Emergency rescue vehicle auxiliary positioning method fusing multi-source data

Abstract

The invention discloses an emergency rescue vehicle auxiliary positioning method fusing multi-source data. The method comprises the following steps: collecting ETC issuing data, transaction data and service area data, and carrying out multi-source fusion; extracting a mobile phone number and alarm time from the alarm event, binding a vehicle license plate number and generating a unique identifier; extracting a vehicle time sequence passing track based on an ETC door frame passing database, and taking a nearest door frame node as an upstream reference point; service area data and road network topology information are fused to calculate actual passing time; generating path probability distribution and a prior context vector by using a global topology attention mechanism through a path prediction module, and predicting a speed reference line through a task condition recursion unit based on a speed estimation module; and finally, the current position of the vehicle is calculated in combination with the pile number of the upstream portal, the actual passing time and the predicted speed, and abnormity correction positioning is performed in combination with GPS data of the mobile phone to obtain the route number, the downstream portal and the pile number information. According to the invention, lane-level accurate positioning is realized, and the rescue efficiency is effectively improved.

Description

Technical Field

[0001] This invention relates to the field of highway technology, and in particular to an emergency rescue vehicle auxiliary positioning method that integrates multi-source data. Background Technology

[0002] In the process of digital upgrading of highways, the emergency call platform, as a crucial entry point for highway emergency management, is the first channel for drivers and passengers to seek help in emergencies. The platform's operational efficiency and information processing capabilities directly impact the speed and quality of emergency response. In vehicle accident alarm processing, rescue efficiency and location accuracy are directly related to the safety of drivers and passengers, smooth traffic flow, and emergency management capabilities. If the specific location and direction of travel of vehicles cannot be obtained in a timely and accurate manner, rescue personnel will face problems of insufficient information and chaotic dispatching, potentially leading to delays in rescue or even secondary accidents, causing more serious casualties and property damage. Currently, highway emergency management departments urgently need a new method that can accurately predict the real-time location and direction of travel of vehicles to achieve rapid rescue and scientific control at accident sites. Simultaneously, accurately predicting the location and direction of travel of vehicles can also provide basic data support for the refined management of highways, promote the digital and intelligent development of the highway network, and ultimately improve overall operational efficiency and safety levels.

[0003] Under the existing technology and management system, emergency management of highway accidents faces the following problems: 1. Insufficient positioning accuracy and high misjudgment rate. Currently, vehicle positioning on rescue call platforms mainly relies on GPS information from mobile phone signaling data and manual inquiry. However, GPS positioning using mobile phone signaling data can only determine a general range, and in highway scenarios, it cannot accurately locate the vehicle's position, let alone distinguish the specific lane direction and lane position of the vehicle. During manual inquiry, when the caller provides station information verbally, due to the complex highway network structure, discontinuous station markers, or unclear descriptions from drivers and passengers, the positioning information is often inaccurate. This positioning ambiguity directly leads to rescue vehicles being unable to determine their driving route, resulting in significant uncertainty in the dispatch of rescue resources, delaying rescue speed, and reducing rescue efficiency. 2. Positioning delay and inability to respond in real time. Traditional methods such as manual patrols or video surveillance have long information collection and transmission chains and slow responses, failing to meet real-time requirements. Especially in the initial stage of an accident, positioning delays prevent rescue vehicles from reaching the scene immediately, missing the golden rescue time and posing a serious threat to personal safety and traffic order. 3. Data fragmentation and lack of unified integration. Currently, highway management involves multiple data sources, including ETC transaction records, service area stop information, license plate recognition data, and telephone alarm information, which are scattered across different systems. These data have different formats, inconsistent encoding, and lack effective interfaces between systems, making efficient data fusion impossible. Data fragmentation means that prediction algorithms lack reliable input, thus reducing the accuracy and reliability of prediction results. 4. Insufficient emergency decision support. Due to information delays and inaccurate positioning, emergency management departments lack real-time and reliable data support when formulating rescue plans. Dispatch decisions often rely on human experience, which may lead to blind and delayed rescue operations, and even further deterioration of on-site traffic, resulting in severe traffic congestion and increasing the risk of secondary accidents. The lack of scientific decision support has become a major bottleneck restricting the improvement of highway accident emergency management capabilities. 5. Coupling gap in existing algorithm models: Emergency calculations rely on kinematic formulas (position = speed × time), and the core challenge lies in the strong coupling between "driving direction" and "speed baseline". The vehicle's speed baseline is highly dependent on its selected path (e.g., main road vs. ramp). Existing technologies, whether basic statistical models (such as multiple linear regression) or single, homogeneous deep learning models, lack an explicit mechanism to decouple this dependency.

[0004] These "path-blind" algorithms cannot effectively determine the macroscopic intentions of vehicles when facing intersections, resulting in fundamentally inaccurate speed predictions. Summary of the Invention

[0005] The purpose of this invention is to solve common problems in highway emergencies, such as insufficient vehicle positioning accuracy, delayed positioning information, inconsistent data, and insufficient emergency decision support. It provides an emergency rescue vehicle auxiliary positioning method that integrates multi-source data. By integrating ETC transaction data, service area passage records, deep learning technology, and road network topology information, it can achieve accurate positioning and prediction of driving direction of vehicles on highways.

[0006] The technical solution adopted in this invention is:

[0007] An emergency rescue vehicle assisted positioning method that integrates multi-source data includes the following steps:

[0008] Multi-source datasets are obtained by collecting and fusing data from multiple sources.

[0009] Receive alarm events from the rescue call platform and extract the alarm mobile phone number and alarm time;

[0010] ETC card issuance data can be queried based on mobile phone numbers, and vehicle license plate numbers can be bound to generate unique identifiers.

[0011] Extract the time-series passage trajectory of the corresponding vehicle from the ETC gantry passage database, and use the last gantry node with the closest alarm time as the upstream reference gantry;

[0012] By integrating service area data and road network topology information, the actual travel time of vehicles in the accident section can be calculated.

[0013] Cascaded prediction is performed using a multi-task relationship model (ST-MTR) based on a spatiotemporal network: first, the path prediction module (PPM) generates the path probability distribution and path prior context vector, and then the speed estimation module (VPM) uses the context vector as a conditional input to predict the vehicle speed baseline.

[0014] The estimated vehicle position is calculated based on the gantry station number of the upstream benchmark, the actual travel time, and the predicted speed, and anomaly correction is performed by combining mobile phone GPS data.

[0015] Specifically, based on the corresponding position of the upstream reference gantry Actual travel time of vehicles in the accident section and the baseline value of the estimated future speed of the vehicle The estimated station position of the current vehicle is calculated. ; .

[0016] It outputs vehicle route number, downstream gantry and mileage information to the rescue platform to provide location information support for traffic management and emergency rescue.

[0017] Furthermore, the steps for extracting the temporal travel trajectory of the corresponding vehicle are as follows:

[0018] Extract all transaction records for the corresponding vehicle from the ETC gantry toll database. Each transaction record is in the following format:

[0019] ;

[0020] Among them, Passid is a unique track number, Node is the identifier of the toll station or gantry, and Time is the corresponding transaction time;

[0021] By sorting transaction record data chronologically and grouping it according to the Passid number, a time-series passage trajectory for vehicles is generated. The expression for the time-series passage trajectory is:

[0022] ;

[0023] in, This is the first node of the time-series travel trajectory; The transaction time of the first node in the time-series travel trajectory; and so on. The first time-series travel trajectory One node; The first time-series travel trajectory The transaction time of each node;

[0024] Based on the alarm time (TimeAlarm), the trajectory closest to the alarm time is extracted from the time-series traffic trajectory data, and the last node of the corresponding trajectory is regarded as the latest node before the accident occurred.

[0025] Extract the latest node before the event from the gantry information table. Lane information is used as the upstream reference gantry for the accident. The expression is:

[0026] ;

[0027] in, The route name; For lanes; This refers to the station number.

[0028] Furthermore, the steps for calculating the actual travel time of the vehicle in the accident section are as follows:

[0029] Calculate the time difference between the two points in time when the vehicle enters and leaves the accident area. The expression is:

[0030] ;

[0031] in, The time the vehicle left the accident area; This refers to the time the vehicle entered the accident area.

[0032] Based on the vehicle's actual driving trajectory, combined with information from toll stations and service area entrances / exits, the actual driving time of the vehicle in the accident section is calculated; by integrating service area data, the actual passage time of the vehicle in the accident section is calculated.

[0033] upstream reference gantry Starting from a node, the system retrieves all adjacent road segments originating from that node based on the road network topology, and integrates toll station information to construct a target toll station set. Its expression is as follows:

[0034]

[0035] in, Indicates the relationship with the reference node The total number of downstream nodes or toll stations with direct topological connections; the purpose of constructing this set is to delineate the possible departure range of vehicles, thereby retrieving whether there is a corresponding departure record for the vehicle.

[0036] Traverse the segment to find its road topology and related service area entrance / exit information, then traverse the service area entrance / exit data to check if the corresponding vehicle license plate number exists. The corresponding passage record; if it exists, the dwell time of the corresponding vehicle in the service area is calculated. Calculate the actual travel time of the vehicle within the accident area. :

[0037] ;

[0038] in, A function to calculate the time difference between two points in time. For alarm time, This refers to the passage time of a vehicle through the upstream reference gantry.

[0039] Furthermore, the path prediction module (PPM) employs a global topology attention mechanism to perform the following steps:

[0040] Convert vehicle gantry passage records into a spatiotemporal state sequence;

[0041] The path probability distribution is generated by calculating the road network node dependency weight matrix through self-attention;

[0042] Specifically, depending on the weight matrix The formula is as follows:

[0043] ;

[0044] in, Let be the input trajectory feature tensor. is the scaling factor for the feature space.

[0045] ;

[0046] in, This is the weight matrix of the path prediction layer. These are the bias parameters for the path prediction layer. For global intent distribution, It is a normalized exponential function.

[0047] The predicted probability distribution vector generates a path prior context vector through a learnable feature compression operator. .

[0048] in, Here is the weight matrix of the feature compression operator. These are the bias parameters for the feature compression operator. This is the activation function.

[0049] vector The topological intent of the macroscopic path is condensed and will be passed to the subsequent VPM module as a "conditional constraint".

[0050] Furthermore, the velocity estimation module (VPM) employs a task conditional recursive unit (TC-RU) to perform the following steps:

[0051] Historical speed features are concatenated with path prior context vectors to form a hybrid vector;

[0052] Specifically, at each time step of VPM The input space is reconstructed into a mixture of observation and prior information vectors. ; ;in, This represents the historical velocity characteristics at the current moment.

[0053] Through the information retention gate Information injection gate and status output gate The dynamic gating mechanism is used to dynamically evolve and update the memory manifold, and output the velocity prediction value under the path prior constraint.

[0054] Specifically, the dynamic evolution equation is as follows:

[0055] (1) Gated activation calculation:

[0056] ;

[0057] ;

[0058] in, These represent the gate control values ​​for the information retention gate and the information injection gate, respectively. This represents the hidden state at the previous moment; This represents a hybrid vector at the current moment, which concatenates historical velocity features with path prior context. This represents the Sigmoid activation function; and These represent the weight matrix and bias parameters of the information retention gate, respectively. and These represent the weight matrix and bias parameters of the information injection gate, respectively.

[0059] (2) Memory manifold update: Calculate the candidate memory states at the current time step. And update the global memoiform manifold. :

[0060] ;

[0061] ;

[0062] in, This represents the candidate memory state calculated at the current moment; and Let represent the global memory manifold at the previous time step and the global memory manifold at the current time step, respectively; and These represent the weight matrix and bias parameters of the candidate memory state computation layer, respectively. Represents the hyperbolic tangent activation function; It represents the Hadamardi (or Hadama) stack.

[0063] (3) Hidden state and velocity mapping: Update the hidden state at the current time step. And mapped to the final velocity baseline value. :

[0064]

[0065]

[0066]

[0067] in, Indicates the gate value of the state output gate; This represents the hidden state at the current moment; This is the final velocity baseline value output after mapping; and These represent the weight matrix and bias parameters of the state output gate, respectively; This represents the regression function used for velocity prediction.

[0068] Specifically, during training, the velocity estimation module (VPM) divides the original historical velocity sequence and path prior context vector into multiple training samples according to a preset fixed window length. Each training sample consists of an input sequence (containing feature vectors from multiple time steps) and a corresponding target velocity value (i.e., a "personalized normal velocity baseline" for future time steps).

[0069] Furthermore, the steps for anomaly correction using mobile phone GPS data are as follows:

[0070] Compare the estimated station location with the feasible station range determined by mobile phone GPS data;

[0071] If the estimated station location is not within the feasible station range, the location shall be recalculated based on the downstream gantry.

[0072] If the estimated station location is within the range but the deviation exceeds the threshold, a weighted average fusion of the GPS location and the predicted location is used.

[0073] Specifically, the estimated station number is compared with the GPS location, and the mobile phone GPS location information is correlated with the gantry GPS location information. Real-time GPS coordinates provided by the vehicle's mobile phone are used, based on the road network topology and Geographic Information System (GIS) data. By mapping the mobile phone GPS point to the nearest road segment and combining it with the gantry information, the gantry section is obtained. Together with the starting and ending station numbers, a feasible station range is formed. The estimated station position obtained through foundation location calculation ( The estimated location is compared with the possible road segment range determined by GPS information. If the estimated location exceeds the feasible road segment range determined by mobile phone GPS data, the estimated location is considered abnormal.

[0074] Anomaly correction mechanism, when the estimated station position is detected. There is an anomaly (i.e.) or The system will activate the anomaly correction mechanism. The correction strategy is as follows:

[0075] (1) Incorrect prediction of downstream gantry; if the estimated location It clearly exceeded the downstream gantry through which the vehicle could pass. If the mobile phone GPS data points to the vicinity of the downstream gantry, then the mobile phone GPS data is prioritized, and the downstream gantry is used as the new reference point to recalculate the vehicle's position. At this point, the new estimated position... It can be updated to:

[0076] .

[0077] (2) Location information correction; if the estimated location Even if the location is within a feasible road segment, if the distance between the mobile phone GPS location and the nearest road marker deviates significantly from the set tolerance threshold, a weighted average method is used to fuse the basic location estimation results and the mobile phone GPS location information to obtain a more accurate location estimate. The fused location... Represented as:

[0078] ;

[0079] in, It is a weighting factor that is dynamically adjusted based on the confidence level of the mobile phone GPS data, the reliability of the gantry data, and the confidence level of the prediction model output. This is the estimated location before fusion.

[0080] Furthermore, the multi-source data includes ETC transaction records, service area parking information, road network topology, time-periodic characteristics, and group preference profiles.

[0081] Furthermore, it also includes a sliding update mechanism: dynamically updating the vehicle's travel trajectory and gantry recognition results during the alarm process, supporting license plate number changes or re-entry.

[0082] Furthermore, data involving sensitive user information is encrypted during transmission.

[0083] This invention employs the above technical solution, utilizing path prediction to optimize speed estimation. The architecture first uses a self-attention mechanism to learn the road network topology and vehicle global path preferences, constructing a path prediction (PPM) module to solve the trajectory prediction problem. Subsequently, the path prior context vector output by the PPM module is used as a conditional input to model the relationships between multiple tasks, constructing and optimizing a speed prediction (VPM) module to estimate the expected driving speed for the predicted path.

[0084] The present invention has the following advantages and beneficial effects compared with existing methods:

[0085] 1. Rapid positioning and second-level response: By integrating the call information from the rescue call platform with real-time data from the ETC gantry, the positioning calculation and driving direction judgment can be completed within seconds after the call is connected, achieving a true second-level response, greatly shortening the rescue response time and reducing the risk of casualties and traffic congestion.

[0086] 2. High-precision positioning and direction prediction (bridging the coupling gap). Traditional mobile phone signaling or GPS positioning can only provide approximate latitude and longitude information, and cannot pinpoint the exact location of highway lanes, lane A / B, and station markers. More importantly, traditional methods cannot resolve the strong coupling between vehicle "path selection" and "speed dynamics".

[0087] 3. Data Fusion and Generalization Capabilities. Through a unified data access interface, the system integrates multi-source data, including alarm calls, ETC gantries, group preference profiles, and geographical topology, enabling end-to-end collaborative analysis. Crucially, this invention does not rely on personalized historical vehicle profiles; instead, it learns road network topology and group preferences through a global topological attention mechanism, achieving high prediction accuracy even for "unfamiliar vehicles" completely absent from the validation set. Furthermore, the introduction of a sliding update mechanism allows vehicle information to be dynamically updated and supplemented during the alarm process. Even if license plate numbers change or are supplemented, calculation results can be corrected in real time, maintaining data continuity and accuracy, significantly enhancing the system's fault tolerance and practicality.

[0088] 4. Multiple location comparisons improve analysis efficiency, supporting the generation of multiple location results for real-time comparison by dispatchers, effectively reducing reliance on manual inquiries. This reduces location errors caused by unclear descriptions from the caller or inconsistent station numbers, making rescue command more efficient and streamlined.

[0089] 5. High compatibility and easy integration and promotion. The system adopts standardized interface specifications, seamlessly connecting with rescue call platforms and highway emergency management platforms. Deployment is simple, and it can be quickly promoted on existing infrastructure, reducing upgrade costs and operation and maintenance difficulties. Attached Figure Description

[0090] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments;

[0091] Figure 1 This is a flowchart illustrating the emergency rescue vehicle auxiliary positioning method that integrates multi-source data according to the present invention.

[0092] Figure 2 This is a schematic diagram of the vehicle probability profile calculation process of the present invention;

[0093] Figure 3 This is a schematic diagram of the multi-task relationship model (ST-MTR) algorithm architecture of the spatiotemporal network of the present invention;

[0094] Figure 4 This is a schematic diagram of the access algorithm architecture for the rescue call platform of the present invention. Detailed Implementation

[0095] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0096] like Figures 1 to 4 As shown in the figure, the present invention discloses an emergency rescue vehicle assisted positioning method that integrates multi-source data, including the following steps:

[0097] Step 1, Data Collection and Preprocessing, primarily utilizes big data technology to collect and integrate multi-source data. Key data sources include: ETC Issuance Data: Data collected during ETC card issuance, including vehicle license plate number, ETC card number, and registered phone number. ETC System Transaction Data: Vehicle passage data collected through the ETC toll collection system on highways, including license plate number, passid (unique toll identifier), transaction time, and transaction node, where transaction nodes include ETC gantries and toll stations. Service Area Data: Records of vehicles entering and exiting service areas, including service area name, license plate number, entry time, whether exited, and exit time, used to determine whether a vehicle has entered a service area, thus eliminating misjudgments of vehicles stranded due to stops at service areas.

[0098] Step 2: Receive alarm events and extract vehicle identifiers; Upon receiving an alarm event from the emergency call platform, the system extracts the phone number (PhoneID) and alarm time (TimeAlarm). By querying ETC issuance data, the system binds the PhoneID to the license plate number, thereby obtaining the vehicle's unique identifier (PlateID). This process provides necessary identifier data for subsequent vehicle location and accident analysis.

[0099] Step 3: Extraction of vehicle's current trajectory information; Once the vehicle's identifier PlateID is determined, the system will extract all transaction records for that vehicle from the ETC gantry toll database. Each record consists of the following:

[0100] ;

[0101] Among them, Passid is a unique track number, Node is the identifier of the toll station or gantry, and Time is the corresponding transaction time.

[0102] By sorting transaction data chronologically and grouping it according to Passid, the system generates the time-series passage trajectory of vehicles:

[0103] ;

[0104] in, This is the first node of the time-series travel trajectory; The transaction time of the first node in the time-series travel trajectory; and so on. The first time-series travel trajectory One node; The first time-series travel trajectory The transaction time of each node.

[0105] Based on the alarm time (TimeAlarm), the system extracts the trajectory closest to the alarm time from the above trajectory data, and sets the last node of that trajectory as... This is considered the latest point in time before the accident occurred.

[0106] Extract the latest node before the accident from the gantry information table. Lane information (such as lane) , station number Route Name (Information on the upstream reference gantry where the accident occurred) The expression is: .

[0107] Step 4: Integrate service area data with road network topology analysis; As shown in Table 1, to further improve the accuracy of location prediction, the system integrates service area data with the road network topology to calculate the actual travel time of vehicles in the accident section. First, the system calculates the time difference between two time points using the following formula: ; The time the vehicle left the accident area; This refers to the time the vehicle entered the accident area.

[0108] Then, based on the vehicle's actual driving trajectory and information from toll stations and service area entrances / exits, the vehicle's actual travel time within the accident area is calculated. This calculation integrates service area data to determine the vehicle's actual travel time within the accident area, using the upstream reference gantry. Starting from a node, the system retrieves all adjacent road segments originating from that node based on the road network topology, and integrates toll station information to construct a target toll station set. Its expression is as follows:

[0109] ;

[0110] in, Indicates the relationship with the reference node The total number of downstream nodes or toll stations with direct topological connections; the purpose of constructing this set is to delineate the possible departure range of vehicles, thereby retrieving whether there is a corresponding departure record for the vehicle.

[0111] Traverse the segment to find its topology and related service area entrance / exit information, then traverse the service area entrance / exit data to check if they exist. The vehicle's license plate number corresponds to its passage record. If it exists, calculate its dwell time in the service area. Based on this, the actual travel time of the vehicle within the accident area can be estimated. :

[0112] ;

[0113] in, A function to calculate the time difference between two points in time. For alarm time, This refers to the passage time of a vehicle through the upstream reference gantry.

[0114] Table 1 Comparison of location prediction performance of different models

[0115]

[0116] Step 5, Path Prediction Based on ST-MTR Architecture (PPM Module Construction); This invention constructs a path prediction module (PPM) to deconstruct the potential topological intent of vehicles from their historical discrete trajectories. Unlike traditional sequence models, this module employs a Global Topological Attention Mechanism, which captures the global path preferences of vehicles in complex road networks by calculating the correlation strength between road network nodes.

[0117] Vehicle spatiotemporal trajectory tensor quantization: Extract vehicle gantry passage records and construct vehicle spatiotemporal state sequences. ,in Includes node identifier embedding and timestamp encoding.

[0118] Global topological feature capture: Construct a self-attention-based topological encoder, which transforms the input sequence into a query space, a key space, and a feature space through a mapping function, denoted as […]. The dependency weight matrix between road network nodes is calculated using multi-subspace projection. The formula is as follows:

[0119] ;

[0120] in, Let be the input trajectory feature tensor. This is a scaling factor for the feature space. The process outputs the hidden layer states. This represents the global intent distribution of vehicles within the current road network topology.

[0121] Path probability distribution generation: through a fully connected mapping layer Projected onto the road network node space, the predicted probability distribution vector of the next gantry node is output. .

[0122] ;

[0123] in, This is the weight matrix of the path prediction layer. These are the bias parameters for the path prediction layer. For global intent distribution, It is a normalized exponential function.

[0124] Building bridges between task dependencies: To achieve asymmetric conditional dependencies between multiple tasks, the PPM module not only outputs prediction results but also generates "path prior context vectors" through a learnable feature compression operator. :

[0125] ;

[0126] in, Here is the weight matrix of the feature compression operator. These are the bias parameters for the feature compression operator. This is the activation function.

[0127] The vector The topological intent of the macroscopic path is condensed and will be passed to the subsequent VPM module as a "conditional constraint".

[0128] Step 6: Velocity estimation based on asymmetric conditional dependence (VPM module construction). As shown in Table 2, this invention constructs a velocity prediction module (VPM). Its core innovation lies in breaking the limitation of traditional independent velocity prediction and constructing a Task Conditional Recursive Unit (TC-RU). This unit, when performing time-series simulations, incorporates the velocity prediction module generated in step 5. As a global constraint, it realizes a logical closed loop of "path optimization speed".

[0129] State space reconstruction: at each time step of VPM The input space is reconstructed as a mixture of "observation-prior" vectors. Definition If the historical velocity characteristics at the current moment are given, then the reconstructed input tensor... Defined as:

[0130] ;

[0131] Task-condition recursive dynamic equations: Utilizing the temporal evolution of velocity data captured by the TC-RU unit. To precisely control the information flow, three dynamic gating mechanisms are defined: information retention gates. Information injection gate and status output gate The dynamic evolution equations are as follows:

[0132] (1) Gated activation calculation:

[0133] ;

[0134] ;

[0135] in, These represent the gate control values ​​for the information retention gate and the information injection gate, respectively. This represents the hidden state at the previous moment; This represents a hybrid vector at the current moment, which concatenates historical velocity features with path prior context. This represents the Sigmoid activation function; and These represent the weight matrix and bias parameters of the information retention gate, respectively. and These represent the weight matrix and bias parameters of the information injection gate, respectively.

[0136] (2) Memory manifold update: Calculate the candidate memory states at the current time step. And update the global memoiform manifold. :

[0137] ;

[0138] ;

[0139] in, This represents the candidate memory state calculated at the current moment; and Let represent the global memory manifold at the previous time step and the global memory manifold at the current time step, respectively; and represent the weight matrix and bias parameters of the candidate memory state computation layer, respectively; tanh represents the hyperbolic tangent activation function. It represents the Hadamardi (or Hadama) stack.

[0140] (3) Hidden state and velocity mapping: Update the hidden state at the current time step. And mapped to the final velocity baseline value. :

[0141] ;

[0142] ;

[0143] ;

[0144] in, Indicates the gate value of the state output gate; This represents the hidden state at the current moment; This is the final velocity baseline value output after mapping; and These represent the weight matrix and bias parameters of the state output gate, respectively; This represents the regression function used for velocity prediction. Through the above mechanism, the VPM module is not merely performing time series regression, but rather... Within the defined "path intent range", deduce the speed baseline that best matches the characteristics of the path.

[0145] Time series partitioning during model training: The original historical velocity sequence and the path prior context vector (i.e., the one in step 6.1) are used to partition the time series. The system is divided into multiple training samples according to a preset fixed window length. Each training sample consists of an input sequence (containing feature vectors of multiple time steps) and a corresponding target velocity value (i.e., the "personalized normal velocity baseline" for future time steps).

[0146] Feature normalization: Z-score standardization is performed on all input features to unify features with different dimensions and value ranges into a similar numerical range, thereby improving the stability and efficiency of model training.

[0147] Table 2. Performance comparison of different models on velocity prediction (velocity baseline) task.

[0148]

[0149] Step 7: Calculation of the vehicle's precise location based on ST-MTR; this step aims to comprehensively utilize historical vehicle trajectory information, gantry marker data, speed prediction results, and time extrapolation algorithms to achieve high-precision calculation of the vehicle's current location. Through multi-source information fusion technology, the system can accurately estimate the vehicle's real-time location in the event of vehicle disconnection or abnormality, providing crucial spatial positioning support for emergency rescue and traffic control.

[0150] Basic position calculation: The system uses the last successfully detected gantry as the reference point. This is combined with the upstream gantry station position from step 3. The actual travel time of vehicles in step 5 And the predicted future vehicle speed in step 7 The current station number of the vehicle can be deduced. Its mathematical expression is as follows:

[0151] ;

[0152] Anomaly correction through multi-source data fusion: This section is the key anomaly detection and position correction step. By fusing data from multiple sources, it identifies and corrects vehicle position estimation biases caused by model prediction errors or missing data.

[0153] (1) The estimated station number is compared with the GPS location. The mobile phone GPS location information is associated with the gantry GPS location information. The real-time GPS coordinates provided by the vehicle's mobile phone are used, based on the road network topology and Geographic Information System (GIS) data. By mapping the mobile phone GPS point to the nearest road segment and combining it with the gantry information, the gantry segment is obtained. Together with the starting and ending station numbers, a feasible station range is formed. The estimated station position obtained through foundation location calculation ( The estimated location is compared with the possible road segment range determined by GPS information. If the estimated location exceeds the feasible road segment range determined by mobile phone GPS data, the estimated location is considered abnormal.

[0154] (2) Anomaly correction mechanism: when the estimated station position is detected There is an anomaly (i.e.) or The system will activate the anomaly correction mechanism. The correction strategy is as follows:

[0155] (a) Incorrectly predicted downstream gantry. If the predicted location... It clearly exceeded the downstream gantry through which the vehicle could pass. If the mobile phone GPS data points to the vicinity of the downstream gantry, then the mobile phone GPS data is prioritized, and the downstream gantry is used as the new reference point to recalculate the vehicle's position. At this point, the new estimated position... It can be updated to: ;

[0156] (b) Location information correction. If the estimated location Even if the location is within a feasible road segment, if the deviation from the mobile phone GPS location to the nearest road marker is significant and exceeds the set tolerance threshold, a weighted average method is used to fuse the basic location estimation results and the mobile phone GPS location information to obtain a more accurate location estimate. The fused location... It can be represented as:

[0157] ;

[0158] in, It is a weighting factor that is dynamically adjusted based on the confidence level of the mobile phone GPS data, the reliability of the gantry data, and the confidence level of the prediction model output.

[0159] Step 8: The system outputs information including vehicle route number, downstream gantry, and vehicle marker number, providing comprehensive location information support for traffic management and emergency rescue.

[0160] This invention employs a multi-task inter-relationship architecture model (ST-MTR) based on a spatiotemporal network: first, a path prediction module (PPM) is used to solve the path problem; then, this path prior is used as a condition to optimize a velocity estimation module (VPM) based on asymmetric conditional dependence to solve the velocity baseline problem. This cascaded mechanism enables higher accuracy in velocity baseline prediction than independent models without path guidance, thereby achieving high-precision reconstruction of vehicle travel paths and accurate inference of the vehicle's current position, direction of travel, and station location.

[0161] The present invention has the following advantages and beneficial effects compared with existing methods:

[0162] 1. Rapid positioning and second-level response: By integrating incoming call information from the rescue call platform with real-time data from the ETC gantry, positioning calculation and driving direction determination can be completed within seconds of the call being connected, achieving a true second-level response, significantly shortening rescue response time, and reducing the risk of casualties and traffic congestion.

[0163] 2. High-precision positioning and direction prediction (bridging the coupling gap). Traditional mobile phone signaling or GPS positioning can only provide approximate latitude and longitude information, and cannot pinpoint the exact location of highway lanes, lane A / B, and station markers. More importantly, traditional methods cannot resolve the strong coupling between vehicle "path selection" and "speed dynamics".

[0164] 3. Data Fusion and Generalization Capabilities. Through a unified data access interface, the system integrates multi-source data, including alarm calls, ETC gantries, group preference profiles, and geographical topology, enabling end-to-end collaborative analysis. Crucially, this invention does not rely on personalized historical vehicle profiles; instead, it learns road network topology and group preferences through a global topological attention mechanism, achieving high prediction accuracy even for "unfamiliar vehicles" completely absent from the validation set. Furthermore, the introduction of a sliding update mechanism allows vehicle information to be dynamically updated and supplemented during the alarm process. Even if license plate numbers change or are supplemented, calculation results can be corrected in real time, maintaining data continuity and accuracy, significantly enhancing the system's fault tolerance and practicality.

[0165] 4. Multiple location comparisons improve analysis efficiency, supporting the generation of multiple location results for real-time comparison by dispatchers, effectively reducing reliance on manual inquiries. This reduces location errors caused by unclear descriptions from the caller or inconsistent station numbers, making rescue command more efficient and streamlined.

[0166] 5. High compatibility and easy integration and promotion. The system adopts standardized interface specifications, seamlessly connecting with rescue call platforms and highway emergency management platforms. Deployment is simple, and it can be quickly promoted on existing infrastructure, reducing upgrade costs and operation and maintenance difficulties.

[0167] Obviously, the described embodiments are only a part of the embodiments of this application, not all of them. Without conflict, the embodiments and features in the embodiments of this application can be combined with each other. The components of the embodiments of this application described and illustrated herein can generally be arranged and designed in various different configurations. Therefore, the detailed description of the embodiments of this application is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

Claims

1. An emergency rescue vehicle assisted positioning method integrating multi-source data, characterized in that: It includes the following steps: Multi-source datasets are obtained by collecting and fusing data from multiple sources. Receive alarm events from the rescue call platform and extract the alarm mobile phone number and alarm time; ETC card issuance data can be queried based on mobile phone numbers, and vehicle license plate numbers can be bound to generate unique identifiers. Extract the time-series passage trajectory of the corresponding vehicle from the ETC gantry passage database, and use the last gantry node with the closest alarm time as the upstream reference gantry; By integrating service area data and road network topology information, the actual travel time of vehicles in the accident section can be calculated. Cascaded prediction is performed using a multi-task relationship model of a spatiotemporal network: first, the path prediction module generates the path probability distribution and the path prior context vector, and then the speed estimation module uses the context vector as a condition to input the predicted vehicle speed baseline. The estimated vehicle position is calculated based on the gantry station number of the upstream benchmark, the actual travel time, and the predicted speed, and anomaly correction is performed by combining mobile phone GPS data. It outputs vehicle route number, downstream gantry and mileage information to the rescue platform to provide location information support for traffic management and emergency rescue.

2. The emergency rescue vehicle assisted positioning method based on multi-source data according to claim 1, characterized in that: The steps to extract the time-series travel trajectory of the corresponding vehicle are as follows: Extract all transaction records for the corresponding vehicle from the ETC gantry passage database; By sorting transaction record data in chronological order and grouping it according to the trajectory number Passid, a time-series passage trajectory of the vehicle is generated. Extract the trajectory closest to the alarm time from the time-series traffic trajectory data, and then extract the last node of the corresponding trajectory. This is considered the latest point in time before the accident occurred. Extract the latest node before the event from the gantry information table. The lane information is used as the upstream reference gantry for the accident.

3. The emergency rescue vehicle auxiliary positioning method based on multi-source data according to claim 1, characterized in that: The steps for calculating the actual travel time of a vehicle in the accident area are as follows: Traverse the segment to find its road topology and related service area entrance / exit information, then traverse the service area entrance / exit data to check if the corresponding vehicle license plate number exists. The corresponding passage record; if it exists, the dwell time of the corresponding vehicle in the service area is calculated. Calculate the actual travel time of the vehicle within the accident area. : ； in, A function to calculate the time difference between two points in time. For alarm time, This refers to the passage time of a vehicle over the upstream reference gantry.

4. The emergency rescue vehicle auxiliary positioning method based on multi-source data according to claim 1, characterized in that: The path prediction module employs a global topology attention mechanism, which includes the following steps: Convert vehicle gantry passage records into a spatiotemporal state sequence; The global intent distribution of vehicles in the current road network topology is obtained by calculating the dependency weight matrix of road network nodes through self-attention. Global intent distribution Projecting the path prediction probability distribution vector onto the road network node space generates the path prediction probability distribution vector for the next gantry node. ; ;in, This is the weight matrix of the path prediction layer. These are the bias parameters for the path prediction layer. For global intent distribution, It is a normalized exponential function; The predicted probability distribution vector generates a path prior context vector through a learnable feature compression operator. ;in, Here is the weight matrix of the feature compression operator. These are the bias parameters for the feature compression operator. This is the activation function.

5. The emergency rescue vehicle auxiliary positioning method based on multi-source data according to claim 1, characterized in that: The velocity estimation module is implemented using task-conditional recursive units, including the following steps: Historical speed features are concatenated with path prior context vectors to form a hybrid vector; Through the information retention gate Information injection gate and status output gate The dynamic gating mechanism is used to dynamically evolve and update the memory manifold, and output the velocity prediction value under the path prior constraint.

6. The emergency rescue vehicle auxiliary positioning method based on multi-source data according to claim 5, characterized in that: The dynamic evolution equations are as follows: (1) Gated activation calculation: ； ； in, These represent the gate control values ​​for the information retention gate and the information injection gate, respectively. This represents the hidden state at the previous moment; This represents a hybrid vector at the current moment, which concatenates historical velocity features with path prior context. This represents the Sigmoid activation function; and These represent the weight matrix and bias parameters of the information retention gate, respectively. and These represent the weight matrix and bias parameters of the information injection gate, respectively. (2) Memory manifold update: Calculate the candidate memory states at the current time step. And update the global memoiform manifold. : ； ； in, This represents the candidate memory state calculated at the current moment; and Let represent the global memory manifold at the previous time step and the global memory manifold at the current time step, respectively; and These represent the weight matrix and bias parameters of the candidate memory state computation layer, respectively. Represents the hyperbolic tangent activation function; It represents the Hadamardi (or Hadama) stack; (3) Hidden state and velocity mapping: Update the hidden state at the current time step. And mapped to the final velocity baseline value. : ； ； ； in, Indicates the gate value of the state output gate; This represents the hidden state at the current moment; This is the final velocity baseline value output after mapping; and These represent the weight matrix and bias parameters of the state output gate, respectively; This represents the regression function used for velocity prediction.

7. The emergency rescue vehicle auxiliary positioning method based on multi-source data according to claim 1, characterized in that: The method for calculating the estimated vehicle station position is as follows: Based on the corresponding position of the upstream reference gantry Actual travel time of vehicles in the accident section and the baseline value of the estimated future speed of the vehicle The estimated station position of the current vehicle is calculated. ; .

8. The emergency rescue vehicle auxiliary positioning method based on multi-source data according to claim 1 or 6, characterized in that: The steps for anomaly correction using mobile phone GPS data are as follows: Compare the estimated station location with the feasible station range determined by mobile phone GPS data; If the estimated station location is not within the feasible station range, the location shall be recalculated based on the downstream gantry. If the estimated station location is within the range but the deviation exceeds the threshold, a weighted average fusion of the GPS location and the predicted location is used.

9. The emergency rescue vehicle assisted positioning method based on multi-source data according to claim 1, characterized in that: Multi-source data includes ETC transaction records, service area parking information, road network topology, time periodic characteristics, and group preference profiles.

10. The emergency rescue vehicle assisted positioning method according to claim 1, characterized in that: It also includes a sliding update mechanism: dynamically updating the vehicle's travel trajectory and gantry recognition results during the alarm process, supporting license plate number changes or re-entry.

Images