit GRU layer, a multi-head self-attention mechanism layer, a second-level gated time unit GRU layer, and a second-level FC fully connected layer connected in sequence;

[0037] The first-level FC fully connected layer is used to perform feature mapping on the original high-dimensional time series data and random matrices containing missing values ​​in the high-dimensional trajectory data, and generate high-dimensional features; understandably, the first-level FC fully connected layer receives the original high-dimensional time series data, and maps the original data with missing values ​​and the random matrix to a higher-dimensional feature space, so that the subsequent GRU layer can process more information and features.

[0038] S1032. Use the first-level gated time unit GRU layer to capture the dependencies in the original high-dimensional time series data, and process and generate coherent and logical sequence data; understandably, the first-level gated time unit GRU layer captures the dependencies in the input features, updates and maintains the internal state, generates coherent and logical sequence data, and provides optimized sequence representation for the subsequent self-attention mechanism and the second GRU.

[0039] S1033. Use the multi-head self-attention mechanism layer to learn the relationship between long-distance features in the original high-dimensional time series data. Each head learns different feature representations and generates global information, and converts the global information into a weight matrix. Understandably, the multi-head self-attention mechanism layer can make up for the defect of the first GRU in extracting the dependency relationship of long-distance features inaccurately, and convert the learned information into a weight matrix. Through multiple attention heads working in parallel, each head learns different feature representations to form a diversified feature representation, and generates global information and converts it into a weight matrix. It can provide richer contextual information and improve the feature expression ability of the generated network. Integrating the multi-head self-attention mechanism into the autoencoder effectively improves the performance of the model in missing value filling, especially when dealing with complex and heterogeneous traffic data, it can better retain and reconstruct important information in the data.

[0040] S1034. Use the second-level gated time unit GRU layer to receive global information and refine and adjust the sequence data; understandably, the second-level gated time unit GRU layer uses the global information provided by the multi-head self-attention mechanism to further refine and adjust the sequence data, ensure the coherence of the sequence data, and optimize it according to the global information.

[0041] S1035, using the second-level FC fully connected layer to map the high-dimensional features generated by the first-level FC fully connected layer back to the dimension of the target data space, generating reconstructed time series trajectory data without missing values. Understandably, this step ensures that the format of the output data is consistent with the target data, completing the reconstruction of the data.

[0042] In this embodiment, through steps S1031-S1035, the deep learning network can effectively process time series data with missing values, especially in the field of complex and heterogeneous traffic data. The deep learning network uses an integrated multi-head self-attention mechanism to focus on different aspects of the data through multiple attention heads working in parallel, effectively capturing the dependencies between long-distance features in the data, which is difficult to accurately achieve with traditional sequence models such as GRU; the parallel processing method not only makes up for the shortcomings of GRU in capturing long-distance dependencies, but also enhances the ability to understand contextual information through rich feature representation, generating a more accurate and comprehensive weight matrix.

[0043] The GRU layer of the deep learning network further refines and adjusts the sequence data to ensure the logical coherence of the generated sequence. Finally, the high-dimensional features are mapped back to the target data space through the FC fully connected layer to generate a complete time series without missing values. This process not only improves the integrity of the data, but also significantly improves the performance in the missing value filling task, especially when dealing with highly dynamic and changeable traffic data sets, it can more effectively retain and reconstruct the key information of the data, thereby enhancing the prediction and reconstruction capabilities of the overall model.

[0044] S104, the discriminator receives the original high-dimensional time series data and the reconstructed time series trajectory data without missing values ​​generated by the generator. Under the same condition vector, the discriminator distinguishes the filled data and the real data in the original high-dimensional time series data, and optimizes the generator and the discriminator through error back propagation. It can be understood that the discriminator can distinguish the input original high-dimensional time series data and the reconstructed data generated by the generator, and provide feedback to the generator by learning to distinguish the original data from the reconstructed data. If the discriminator can easily distinguish the original data from the reconstructed data, it means that the output of the generator is not realistic enough. If the discriminator is difficult to distinguish, it means that the performance of the generator is better. The output error of the discriminator is passed back to the generator through the back propagation algorithm to optimize the parameters of the generator. In this way, through the feedback mechanism of the discriminator, the performance of the generator is optimized, so that it can generate more accurate and realistic reconstructed time series data, and improve the overall effect of the data interpolation task.

[0045] S105. By stepwise iteration, the time series trajectory data generated by the generator without missing values ​​is optimized until the discriminator cannot distinguish between the filled part and the real data part, and the iteration is stopped. It can be understood that stepwise iteration makes the data generated by the generator gradually approach the distribution of the real data through continuous loop feedback and adjustment. The stepwise iteration process can improve the quality of the generated data until the discriminator cannot distinguish between the generated data and the real data, thereby ensuring the accuracy and reliability of data interpolation.

[0046] In one embodiment, if Figure 5 and Figure 6 As shown, step S20, that is, according to the KSTV-DBSCAN algorithm that introduces threshold rules and adaptive mechanisms, cluster analysis is performed on the time series trajectory data without missing values ​​to identify a list of valid stop points for transport vehicles, including the following steps S201-S205:

[0047] S201. Based on the DBSCAN algorithm, the threshold rule and adaptive mechanism are introduced to form the KSTV-DBSCAN algorithm; understandably, the KSTV-DBSCAN algorithm is an extension of the classic DBSCAN algorithm, which aims to more accurately identify the stop points of dangerous goods transport vehicles by combining multi-dimensional information such as space, time, speed and threshold. This method is particularly suitable for processing vehicle trajectory data in complex dynamic environments, while the classic DBSCAN algorithm that relies solely on spatial position may lead to the misjudgment of a large number of false stop points.

[0048] Specifically, threshold rules are introduced based on the DBSCAN algorithm. The thresholds can be set according to requirements. For example, the time threshold limits the minimum time that a vehicle stays at the same or near location. The distance threshold, in addition to the spatial density, also limits the maximum distance that a vehicle moves in the same time period to confirm whether the vehicle is indeed in a parked state. The speed threshold directly identifies the parked state by the vehicle speed, which is usually set to a very low value or zero.

[0049] Based on the DBSCAN algorithm, an adaptive mechanism is introduced to dynamically adjust parameters.

[0050] S202, determining the optimal hyperparameter combination of the DBSCAN algorithm using an adaptive mechanism;

[0051] In one embodiment, if Figure 6 As shown, the step S202, how to use the adaptive mechanism to determine the optimal hyperparameter combination of the DBSCAN algorithm, includes the following steps S2021-S2024:

[0052] S2021. Use the silhouette coefficient as an optimization indicator to measure the closeness and separation of clusters. Understandably, the value range of the silhouette coefficient is from -1 to 1. The closer the silhouette coefficient is to 1, the better the clustering effect, that is, the similarity of samples within a class is high and the similarity of samples between classes is low. Through the silhouette coefficient, we can quantify the clustering effect of the DBSCAN algorithm under different hyperparameter combinations.

[0053] S2022. Identify the domain radius eps and the minimum number of points within the domain radius MinPTs in the DBSCAN algorithm, determine that the domain radius eps and the minimum number of points within the domain radius MinPTs are the optimal hyperparameter combination, and determine their reasonable value range to limit the search space; understandably, since the value range of the domain radius eps and the minimum number of points within the domain radius MinPTs may be very wide, it is inefficient to directly search all combinations, and it is necessary to determine the reasonable value range of these two parameters based on data characteristics and prior knowledge. By limiting the search space, not only can unnecessary calculations be reduced and search efficiency be improved, but also extreme or unreasonable parameter values ​​can be avoided.

[0054] S2023. Use Bayesian optimization to dynamically search for optimal hyperparameter combinations. Understandably, using Bayesian optimization can intelligently balance the relationship between exploration (finding new hyperparameter combinations) and exploitation (fine searching in known high-performance areas), thereby efficiently finding optimal or near-optimal hyperparameter combinations. In addition, the Bayesian optimization method takes into account previous search results, so the search becomes more and more accurate as the number of iterations increases.

[0055] S2024. Through multiple iterations, find the optimal hyperparameter combination that makes the silhouette coefficient the highest. Understandably, after determining the optimization index, the value range of the hyperparameter and the optimization method, it is necessary to continuously adjust the hyperparameter combination through multiple iterations until the set of parameters with the highest silhouette coefficient is found. Through multiple iterations, the search range can be gradually narrowed, and finally the hyperparameter combination that makes the clustering effect optimal can be found. This can not only improve the accuracy and stability of clustering, but also reduce the influence of human intervention and subjective factors.

[0056] In this embodiment, the adaptive mechanism can more accurately identify dense and sparse areas in the data by dynamically adjusting the values ​​of the domain radius eps and the minimum number of points MinPTs within the domain radius, thereby generating clustering results that are more in line with the natural distribution of the data. Since different data sets have different feature distributions and density changes, fixed hyperparameter settings are often difficult to apply to all situations. The adaptive mechanism can automatically adjust parameters according to the characteristics of different data sets, enhancing the flexibility and robustness of the KSTV-DBSCAN algorithm when processing different types of data. The adaptive mechanism greatly reduces the need for manual intervention by automatically finding the optimal parameter combination, and can also be used to achieve multi-scale clustering, that is, analyzing the clustering structure of data sets at different scales.

[0057] S203, input the optimal hyperparameter combination into the KSTV-DBSCAN algorithm, perform spatial clustering analysis on the time series trajectory data without missing values, and determine the number of clusters K, each cluster represents a potential stop point area; understandably, the optimal hyperparameter combination is applied to the KSTV-DBSCAN algorithm, spatial clustering analysis is performed on the time series trajectory data, the number of clusters is automatically determined based on the density distribution of the data itself, and potential stop point areas are effectively identified. In the time series trajectory data, the stop point area is usually manifested as a high-density aggregation area of ​​trajectory points. By identifying these high-density areas, KSTV-DBSCAN can automatically divide the trajectory data into different clusters, each cluster corresponding to a potential stop point area.

[0058] S204. According to the requirements of the number of trajectory points and the surface conditions, each cluster is evaluated, and the clusters that meet the conditions are determined as the primary stop point set, and the clusters that do not meet the conditions are determined as noise points; understandably, according to the spatial distribution characteristics of the trajectory points, the KSTV-DBSCAN algorithm is used to group the trajectory points into multiple clusters. Based on the number of trajectory points, that is, the number of trajectory points contained in the cluster must meet certain threshold requirements to ensure that the cluster represents a sufficiently significant location. Based on the surface conditions, this usually means that the spatial range or shape of the cluster must meet certain specific geographical or logical constraints, such as whether it is located in a specific area, whether the shape is compact, etc., to further verify its rationality as a stop point. The clusters that meet the conditions are determined as the primary stop point set. These clusters represent candidate stop locations, which are of great significance for subsequent analysis or application.

[0059] S205, set the rule conditions of time threshold, distance threshold and speed threshold to reflect the behavior characteristics of the transport vehicle at the primary stop point, based on the rule conditions, filter out the pseudo stop points from the primary stop point set, and output a valid stop point list containing only the valid stop point set. Understandably, by setting the time threshold, distance threshold and speed threshold as rule conditions, the pseudo stop points and valid stop points in the primary stop point set can be effectively distinguished and filtered out. The rule conditions are intended to capture the behavior characteristics of the transport vehicle during the stop process, such as long-term stillness (determined by the time threshold), very small or no displacement (determined by the distance threshold) and a speed state close to zero (determined by the speed threshold). Through the comprehensive application of these rules, the system can automatically identify and eliminate pseudo stop points generated due to short stays, data anomalies or erroneous records, thereby ensuring that the valid stop point list output in the end only contains stops that are of practical significance to transportation management decisions.

[0060] In one embodiment, the step S30, i.e., performing behavioral feature analysis and duration analysis on the valid stop point list to obtain classified stop information of the transport vehicle at the valid stop points, wherein the classified stop information includes the stop position information of all valid stop points; comprises the following steps S301-S303:

[0061] S301. Obtain the longitude and latitude information of each cluster in the valid stop point list, and use geocoding technology to convert the longitude and latitude information of each cluster into stop location information, wherein the stop location information includes the stay time of the transport vehicle, the stop location POI, and the characteristics and activity type of the stop location; understandably. By using geocoding / reverse geocoding services such as Amap API, users can easily convert address data into a location on a map, or convert a location on a map into specific address information. Of course, it is not limited to the above method. Among them, the stop location information includes the stay time of the transport vehicle at the location, the type of the location as a point of interest POI (such as a gas station, factory, warehouse, bus stop, etc.), and the characteristics of the location (such as a commercial area, residential area, pickup area, unloading area, dining area) and possible activity types (such as loading and unloading, rest, passengers getting on and off, etc.).

[0062] S302, classifying the effective stop point list into types according to the stop position information, analyzing the behavioral characteristics of each type of effective stop point, outputting a characteristic description of each type of effective stop point, wherein the characteristic description includes the stop purpose and the duration of stay, and generating a list of effective stop points with type labels; understandably, classifying the effective stop point list into types according to the stop position information, and the classification basis may be factors such as the type of POI, the characteristics of the location, and the type of activity, and analyzing the unique behavioral characteristics of each type of effective stop point, which may include the main purpose of the stop (such as loading and unloading, passenger service), the regularity of the stop time (such as peak hours), the stop frequency, etc.; generating a characteristic description for each type of stop point, including the stop purpose, the average / longest / shortest duration of stay, etc., and generating a list of effective stop points with type labels, for subsequent analysis and management.

[0063] S303, based on each type of stop in the list of valid stops with type labels, a stay time analysis is performed to reflect the stay time of the transport vehicle in different types of activities, and an analysis report of the stay time of each type of stop is output. Understandably, the stay time analysis mainly includes calculating the average stay time, stay time distribution (such as the proportion of different time intervals), the longest and shortest stay time, etc. of each type of stop, and outputting a detailed analysis report of the stay time of each type of stop. The analysis report can help managers understand the time allocation of transport vehicles in different types of activities.

[0064] In the above embodiment, in-depth feature analysis and duration analysis of effective stop points facilitates understanding of the stop behavior of transport vehicles at different locations and the types of activities behind them, which helps to better understand the operation mode of transport vehicles, optimize transportation plans, improve transportation efficiency, and provide data support for supervision. Through this analysis, atypical or abnormal stop behaviors can be identified, providing a basis for further investigation or adjustment.

[0065] In one embodiment, the step S40, i.e., correlating and comparing the classified stop information of the same transport vehicle with the pre-filled waybill information, and generating an enhanced waybill according to the consistency between the classified stop information of the time series trajectory data and the pre-filled waybill information, includes the following steps S401-S404:

[0066] S401. Based on the classified stop information of valid stops, according to the stay time of the transport vehicle and the stop location POI, exclude the types of stop location POI that do not meet the characteristics of dangerous goods transport vehicles, screen out valid stops that meet the characteristics of dangerous goods transport vehicles, and determine them as potential starting and ending points; understandably, considering the particularity of dangerous goods transportation, the system needs to exclude those locations that are not suitable for dangerous goods to stop, such as service areas, highways, gas stations (unless specifically designated as dangerous goods), residential areas, hotels, restaurants, etc., because these places may increase the risk of accidents or fail to meet safety regulations. The remaining stops are regarded as valid stops that meet the characteristics of dangerous goods transportation, and these points are further determined as potential starting and ending points.

[0067] S402. Use reverse geocoding technology to convert the origin and destination information in the pre-filled waybill information into longitude and latitude information; understandably, it can be directly compared with the GPS track data to improve the accuracy of the match.

[0068] S403, comparing the classified stop information and the pre-filled waybill information of the same transport vehicle, verifying the consistency of the time series trajectory data and the pre-filled waybill information, determining the consistency of the start and end points in the classified stop information and the start and end points in the pre-filled waybill information, and adjusting and updating the pre-filled waybill information;

[0069] In one embodiment, the step S403 includes the following steps S4031-S4032:

[0070] S4031. Match the classified stop information with the pre-filled waybill information according to the transport vehicle number;

[0071] It can be understood that matching based on the transport vehicle number can ensure that the same transport vehicle is being compared.

[0072] S4032. Compare the distance and time correlation between the potential starting and ending points and the starting and ending points of the pre-filled waybill information; if the distance difference and time range of the two are consistent, it is regarded as a valid waybill and recorded.

[0073] If the first waybill that is in the time series trajectory data but not in the pre-filled waybill information is filtered out, it means that the first waybill reveals a new transportation activity, and the first waybill is added to the pre-filled waybill information as a quasi-new waybill; understandably, this indicates that the waybill reveals a new, previously unplanned transportation activity, and this new waybill is added to the pre-filled waybill information as a quasi-new waybill for subsequent management and monitoring.

[0074] If the second waybill is found in the pre-filled waybill information but not in the time series trajectory data, it indicates that the second waybill does not match the actual transportation activity. The second waybill is removed from the pre-filled waybill information to complete the update of the pre-filled waybill information. Understandably, this indicates that the waybill does not match the actual transportation activity, which may be caused by a change in plan or an erroneous operation. This waybill is removed from the pre-filled waybill information to ensure the accuracy and real-time nature of the data.

[0075] In this embodiment, by real-time verification and adjustment of the waybill information of the transport vehicle, the accuracy and efficiency of logistics transportation are ensured, and potential transportation problems such as waybill omissions, misoperation or transportation delays can be discovered and resolved in a timely manner.

[0076] S404, integrate the valid waybill and the updated pre-filled waybill information to generate an enhanced waybill. Understandably, based on the trajectory data, extract the specific location information of the departure and arrival of the transport vehicle, including the longitude and latitude coordinates accurate to the 4th decimal place, which will replace or supplement the starting and ending point information in the original waybill. Use the timestamp in the trajectory data to record the specific time of cargo transportation, including the departure time, the time of key nodes on the way, and the arrival time, accurate to the second level. Analyze the trajectory data, identify the vehicle's stops on the way, and record the longitude and latitude coordinates and the stay time of each stop, also accurate to the longitude and latitude and second level. Integrate the above processed information into the new waybill to form an enhanced waybill. The waybill contains more detailed and accurate starting and ending point information, specific time of cargo transportation, and vehicle stop information.

[0077] In one embodiment, if Figure 2 As shown, the step S404 includes the following steps S4041-S4042:

[0078] S4041. Record the valid waybill and the updated pre-filled waybill information, input them into the corresponding system, and output the relevant report of the enhanced waybill, which includes the latitude and longitude coordinates of the starting and ending points of the transport vehicle, the time of dangerous goods transportation, the stop duration and the stop location POI;

[0079] S4042. Evaluate and manage the transportation progress of transport vehicles based on the enhanced waybill.

[0080] In this embodiment, the transportation progress of the transport vehicle is evaluated based on the enhanced waybill, and the evaluation can be performed from the aspects of progress comparison, performance evaluation and data visualization. For example, in terms of transportation progress, the actual transportation progress is compared with the preset plan to analyze the cause of the delay. In terms of performance evaluation, the performance of transport vehicles, carriers and logistics routes is regularly evaluated, including indicators such as punctuality rate, cargo damage rate, and cost-effectiveness. In terms of data visualization, charts, dashboards and other tools are used to visualize data such as transportation progress and performance indicators, so that management can quickly grasp the overall situation.

[0081] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0082] In one embodiment, a waybill enhancement system based on the track data of a dangerous goods transport vehicle is provided, and the waybill enhancement system based on the track data of a dangerous goods transport vehicle corresponds one-to-one to the waybill enhancement method based on the track data of a dangerous goods transport vehicle in the above embodiment. As shown in 7, the waybill enhancement system based on the track data of a dangerous goods transport vehicle includes:

[0083] The interpolation processing module 100 is used to obtain high-dimensional trajectory data of the transport vehicle, and perform interpolation processing on the high-dimensional trajectory data to output time series trajectory data without missing values;

[0084] The cluster analysis module 200 is used to perform cluster analysis on the time series trajectory data without missing values ​​according to the KSTV-DBSCAN algorithm with the introduction of threshold rules and adaptive mechanism, and identify a list of valid stop points for transport vehicles;

[0085] The behavior characteristic and duration analysis module 300 is used to perform behavior characteristic analysis and duration analysis on the effective stop point list, and obtain classified stop information of the transport vehicle at the effective stop points, wherein the classified stop information includes the stop position information of all effective stop points;

[0086] The generation module 400 is used to associate and compare the classified stop information of the same transport vehicle with the pre-filled waybill information, and generate an enhanced waybill based on the consistency between the classified stop information of the time series trajectory data and the pre-filled waybill information.

[0087] In the waybill enhancement system based on the trajectory data of dangerous goods transport vehicles in the above-mentioned embodiment of the present invention, the high-dimensional trajectory data of the dangerous goods transport vehicles is interpolated and improved to ensure the integrity and accuracy of the data. Based on the improved high-dimensional trajectory data, the KSTV-DBSCAN algorithm that introduces threshold rules and adaptive mechanisms is used to automatically extract the actual stop points in the transportation process and identify the valid stop points. By performing behavioral feature analysis and duration analysis on the valid stop points, the stop position information of the valid stop points is obtained. Furthermore, the stop position information is intelligently matched and verified with the pre-filled waybill information, errors are automatically corrected and missing records are supplemented, and an accurate and efficient enhanced waybill is generated. The enhanced waybill not only improves the transparency and controllability of the dangerous goods transportation process, but also provides strong technical support for regulatory authorities, thereby ensuring the safety and compliance of dangerous goods transportation.

[0088] For the specific limitations of the waybill enhancement system based on the trajectory data of dangerous goods transport vehicles, please refer to the limitations of the waybill enhancement method based on the trajectory data of dangerous goods transport vehicles mentioned above, which will not be repeated here. Each module in the above-mentioned waybill enhancement system based on the trajectory data of dangerous goods transport vehicles can be implemented in whole or in part through software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0089] In one embodiment, a controller is provided, comprising a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, wherein when the processor executes the computer-readable instructions, the method for enhancing the waybill based on the trajectory data of a dangerous goods transport vehicle is implemented. Figure 8 As shown, the controller includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the controller is used to provide computing and control capabilities. The memory includes a readable storage medium and an internal memory. The readable storage medium stores an operating system, computer-readable instructions and a database. The internal memory provides an environment for the operation of the operating system and computer-readable instructions in the readable storage medium. When the computer-readable instructions are executed by the processor, a suspension height adjustment method is implemented. The readable storage medium provided in this embodiment includes a non-volatile readable storage medium and a volatile readable storage medium.

[0090] In one embodiment, a computer-readable storage medium is provided, which stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the waybill enhancement method based on the trajectory data of a dangerous goods transport vehicle in the above embodiment is implemented. To avoid repetition, it is not repeated here. The computer-readable storage medium can be non-volatile or volatile.

[0091] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing related hardware through computer-readable instructions, and the computer-readable instructions can be stored in a non-volatile readable storage medium or a volatile readable storage medium. When the computer-readable instructions are executed, they may include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0092] The above is only an embodiment of the waybill enhancement method and system based on the trajectory data of dangerous goods transport vehicles of the present invention, and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A waybill enhancement method based on dangerous goods transport vehicle trajectory data, characterized in that: include: S10, obtaining high-dimensional trajectory data of the transport vehicle, and interpolating the high-dimensional trajectory data to output time series trajectory data without missing values; S20, performing cluster analysis on the time series trajectory data without missing values ​​according to the KSTV-DBSCAN algorithm with the introduction of threshold rules and adaptive mechanism, and identifying a list of valid stop points for transport vehicles; S30, performing behavior feature analysis and duration analysis on the valid stop point list to obtain classified stop information of the transport vehicle at the valid stop points, wherein the classified stop information includes the stop position information of all the valid stop points; S40, associating and comparing the classified stop information of the same transport vehicle with the pre-filled waybill information, and generating an enhanced waybill according to the consistency between the classified stop information of the time series trajectory data and the pre-filled waybill information; The following methods are included: S401, based on the classified docking information of valid docking points, according to the stay time of the transport vehicle and the docking location POI, exclude the types of docking location POI that do not meet the characteristics of dangerous goods transport vehicles, screen out valid docking points that meet the characteristics of dangerous goods transport vehicles, and determine them as potential starting and ending points; S402, using reverse geocoding technology to convert the origin and destination information in the pre-filled waybill information into longitude and latitude information; S403, comparing the classified stop information and the pre-filled waybill information of the same transport vehicle, verifying the consistency of the time series trajectory data and the pre-filled waybill information, determining the consistency of the start and end points in the classified stop information and the start and end points in the pre-filled waybill information, and adjusting and updating the pre-filled waybill information; S404: Integrate the valid waybill and the updated pre-filled waybill information to generate an enhanced waybill.

2. The waybill enhancement method based on dangerous goods transport vehicle trajectory data according to claim 1 is characterized in that: The step S10 further includes the following steps: S101, setting a preset time interval to obtain high-dimensional trajectory data of the transport vehicle, wherein the high-dimensional trajectory data includes the location ID, second-level transport timestamp, longitude and latitude, driving speed, and driving direction angle of the transport vehicle; S102, using Kalman filtering to perform denoising on high-dimensional trajectory data; S103, using a deep learning network to interpolate the high-dimensional trajectory data after denoising, and output time series trajectory data without missing values; S104, the discriminator receives the original high-dimensional time series data and the reconstructed time series trajectory data without missing values ​​generated by the generator. Under the same condition vector, the discriminator distinguishes the filled data and the real data in the original high-dimensional time series data, and optimizes the generator and the discriminator through error back propagation; S105. Optimize the time series trajectory data without missing values ​​generated by the generator through step-by-step iteration until the discriminator cannot distinguish between the filled part and the real data part, and then stop the iteration.

3. The method for enhancing waybills based on trajectory data of dangerous goods transport vehicles according to claim 2, characterized in that: The step S103 further includes the following steps: S1031, inputting the denoised high-dimensional trajectory data into a generator, wherein the generator includes a first-level FC fully connected layer, a first-level gated time unit GRU layer, a multi-head self-attention mechanism layer, a second-level gated time unit GRU layer, and a second-level FC fully connected layer connected in sequence; The first-level FC fully connected layer is used to perform feature mapping on the original high-dimensional time series data and random matrices containing missing values ​​in the high-dimensional trajectory data, and generate high-dimensional features; S1032, using the first-level gated time unit GRU layer to capture the dependencies in the original high-dimensional time series data, and process and generate coherent and logical sequence data; S1033, using a multi-head self-attention mechanism layer to learn the relationship between long-range features in the original high-dimensional time series data, each head learns a different feature representation and generates global information, and converts the global information into a weight matrix; S1034, using the second-level gated time unit GRU layer to receive global information, and refine and adjust the sequence data; S1035. Use the second-level FC fully connected layer to map the high-dimensional features generated by the first-level FC fully connected layer back to the dimension of the target data space, and generate reconstructed time series trajectory data without missing values.

4. The method for enhancing waybills based on trajectory data of dangerous goods transport vehicles according to claim 2, characterized in that: The step S20 further comprises the following steps: S201. Based on the DBSCAN algorithm, the threshold rule and adaptive mechanism are introduced to form the KSTV-DBSCAN algorithm; S202, determining the optimal hyperparameter combination of the DBSCAN algorithm using an adaptive mechanism; S203, inputting the optimal hyperparameter combination into the KSTV-DBSCAN algorithm, performing spatial clustering analysis on the time series trajectory data without missing values, and determining the number of clusters K, each cluster representing a potential stop point area; S204, evaluating each cluster according to the requirement of the number of trajectory points and the area condition, determining the clusters that meet the conditions as the primary stop point set, and determining the clusters that do not meet the conditions as noise points; S205, setting rule conditions of time threshold, distance threshold and speed threshold to reflect the behavior characteristics of the transport vehicle at the primary stop point, filtering out pseudo stop points from the primary stop point set based on the rule conditions, and outputting a valid stop point list containing only the valid stop point set.

5. The method for enhancing waybills based on trajectory data of dangerous goods transport vehicles according to claim 4, characterized in that: The step S202 further includes the following steps: S2021. Use the silhouette coefficient as an optimization indicator to measure the compactness and separation of clusters; S2022. Identify the domain radius eps and the minimum number of points MinPTs within the domain radius in the DBSCAN algorithm, determine that the domain radius eps and the minimum number of points MinPTs within the domain radius are the optimal hyperparameter combination, and determine a reasonable value range thereof to limit the search space; S2023. Use Bayesian optimization method to dynamically search for the optimal hyperparameter combination; S2024. Through multiple iterations, find the optimal hyperparameter combination that gives the highest silhouette coefficient.

6. The method for enhancing waybills based on trajectory data of dangerous goods transport vehicles according to claim 3, characterized in that: The step S30 further includes the following steps: S301, obtaining the longitude and latitude information of each cluster in the valid stop point list, and converting the longitude and latitude information of each cluster into stop location information using geocoding technology, wherein the stop location information includes the stay time of the transport vehicle, the stop location POI, and the characteristics and activity type of the stop location; S302, classifying the valid stop point list into types according to the stop position information, analyzing the behavior characteristics of each type of valid stop point, outputting a characteristic description of each type of valid stop point, the characteristic description including the stop purpose and the stay duration, and generating a valid stop point list with a type label; S303: Based on each type of stop in the valid stop list with type labels, a stay time analysis is performed to reflect the stay time of the transport vehicle in different types of activities, and an analysis report on the stay time of each type of stop is output.

7. The method for enhancing waybills based on trajectory data of dangerous goods transport vehicles according to claim 2, characterized in that: The step S403 further includes the following steps: S4031. Match the classified stop information with the pre-filled waybill information according to the transport vehicle number; S4032, compare the potential starting and ending points with the starting and ending points of the pre-filled waybill information in terms of distance and time; if the distance difference and time range of the two are consistent, it is considered as a valid waybill and recorded; If the first waybill in the time series trajectory data but not in the pre-filled waybill information is screened out, it means that the first waybill reveals a new transport activity, and the first waybill is added to the pre-filled waybill information as a quasi-new waybill; If the second waybill is selected from the pre-filled waybill information but not in the time series trajectory data, it indicates that the second waybill does not match the actual transportation activity. The second waybill is removed from the pre-filled waybill information, and the update of the pre-filled waybill information is completed.

8. The method for enhancing waybills based on trajectory data of dangerous goods transport vehicles according to claim 2, characterized in that: The step S404 further includes the following steps: S4041. Record the valid waybill and the updated pre-filled waybill information, input them into the corresponding system, and output the relevant report of the enhanced waybill, which includes the latitude and longitude coordinates of the starting and ending points of the transport vehicle, the time of dangerous goods transportation, the stop duration and the stop location POI; S4042. Evaluate and manage the transportation progress of transport vehicles based on the enhanced waybill.

9. A waybill enhancement system based on dangerous goods transport vehicle trajectory data, characterized in that: include: An interpolation processing module is used to obtain high-dimensional trajectory data of transport vehicles, perform interpolation processing on the high-dimensional trajectory data, and output time series trajectory data without missing values; The cluster analysis module is used to perform cluster analysis on the time series trajectory data without missing values ​​based on the KSTV-DBSCAN algorithm with threshold rules and adaptive mechanism, and identify the effective stop list of transport vehicles; The behavior characteristic and duration analysis module is used to perform behavior characteristic analysis and duration analysis on the effective stop point list, and obtain the classified stop information of the transport vehicle at the effective stop points, wherein the classified stop information includes the stop position information of all effective stop points; The generation module is used to associate and compare the classified stop information of the same transport vehicle with the pre-filled waybill information, and generate an enhanced waybill according to the consistency between the classified stop information of the time series trajectory data and the pre-filled waybill information; the generation module includes the following methods: S401, based on the classified docking information of valid docking points, according to the stay time of the transport vehicle and the docking location POI, exclude the types of docking location POI that do not meet the characteristics of dangerous goods transport vehicles, screen out valid docking points that meet the characteristics of dangerous goods transport vehicles, and determine them as potential starting and ending points; S402, using reverse geocoding technology to convert the origin and destination information in the pre-filled waybill information into longitude and latitude information; S403, comparing the classified stop information and the pre-filled waybill information of the same transport vehicle, verifying the consistency of the time series trajectory data and the pre-filled waybill information, determining the consistency of the start and end points in the classified stop information and the start and end points in the pre-filled waybill information, and adjusting and updating the pre-filled waybill information; S404: Integrate the valid waybill and the updated pre-filled waybill information to generate an enhanced waybill.

Images