Digital intelligent trajectory prediction method and system based on logistics transportation order tracking

Abstract

The invention relates to a digital intelligent trajectory prediction method and system based on logistics transportation order tracking. The method comprises the steps of collecting multi-source transportation data including vehicle GPS data, AI telephone interaction data, ETC passing record data and freight station transfer operation node data; performing space-time alignment, data cleaning and map matching based on road network information on the multi-source transportation data; utilizing the processed multi-source data to generate a continuous digital intelligent track through an interpolation algorithm and a path deduction method; calculating the latest predicted position of the vehicle through a prediction algorithm model based on the digital intelligent track in combination with real-time traffic information and a historical driving mode; residual paths, real-time road conditions, weather factors and driver behavior modes are comprehensively considered, and the predicted arrival time is calculated through a dynamic ETA model. According to the invention, the GPS signal blind area problem is effectively solved, the whole-course blind-area-free tracking and high-precision arrival time prediction are realized, and the transparency and intelligent management level of logistics transportation are greatly improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent logistics, in particular to a digital intelligent track prediction method and system based on logistics transportation order tracking, and particularly relates to a logistics transportation digital intelligent track tracking and predicted time of arrival prediction method and system based on multi-source data fusion. BACKGROUND

[0002] With the vigorous development of global e-commerce and the increasing refinement of supply chain management, the modern logistics industry is undergoing unprecedented changes. Enterprises have made extremely high demands on the transparency, real-time and precise control of the logistics transportation process. Real-time understanding of the in-transit location of goods and predicting their accurate time of arrival are of great significance for optimizing warehouse operation plans, improving customer satisfaction and reducing operating costs.

[0003] Currently, the tracking of logistics transportation vehicles in the industry mainly relies on Global Positioning System (GPS) technology. By installing a GPS terminal on the vehicle, the geographical coordinates of the vehicle can be obtained regularly, thereby depicting the driving track of the vehicle. However, this technology solution relying on a single data source has many inherent defects:

[0004] Firstly, GPS signals are extremely susceptible to physical environmental interference. When passing through tunnels, under viaducts, underground parking lots, dense urban areas or remote mountainous areas, GPS signals often attenuate or even completely lose, forming a tracking blind area. The vehicle is in a "lost" state during this period, and its actual driving path and state cannot be perceived until the satellite signal is reacquired. This blind area not only causes the track to be interrupted, but more importantly, it makes the system unable to effectively monitor and manage the vehicle at critical moments.

[0005] Secondly, GPS technology essentially provides a record of historical positions. It tells where the vehicle has been, not where it is now or will be in the future. For logistics management, simply knowing the historical position is far from enough, decision-making relies more on accurate judgment of the current state and precise prediction of future events, especially the predicted time of arrival.

[0006] Thirdly, relying solely on GPS data, its reliability and accuracy are limited. GPS data may be incorrect due to device failure, signal drift, etc., and the system lacks other data sources for cross-validation and correction, resulting in false positives or distorted data.

[0007] The logistics industry is undergoing a transformation driven by technological innovation in 2025. The integration of AI prediction and blockchain storage technology has become a core solution for optimizing logistics track management. AI prediction technology integrates massive data to build accurate prediction models, enabling demand prediction, resource scheduling, dynamic path optimization, risk warning, and emergency response. Blockchain storage technology uses distributed ledgers and tamper-proof features to provide full-traceability for logistics tracks, and is applied in areas such as electronic waybill storage, responsibility tracing and dispute resolution, and automatic execution of smart contracts.

[0008] To address the above problems, the existing technology has also made some improvements. For example, in the existing technology, a Chinese patent with publication number CN109188978A discloses an intelligent logistics vehicle monitoring system based on GPS, which includes a monitoring coordination system, a logistics vehicle, a location acquisition device, and a path navigation system. The monitoring coordination system is used for coordinated control of the logistics vehicle, the location acquisition device, and the path navigation system, so that the goods in the cargo list are transported from the goods location to the sorting center. The location acquisition device is used to determine the corresponding goods location of the goods in the cargo list in the inventory data according to the monitoring coordination system, and to indicate the specific location of the goods. The path navigation system is used to plan the path according to the location of the logistics vehicle, the specific location of the goods, and the location of the sorting center, and to find the shortest path. The logistics vehicle with GPS provides its own location information to the path navigation system, and according to the planned shortest path, transports the goods indicated by the location acquisition device from the goods location to the sorting center. The focus is on optimizing the transmission and processing of GPS data, but it only relies on GPS data and does not fundamentally solve the signal loss problem.

[0009] For example, Chinese Patent No. CN119559784A discloses a multi-source data fusion signal intersection vehicle trajectory reconstruction method, which obtains intelligent connected vehicle (CAV) and its perceived surrounding vehicle motion data, signal phase data, and speed camera data, and determines the signal cycle of the vehicle passing through the intersection according to these data. The trajectory of a regular vehicle (RV) passing through the intersection with the same cycle as the known trajectory vehicle in front of it can be reconstructed from back to front in turn using an intelligent driver model (IDM) based trajectory reconstruction optimization model. When there is a known trajectory vehicle with the same cycle passing through the intersection in front of the RV, the trajectory of the vehicle is reconstructed from front to back in turn using an IDM and grid search based trajectory reconstruction method. When there is no known trajectory vehicle with the same cycle passing through the intersection in front of the RV, a virtual vehicle is constructed based on the signal phase data to replace the CAV. The trajectory reconstruction results of the same vehicle are fused using a history weight based trajectory fusion algorithm. In its scheme, the fusion of multi-source data is not deep enough, especially without involving how to use interactive data, and its core goal is to reconstruct the historical trajectory. No effective solution is proposed for real-time vehicle position prediction and accurate and dynamic ETA prediction. There is a lack of consideration of complex factors such as weather and driver behavior, and no accurate ETA prediction is involved. Moreover, the AI voice interactive data is not utilized.

[0010] However, the existing technology still does not effectively solve the GPS signal blind area problem, lacks deep fusion of multi-source data, especially in the application of AI phone interactive data, confidence weight distribution mechanism and dynamic ETA prediction model. Therefore, there is an urgent need in the field for an intelligent logistics tracking method and system that can fuse multi-source data, realize blind area-free tracking, and accurately predict the ETA of various factors. SUMMARY

[0011] In view of the deficiencies of the prior art, the purpose of the present application is to provide a digital trajectory prediction method and system based on logistics transportation order tracking, to solve the tracking interruption problem caused by GPS signal blind area, to accurately predict the current position of the vehicle, and to dynamically and accurately predict the ETA of the transportation vehicle.

[0012] The above invention purpose of the present application is realized by the following technical scheme:

[0013] The present application provides a digital trajectory prediction method based on logistics transportation order tracking, comprising the following steps:

[0014] Step S1, multi-source data acquisition: collect multi-source transportation data including vehicle GPS data, AI phone interactive data, ETC pass record data and freight station transfer operation node data;

[0015] Step S2, Data Fusion Processing: Spatiotemporal alignment and data cleaning are performed on the multi-source transportation data, and map matching is performed based on road network information;

[0016] Step S3, Trajectory Reconstruction: Using the processed multi-source data, a continuous digital trajectory is generated through interpolation algorithms and path extrapolation methods;

[0017] Step S4, Location Prediction: Based on the digital trajectory, combined with real-time traffic information and historical driving patterns, the latest predicted location of the vehicle is calculated through a prediction algorithm model;

[0018] Step S5, Arrival Time Prediction: Taking into account the remaining route, real-time traffic conditions, weather factors, and driver behavior patterns, the estimated arrival time is calculated using a dynamic ETA model.

[0019] The core idea of ​​this method is to no longer rely solely on GPS data, but to innovatively integrate multiple data sources with different characteristics and complement each other. Through a set of intelligent data processing and algorithm models, a continuous, reliable, and intelligent "digital trajectory" is generated, and prediction function is realized on this basis.

[0020] According to one embodiment of the present invention, in step S1, the step of collecting AI telephone interaction data includes: the system automatically making an AI call to the driver at regular intervals or triggered by the system, obtaining the driver's voice response through voice recognition technology, extracting key location information from the voice response through natural language processing technology, and associating the extracted location information with spatiotemporal coordinates to form location anchors.

[0021] According to one embodiment of the present invention, in step S2, different confidence weights are set for data from different sources, wherein the confidence weight of ETC passage record data is the highest, followed by AI telephone interaction data, and the weight of GPS data is dynamically adjusted according to signal strength, and data fusion is performed based on the confidence weights.

[0022] According to one embodiment of the present invention, in step S3, when the GPS signal is lost, a map matching algorithm based on a hidden Markov model is used to deduce the most likely driving path of the vehicle in order to fill the trajectory gap, combining the last known location, the vehicle's driving direction and the road network topology constraints.

[0023] According to one embodiment of the present invention, the dynamic ETA model in step S5 is expressed as: ETA = T_current + Σ + ΣT_stop, where ETA is the estimated arrival time, T_current is the current time, D_i is the distance of the i-th segment, V_i is the predicted speed of the i-th segment based on real-time traffic conditions, historical average speed and weather factors, and T_stop is the intermediate stop time predicted based on historical data and current order information.

[0024] According to one embodiment of the present invention, the predicted speed V_i is obtained through a machine learning model, the input features of which include real-time traffic flow data, historical average vehicle speed of the same road segment during the same period, current weather conditions, and date type.

[0025] Another objective of this invention is to provide a predictive system that implements the above-mentioned method, providing logistics companies with a fully visualized and predictable smart logistics management solution.

[0026] Specifically, this invention provides a digital trajectory prediction system based on logistics transportation order tracking that implements the method described above, comprising:

[0027] The data acquisition module is configured to collect transportation data from multiple data sources;

[0028] The data processing module is configured to perform cleaning, spatiotemporal alignment, confidence weighting, and fusion of multi-source data.

[0029] The trajectory generation module is configured to generate continuous digital trajectories based on the processed data.

[0030] The prediction module is configured to predict the vehicle's real-time location and estimated arrival time.

[0031] The application interface module is configured to provide users with data query, trajectory display, and early warning services.

[0032] According to one embodiment of the present invention, the data acquisition module includes an AI telephone interaction submodule, the AI ​​telephone interaction submodule comprising:

[0033] The call unit is used to automatically initiate a call to the driver according to a preset strategy;

[0034] A voice recognition unit is used to convert the driver's voice responses into text;

[0035] The Natural Language Processing Unit (NLP) is used to identify and extract geolocation keywords from text;

[0036] The data association unit is used to bind the extracted location information with the current time and vehicle information to generate auxiliary positioning data.

[0037] According to one embodiment of the present invention, the prediction module integrates a dynamic ETA prediction submodule, which accesses third-party traffic big data services and weather data services, and has a built-in driver behavior profile database for fine-tuning the predicted arrival time.

[0038] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described above.

[0039] In summary, compared with the prior art, the present invention has at least one of the following beneficial technical effects:

[0040] 1. This invention discloses a digital trajectory prediction method and system based on logistics transportation order tracking, belonging to the field of intelligent logistics technology. The method includes: a multi-source data acquisition step, used to collect multi-source transportation data including vehicle GPS data, AI telephone interaction data, ETC passage record data, and freight station transfer operation node data; a data fusion processing step, used to perform spatiotemporal alignment, data cleaning, and map matching based on road network information on the multi-source transportation data; a trajectory reconstruction step, used to generate a continuous digital trajectory using the processed multi-source data through interpolation algorithms and path extrapolation methods; a location prediction step, used to calculate the latest predicted location of the vehicle based on the digital trajectory, combined with real-time traffic information and historical driving patterns, through a prediction algorithm model; and an arrival time prediction step, used to calculate the estimated arrival time by comprehensively considering the remaining path, real-time road conditions, weather factors, and driver behavior patterns through a dynamic ETA model. This invention effectively solves the GPS signal blind spot problem, achieving blind-spot-free tracking and high-precision arrival time prediction throughout the entire process, greatly improving the transparency and intelligent management level of logistics transportation.

[0041] 2. End-to-end tracking without blind spots: By integrating auxiliary data sources such as AI telephone interaction and ETC records, and effectively complementing GPS data, the signal loss problem in complex environments such as tunnels and mountainous areas is completely solved, enabling continuous tracking and monitoring of transport vehicles throughout the entire process.

[0042] 3. High positioning accuracy and reliability: The system adopts a confidence weight fusion mechanism, which prioritizes high-confidence data sources (such as ETC) and downweights low-quality GPS data, effectively eliminating errors from a single data source and significantly improving the accuracy and reliability of the generated trajectory.

[0043] 4. Powerful Predictive Capabilities: Not only can it reconstruct historical trajectories, but it can also predict the real-time location and future arrival time of vehicles based on algorithmic models. The dynamic ETA model comprehensively considers real-world factors such as road conditions, weather, and driver behavior, resulting in predictions far more accurate than ETAs based on simple distance-speed calculations. Experimental verification shows that in trunk line transportation scenarios, its prediction error can be reduced by more than 60% compared to traditional methods.

[0044] 5. High degree of automation and intelligence: AI-powered automatic telephone interaction replaces traditional manual telephone inquiries, automatically extracts location information, greatly reduces labor costs, and improves the efficiency and timeliness of information collection.

[0045] 6. Enhance logistics management efficiency: Provides enterprises with unprecedented transportation transparency. Precise ETAs help consignees optimize resource allocation and reduce waiting costs; real-time anomaly alerts help managers intervene in a timely manner to mitigate risks, thereby improving the efficiency of the overall supply chain and customer satisfaction. Attached Figure Description

[0046] Figure 1 This is a flowchart of the method of the present invention.

[0047] Figure 2 This is a structural block diagram of the system of the present invention.

[0048] Figure 3 This is a structural block diagram of the data acquisition module in the system of the present invention.

[0049] Figure 4 This is a structural block diagram of the data processing module in the system of the present invention.

[0050] Figure 5 This is a structural block diagram of the prediction module in the system of the present invention. Detailed Implementation

[0051] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0052] In the description of this application, it should be noted that the terms "upper," "lower," "inner," "outer," "top / bottom," etc., indicating the orientation or positional relationship are based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0053] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installed," "equipped with," "sleeved / connected," "connected," etc., should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium; it can be a connection within two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0054] Example 1:

[0055] Reference Figure 1 This invention discloses a digital trajectory prediction method based on logistics transportation order tracking, comprising the following steps:

[0056] Step S1, Multi-source data collection: Collect multi-source transportation data including vehicle GPS data, AI telephone interaction data, ETC passage record data, and freight station transfer operation node data;

[0057] Step S2, Data Fusion Processing: Spatiotemporal alignment and data cleaning are performed on multi-source transportation data, and map matching is performed based on road network information;

[0058] Step S3, Trajectory Reconstruction: Using the processed multi-source data, a continuous digital trajectory is generated through interpolation algorithms and path extrapolation methods;

[0059] Step S4, Location Prediction: Based on the digital trajectory, combined with real-time traffic information and historical driving patterns, the latest predicted location of the vehicle is calculated through a prediction algorithm model;

[0060] Step S5, Arrival Time Prediction: Taking into account the remaining route, real-time traffic conditions, weather factors, and driver behavior patterns, the estimated arrival time is calculated using a dynamic ETA model.

[0061] The core idea of ​​this method is to no longer rely solely on GPS data, but to innovatively integrate multiple data sources with different characteristics and complement each other. Through a set of intelligent data processing and algorithm models, a continuous, reliable, and intelligent "digital trajectory" is generated, and prediction function is realized on this basis.

[0062] In step S1, the AI ​​phone interaction data collection steps include: the system automatically makes an AI phone call to the driver at regular intervals or triggered times, obtains the driver's voice response through voice recognition technology, extracts key location information from the voice response through natural language processing technology, and associates the extracted location information with spatiotemporal coordinates to form location anchors.

[0063] Specifically, the data collected in step S1 includes:

[0064] GPS data: provided by in-vehicle IoT devices, is the basic data source for positioning.

[0065] AI-powered telephone interaction data is a key innovation of this invention. The system automatically dials the driver's number and uses Automatic Speech Recognition (ASR) and Natural Language Processing (NLP) technologies to intelligently extract vehicle location information (such as "I just passed the XX highway toll station" or "I'm resting at the XX service area") from the conversation, converting it into structured location anchor data. This is especially crucial when GPS fails.

[0066] ETC toll record data: obtained through the Electronic Toll Collection (ETC) system. ETC records have precise timestamps and geographical locations (toll station locations), and the data accuracy is extremely high, making them excellent trajectory calibration anchor points.

[0067] Freight station transshipment operation node data: This includes the time and location of operations such as loading, unloading, and transshipment. This data typically comes from a warehouse management system (WMS) or transportation management system (TMS) and marks the key nodes in the journey.

[0068] In step S2, different confidence weights are set for data from different sources. The confidence weight of ETC passage record data is the highest, followed by AI telephone interaction data. The weight of GPS data is dynamically adjusted according to signal strength, and data fusion is performed based on the confidence weights.

[0069] In step S2, the aforementioned multi-source transportation data is preprocessed, including:

[0070] Spatiotemporal alignment: unifying data from different sources and with different timestamps into the same spatiotemporal coordinate system.

[0071] Data cleaning: Removing obviously erroneous, duplicate, or invalid data.

[0072] Map Matching: Matches the original coordinates to the actual road network to correct positioning drift.

[0073] This step innovatively introduces a confidence weighting mechanism. Different confidence weights are assigned to different data sources based on their inherent characteristics. For example, ETC data, with the highest accuracy, is given the highest weight (0.95); AI interaction data, after NLP parsing, is given a relatively high weight (0.85); and GPS data has its weight dynamically adjusted based on its current signal strength index—higher weight for stronger signals and lower weight for weaker signals. Data fusion based on these weights maximizes the utilization of highly reliable information.

[0074] In step S3, when the GPS signal is lost, a map matching algorithm based on a Hidden Markov Model (HMM) is used. This algorithm combines the last known location, vehicle direction, and road network topology constraints to extrapolate the most likely path the vehicle takes to fill the trajectory gaps. Using the fused, high-quality data, an interpolation algorithm (filling in points between data points) and path extrapolation methods are employed to generate a continuous and smooth "intelligent digital trajectory." Specifically, when the GPS signal is lost, a map matching algorithm based on a Hidden Markov Model (HMM) is used. This algorithm combines the last known location, vehicle direction, speed, and road network topology (such as one-way streets, no-turn zones, etc.) as constraints to calculate the probability distribution of the vehicle on all possible paths and extrapolates the path with the highest probability as the vehicle's actual driving path, thus intelligently filling the trajectory blind spots.

[0075] In step S4, based on the generated high-precision digital trajectory, combined with real-time traffic information (such as congestion data from map services) and historical driving pattern big data, the latest predicted position of the vehicle is extrapolated and calculated using machine learning prediction algorithm models (such as Kalman filtering, Long Short-Term Memory networks, etc.). This not only provides the historical trajectory but also the vehicle's real-time estimated position.

[0076] In step S5, considering multiple dynamic factors such as remaining path distance, real-time traffic conditions (congestion), weather factors (the impact of rain, snow, fog, etc. on vehicle speed), and driver behavior patterns (e.g., average rest time at service areas and driving habits of specific drivers), the estimated arrival time is calculated using a dynamic ETA model. The dynamic ETA model is expressed as: ETA = T_current + Σ + ΣT_stop, where ETA is the estimated arrival time, T_current is the current time, D_i is the distance of the i-th segment, V_i is the predicted speed of the i-th segment based on real-time traffic conditions, historical average speed, and weather factors, and T_stop is the predicted stop time based on historical data and current order information. The predicted speed V_i is obtained through a machine learning model. The input features of this model include real-time traffic flow data, historical average speed of the segment during the same period, current weather conditions, and date type. V_i (predicted speed) and T_stop (predicted stop time) are no longer fixed values ​​but are dynamically calculated by the sub-model, allowing the prediction results to be dynamically adjusted according to environmental changes, resulting in greater accuracy.

[0077] This embodiment describes in detail the application process of the method of the present invention in a specific logistics transportation order tracking scenario.

[0078] The method begins with multi-source data acquisition step S1:

[0079] The vehicle-mounted GPS terminal uploads latitude and longitude coordinates, speed, direction, and signal strength data once per minute. If the system monitoring platform detects that a vehicle's GPS signal strength has been below the threshold for more than 5 minutes, it triggers the AI-powered phone interaction process.

[0080] AI-powered phone interaction sub-process begins:

[0081] S101: The system will automatically call the driver's registered mobile phone number.

[0082] S102: After the call is connected, a pre-recorded voice message is played: "Hello, the system has detected a weak location signal. What is your approximate current location?"

[0083] S103: The driver responded: "I just came out of the Qingshui Tunnel, and I'm probably on the G55 Expressway."

[0084] S104: The system uses ASR technology to convert speech into text: "I just exited the Qingshui Tunnel and am probably on the G55 Expressway."

[0085] The S105 NLP engine performs word segmentation and entity recognition on the text, extracting the key geographical entities "Qingshui Tunnel" and "G55 Expressway".

[0086] S106: The system queries the geographic information database, resolves "Qingshui Tunnel" as a specific location (latitude and longitude coordinates) on the G55 expressway, and binds this coordinate with the current time and vehicle ID to generate a high-weight AI positioning data record.

[0087] Simultaneously, the data acquisition module also retrieved the vehicle's passage record from the provincial ETC platform via API, obtaining a passage record from 10 minutes prior at the "G55 Expressway Xishan Toll Station," which included the precise passage time and toll station location. The module also retrieved the order's loading completion timestamp and warehouse address from the TMS system.

[0088] The next step is data fusion processing step S2:

[0089] All data received by the system (GPS points, AI positioning points, ETC records, warehouse addresses) are uniformly converted to the WGS-84 coordinate system.

[0090] The data is cleaned to remove drifting GPS points that are significantly off-road.

[0091] Perform map matching to match all data points to the road lines of the G55 expressway.

[0092] The system assigns confidence weights to the current data: newly acquired ETC records have the highest weight (0.95), followed by AI interactive data (0.85), while GPS data has its weight dynamically reduced to 0.4 due to weak signal.

[0093] The next step is trajectory reconstruction step S3:

[0094] The system detected that GPS data has been missing for some time.

[0095] The system takes the last high-confidence location point (such as an ETC toll station or an AI location point) as the starting point.

[0096] By combining the vehicle's direction of travel at this point (estimated from previous GPS data) and road network information (such as the fact that there is no exit on the G55 expressway in the tunnel section), a hidden Markov model is used to perform path deduction.

[0097] The model calculates the most likely path for the vehicle during the blind spot and connects it with the effective data segments at the beginning and end to form a continuous and complete trajectory, seamlessly filling the GPS blind spot.

[0098] Then comes the location prediction step S4:

[0099] Based on the newly reconstructed trajectory segment, the system combines real-time traffic big data (which indicates that the G55 expressway is currently clear with an average speed of 90 km / h) and extrapolates the data using a Kalman filter algorithm.

[0100] The vehicle's latest predicted position is calculated to be a point on the trajectory projection path, and its current speed is estimated.

[0101] Finally, there is the arrival time prediction step S5:

[0102] The system calculates the remaining path from the predicted location to the destination.

[0103] Dynamic ETA model launch:

[0104] The remaining path is divided into several segments.

[0105] Query the predicted speed V_i of each road segment based on real-time traffic conditions (e.g., 90km / h for a smooth road segment, and 60km / h for a road construction segment ahead).

[0106] According to the weather forecast, there is no severe weather expected within the next 2 hours, so the weather factor coefficient is 1.0.

[0107] According to the driver's behavior profile, he usually takes a 20-minute break after driving for 3 hours. Based on the itinerary, he is about to reach his break point, so the predicted stop time T_stop is 20 minutes.

[0108] Substituting into the formula ETA=T_current+Σ(D_i / V_i)+ΣT_stop, the accurate estimated arrival time can be calculated.

[0109] The ETA results are provided to logistics managers and consignees through the application interface module.

[0110] Example 2:

[0111] This invention also provides a predictive system that implements the method of Embodiment 1, providing logistics companies with a fully visualized and predictable intelligent logistics management solution. The system is deployed on a cloud server, and its modules communicate through a microservice architecture, enabling efficient handling of massive concurrent vehicle tracking requests.

[0112] Specifically, refer to Figure 2 The present invention provides a digital trajectory prediction system based on logistics transportation order tracking that implements the method described in Embodiment 1 above, comprising:

[0113] The data acquisition module is configured to collect transportation data from multiple data sources. For details, please refer to [link / reference]. Figure 3 The data acquisition module includes a GPS receiver submodule, an ETC interface submodule, and a TMS / WMS interface submodule. These submodules connect to their respective external data sources to achieve automated data acquisition. The data acquisition module is responsible for collecting data from various data sources such as GPS devices, ETC systems, AI telephone platforms, and TMS / WMS systems.

[0114] The data processing module is configured to perform cleaning, spatiotemporal alignment, confidence weighting and fusion, and map matching on multi-source data. For details, please refer to [link / reference]. Figure 4 The data processing module includes a data cleaning unit, a spatiotemporal alignment unit, a confidence management unit, and a map matching engine. This module is responsible for standardizing and fusion preprocessing the collected raw data.

[0115] The trajectory generation module is configured to generate continuous digital trajectories based on the processed data. Specifically, it uses a trajectory reconstruction algorithm to generate digital trajectories. The trajectory generation module has built-in interpolation algorithms and HMM-based path inference algorithms to generate digital trajectories.

[0116] The prediction module is configured to predict the real-time location and estimated time of arrival of vehicles. Specifically, it integrates location prediction and dynamic ETA prediction algorithms, as shown in the reference... Figure 5 The prediction module includes a real-time location prediction submodule and a dynamic ETA prediction submodule.

[0117] The application interface module is configured to provide users with data query, trajectory display and early warning services. Specifically, it provides users with services such as trajectory visualization, location query, ETA query and anomaly early warning in the form of application programming interface (API) or graphical interface (GUI).

[0118] Reference Figure 3 The data acquisition module also includes an AI telephone interaction submodule, which includes: a call unit, used to automatically initiate a call to the driver according to a preset strategy; a voice recognition unit, used to convert the driver's voice response into text; a natural language processing unit, used to identify and extract geographic location keywords from the text; and a data association unit, used to bind the extracted location information with the current time and vehicle information to generate auxiliary positioning data.

[0119] The prediction module integrates a dynamic ETA prediction submodule, which connects to third-party traffic big data services and weather data services, and has a built-in driver behavior profile library for fine-tuning predicted arrival times. The application interface module provides a RESTful API 451 for other systems (such as TMS and customer service systems) to call, and provides a web visualization interface 452 for users to directly query and monitor.

[0120] Example 3:

[0121] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in Embodiment 1 above.

[0122] This invention employs a unique combination of multi-source data types, innovatively fusing AI telephone interaction data with GPS data, ETC toll record data, and freight station transfer operation node data. While existing technologies such as CN119559784A mention multi-source data fusion, they do not address the application of AI telephone interaction data. The introduction of AI telephone interaction data enables the system to obtain vehicle location information through intelligent voice interaction even when GPS signals are lost, a feature not addressed in existing technologies.

[0123] This invention employs a confidence weighting mechanism, proposing to assign different confidence weights to different data sources based on their inherent characteristics. ETC toll record data has the highest confidence weight, followed by AI telephone interaction data, while GPS data has its weight dynamically adjusted based on signal strength. This dynamic weighting mechanism is not reported in existing logistics tracking technologies and is a unique innovation of this invention.

[0124] This invention presents a path deduction algorithm based on Hidden Markov Models (HMM). In the trajectory reconstruction step, when GPS signals are lost, a map matching algorithm based on HMM is employed, combined with the last known location, vehicle direction, and road network constraints, to deduce the most probable vehicle path. This path deduction method based on a probabilistic model is innovative in the field of logistics tracking.

[0125] This invention employs a dynamic ETA model, which comprehensively considers remaining route, real-time traffic conditions, weather factors, and driver behavior patterns, calculating the estimated arrival time through multi-factor fusion. Compared to traditional ETA models based on simple distance-speed calculations, this invention's model is more accurate and dynamic.

[0126] The implementation principle of this invention is as follows: This invention discloses a digital intelligent trajectory prediction method and system based on logistics transportation order tracking, belonging to the field of intelligent logistics technology. The method includes: a multi-source data acquisition step, used to collect multi-source transportation data including vehicle GPS data, AI telephone interaction data, ETC passage record data, and freight station transfer operation node data; a data fusion processing step, used to perform spatiotemporal alignment, data cleaning, and map matching based on road network information on the multi-source transportation data; a trajectory reconstruction step, used to generate a continuous digital intelligent trajectory using the processed multi-source data through interpolation algorithms and path extrapolation methods; a location prediction step, used to calculate the latest predicted location of the vehicle based on the digital intelligent trajectory, combined with real-time traffic information and historical driving patterns, through a prediction algorithm model; and an arrival time prediction step, used to comprehensively consider the remaining path, real-time road conditions, weather factors, and driver behavior patterns, and calculate the estimated arrival time through a dynamic ETA model. This invention effectively solves the GPS signal blind spot problem, achieving blind-spot-free tracking and high-precision arrival time prediction throughout the entire process, greatly improving the transparency and intelligent management level of logistics transportation.

[0127] This invention provides a novel logistics transportation tracking solution. Through technological innovations such as multi-source data fusion, confidence weight allocation, route extrapolation, and dynamic prediction, it solves the long-standing problems of GPS signal blind spots and insufficient prediction accuracy that have plagued the logistics industry. Compared with the closest existing technology, CN119559784A, this invention has significant improvements and innovations in data sources, processing methods, and prediction models, possessing outstanding substantive features. This invention achieves end-to-end blind-spot-free tracking and high-precision arrival time prediction, with significantly superior technical performance compared to existing technologies. Experimental verification shows that in trunk transportation scenarios, its prediction error can be reduced by more than 60% compared to traditional methods. This improvement in technical performance is not easily anticipated by those skilled in the art and represents a significant advancement.

[0128] This invention expands its application scope beyond general cargo transportation to include specialized fields such as cold chain logistics, hazardous materials transport, and high-value goods transport. By integrating multimodal sensors for temperature, humidity, and vibration, and combining this with blockchain-based evidence storage technology, this invention provides end-to-end monitoring and traceability services for special goods—a capability unattainable by existing technologies. By enhancing the transparency and predictive accuracy of logistics transportation, this invention brings significant commercial value to logistics companies. Precise ETAs help consignees optimize resource allocation and reduce waiting costs; real-time anomaly warnings enable managers to intervene promptly, mitigate risks, and thus improve overall supply chain efficiency and customer satisfaction. This enhanced commercial value was unexpected by those skilled in the art.

[0129] The embodiments described herein are preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape, and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A digital trajectory prediction method based on logistics transportation order tracking, characterized in that, Includes the following steps: Step S1, Multi-source data collection: Collect multi-source transportation data including vehicle GPS data, AI telephone interaction data, ETC passage record data, and freight station transfer operation node data; Step S2, Data Fusion Processing: Spatiotemporal alignment and data cleaning are performed on the multi-source transportation data, and map matching is performed based on road network information; Step S3, Trajectory Reconstruction: Using the processed multi-source data, a continuous digital trajectory is generated through interpolation algorithms and path extrapolation methods; Step S4, Location Prediction: Based on the digital trajectory, combined with real-time traffic information and historical driving patterns, the latest predicted location of the vehicle is calculated through a prediction algorithm model; Step S5, Arrival Time Prediction: Taking into account the remaining route, real-time traffic conditions, weather factors, and driver behavior patterns, the estimated arrival time is calculated using a dynamic ETA model.

2. The digital trajectory prediction method based on logistics transportation order tracking according to claim 1, characterized in that, In step S1, the AI ​​telephone interaction data collection steps include: the system automatically makes an AI call to the driver at regular intervals or triggered times, obtains the driver's voice response through voice recognition technology, extracts key location information from the voice response through natural language processing technology, and associates the extracted location information with spatiotemporal coordinates to form location anchors.

3. A digital trajectory prediction method based on logistics transportation order tracking according to claim 1 or 2, characterized in that, In step S2, different confidence weights are set for data from different sources. The confidence weight of ETC passage record data is the highest, followed by AI telephone interaction data. The weight of GPS data is dynamically adjusted according to signal strength, and data fusion is performed based on the confidence weights.

4. The digital trajectory prediction method based on logistics transportation order tracking according to claim 1, characterized in that, In step S3, when the GPS signal is lost, a map matching algorithm based on a hidden Markov model is used to deduce the most likely driving path of the vehicle in order to fill the trajectory gap, combining the last known location, the vehicle's driving direction and the road network topology constraints.

5. The digital trajectory prediction method based on logistics transportation order tracking according to claim 1, characterized in that, The dynamic ETA model in step S5 is expressed as: ETA = T_current + Σ(D_i / V_i) + ΣT_stop, where ETA is the estimated arrival time, T_current is the current time, D_i is the distance of the i-th segment, V_i is the predicted speed of the i-th segment based on real-time traffic conditions, historical average speed and weather factors, and T_stop is the intermediate stop time predicted based on historical data and current order information.

6. The digital trajectory prediction method based on logistics transportation order tracking according to claim 5, characterized in that, The predicted speed V_i is obtained through a machine learning model whose input features include real-time traffic flow data, historical average vehicle speeds for the same road segment during the same period, current weather conditions, and date type.

7. A digital trajectory prediction system based on logistics transportation order tracking that implements the method described in any one of claims 1-6, characterized in that, include: The data acquisition module is configured to collect transportation data from multiple data sources; The data processing module is configured to perform cleaning, spatiotemporal alignment, confidence weighting, and fusion of multi-source data. The trajectory generation module is configured to generate continuous digital trajectories based on the processed data. The prediction module is configured to predict the vehicle's real-time location and estimated arrival time. The application interface module is configured to provide users with data query, trajectory display, and early warning services.

8. The intelligent trajectory prediction system based on logistics transportation order tracking according to claim 7, characterized in that, The data acquisition module includes an AI telephone interaction submodule, which includes: The call unit is used to automatically initiate a call to the driver according to a preset strategy; A voice recognition unit is used to convert the driver's voice responses into text; The Natural Language Processing Unit (NLP) is used to identify and extract geolocation keywords from text. The data association unit is used to bind the extracted location information with the current time and vehicle information to generate auxiliary positioning data.

9. The intelligent trajectory prediction system based on logistics transportation order tracking according to claim 7, characterized in that, The prediction module integrates a dynamic ETA prediction submodule, which connects to third-party traffic big data services and weather data services, and has a built-in driver behavior profile database for fine-tuning the predicted arrival time.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the method as described in any one of claims 1-6.

Images