Emergency rescue vehicle intelligent scheduling method based on beidou positioning and related equipment

Abstract

The application discloses an emergency rescue vehicle intelligent scheduling method based on Beidou positioning and related equipment, and relates to the technical field of emergency management.The emergency rescue vehicle intelligent scheduling method based on Beidou positioning comprises the following steps: acquiring multi-modal state sensing data that is transmitted back by multiple emergency rescue nodes in real time, wherein the multi-modal state sensing data is composed of Beidou high-precision positioning information and a space-time label; inputting the multi-modal state sensing data into a preset intelligent scheduling model to generate a dynamic vehicle scheduling strategy and a cooperative path planning, wherein the intelligent scheduling model is obtained by pre-training based on historical disaster data, a rescue resource game relationship and a group survival entropy optimization target; and sending a scheduling instruction to a corresponding target emergency rescue node according to the dynamic vehicle scheduling strategy and the cooperative path planning.The application solves the technical problem of low scheduling efficiency in a multi-concurrent scene.

Description

Technical Field

[0001] This application relates to the field of emergency management technology, and in particular to an intelligent dispatching method and related equipment for emergency rescue vehicles based on BeiDou positioning. Background Technology

[0002] In the field of emergency rescue, efficient and precise vehicle dispatching is crucial for saving lives and minimizing losses. Traditional emergency rescue vehicle dispatching relies heavily on human experience. The dispatch center communicates with vehicles via voice radio, issuing commands based on static maps and limited road condition information. This lack of real-time and accurate information regarding the precise location, status, and dynamic situation of all rescue vehicles leads to a lack of data support for decision-making and creates "invisible" blind spots. The dispatching process is highly dependent on individual experience, making it difficult to quickly generate globally optimal vehicle assignment and route planning solutions under complex and ever-changing disaster situations (such as multiple disaster sites occurring simultaneously or dynamic road blockages), resulting in low efficiency and a high risk of errors. Over-reliance on public mobile communication networks means that in scenarios where disasters easily lead to public network outages or congestion, instructions cannot be reliably issued, and vehicles easily become "information islands." This results in low efficiency in emergency rescue vehicle dispatching under multi-concurrent emergency scenarios. Summary of the Invention

[0003] The main purpose of this application is to provide an intelligent dispatching method and related equipment for emergency rescue vehicles based on BeiDou positioning, aiming to solve the technical problem of low dispatching efficiency of emergency rescue vehicles in emergency multi-concurrency scenarios.

[0004] To achieve the above objectives, this application proposes an intelligent dispatching method for emergency rescue vehicles based on BeiDou positioning. The method includes:

[0005] Acquire multimodal state perception data transmitted in real time from multiple emergency rescue nodes, wherein the multimodal state perception data consists of BeiDou high-precision positioning information and spatiotemporal tags;

[0006] The multimodal state perception data is input into a preset intelligent scheduling model to generate dynamic vehicle scheduling strategies and collaborative path planning. The intelligent scheduling model is pre-trained based on historical disaster data, the game relationship of rescue resources, and the optimization objective of group survival entropy.

[0007] Based on the dynamic vehicle scheduling strategy and collaborative path planning, scheduling instructions are sent to the corresponding target emergency rescue nodes.

[0008] In one embodiment, the step of inputting the multimodal state perception data into a preset intelligent scheduling model to generate a dynamic vehicle scheduling strategy and cooperative path planning includes:

[0009] Based on the multimodal state perception data, it is input into the dynamic game environment model in the preset intelligent scheduling model;

[0010] Based on the dynamic game environment model, the rescue effectiveness weight of each emergency rescue node for different disaster points and the group survival entropy change rate of each disaster point are calculated. The rescue effectiveness weight is calculated by comprehensively considering the node type, real-time location, equipment status and historical efficiency. The group survival entropy change rate is calculated based on the scale of trapped personnel, environmental threat level and time decay factor.

[0011] With the optimization objective of minimizing the weighted sum of global group survival entropy and total rescue time, dynamic deduction is performed in the dynamic game environment model to obtain the optimal vehicle task allocation matrix and spatiotemporal path sequence, which serve as the dynamic vehicle scheduling strategy and collaborative path planning.

[0012] In one embodiment, the step of acquiring multimodal state perception data transmitted in real time by multiple emergency rescue nodes includes:

[0013] A node agent training framework based on federated causal perception is constructed, wherein each of the emergency rescue nodes acts as a distributed agent;

[0014] Each of the aforementioned intelligent agents trains a local causal reasoning sub-model based on local historical perception data, which is used to identify potential causal relationships between key state features and rescue outcomes;

[0015] Each of the aforementioned intelligent agents uploads the lightweight causal primitives obtained from training to the scheduling center for secure aggregation, generating a global causal knowledge graph;

[0016] Based on the global causal knowledge graph, the intelligent scheduling model is trained through knowledge distillation and reinforcement learning to improve the interpretability and predictive ability of its policy generation.

[0017] In one embodiment, the step of performing knowledge distillation and reinforcement learning training on the intelligent scheduling model based on the global causal knowledge graph includes:

[0018] High-frequency causal rules are extracted from the global causal knowledge graph and transformed into soft constraints for the scheduling strategy.

[0019] During the training process of the game deduction and scheduling model, a policy distillation loss function based on the soft constraint is introduced to guide the model to learn a policy space that conforms to causal priors.

[0020] In a pre-defined simulated disaster environment scenario, a pre-defined near-end policy optimization algorithm is used to conduct adversarial training on a model that incorporates causal constraints, with the success rate and efficiency of rescue as reward signals. This enables the model to stably output robust policies in complex games.

[0021] In one embodiment, the step of sending a dispatch instruction to the corresponding target emergency rescue node according to the dynamic vehicle dispatch strategy and cooperative path planning includes:

[0022] The optimal communication link combination is dynamically selected based on the key intent level of the scheduling instruction and the real-time channel state entropy of the target node. The key intent level is comprehensively evaluated by the spatiotemporal sensitivity, action necessity and information complexity of the instruction, and the channel state entropy comprehensively evaluates the uncertainty of link stability, bandwidth and delay.

[0023] For survivability-level commands, transmission is prioritized via the BeiDou short message communication link, and forward error correction coding and duplicate confirmation mechanisms are employed.

[0024] For situational data that is not survivability-level but has high bandwidth requirements, it is transmitted through a backup link pre-configured based on digital twin network simulation. The backup link is a converged communication channel established in advance based on the network twin simulation results.

[0025] In one embodiment, the step of sending the dispatch instruction to the corresponding target emergency rescue node further includes:

[0026] A unique digital thread identifier is generated for each rescue mission, and data from triggering, decision-making, execution to completion is recorded throughout the entire lifecycle.

[0027] The dynamic vehicle scheduling strategy, collaborative path planning, actual vehicle trajectory and status feedback data, and the final task results are correlated and compared along the digital thread.

[0028] Based on the comparison results, the strategy deviation and comprehensive effectiveness index of this task are calculated and fed back as incremental training data to the game inference scheduling model for online fine-tuning and continuous optimization of the model.

[0029] Furthermore, to achieve the above objectives, this application also proposes an intelligent dispatching device for emergency rescue vehicles based on BeiDou positioning, wherein the intelligent dispatching device for emergency rescue vehicles based on BeiDou positioning includes:

[0030] The acquisition module is used to acquire multimodal state perception data transmitted back in real time by multiple emergency rescue nodes, wherein the multimodal state perception data consists of Beidou high-precision positioning information and spatiotemporal tags;

[0031] The generation module is used to input the multimodal state perception data into a preset intelligent scheduling model to generate dynamic vehicle scheduling strategies and collaborative path planning. The intelligent scheduling model is pre-trained based on historical disaster data, the game relationship of rescue resources, and the optimization objective of group survival entropy.

[0032] The scheduling module is used to send scheduling instructions to the corresponding target emergency rescue nodes according to the dynamic vehicle scheduling strategy and collaborative path planning.

[0033] Furthermore, to achieve the above objectives, this application also proposes an intelligent dispatching device for emergency rescue vehicles based on BeiDou positioning. The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. The computer program is configured to implement the steps of the intelligent dispatching method for emergency rescue vehicles based on BeiDou positioning as described above.

[0034] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the intelligent dispatching method for emergency rescue vehicles based on BeiDou positioning as described above.

[0035] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the intelligent dispatching method for emergency rescue vehicles based on BeiDou positioning as described above.

[0036] One or more technical solutions proposed in this application have at least the following technical effects:

[0037] In traditional emergency rescue vehicle dispatching, which relies heavily on human experience, the dispatch center communicates with vehicles via voice radio and issues commands based on static maps and limited road condition information. This lack of real-time and accurate information regarding the precise location, status, and dynamic situation of all rescue vehicles makes decision-making difficult and data-driven, and the dispatching process is highly dependent on personal experience. In contrast, this application acquires multimodal state perception data transmitted in real-time from multiple emergency rescue nodes. This multimodal state perception data consists of BeiDou high-precision positioning information and spatiotemporal tags. The multimodal state perception data is then input into a pre-set intelligent dispatching model to generate dynamic vehicle dispatching strategies and collaborative path planning. This intelligent dispatching model is pre-trained based on historical disaster data, the game theory relationship of rescue resources, and the optimization objective of group survival entropy. Based on the dynamic vehicle dispatching strategy and collaborative path planning, dispatching instructions are sent to the corresponding target emergency rescue nodes. Understandably, this application adopts a core architecture of "data-driven dynamic game theory simulation". When faced with complex emergency scenarios involving multiple disaster points, resource competition, and changing environments, it provides accurate and real-time global situational input to the decision-making model by integrating spatiotemporal tags from BeiDou high-precision positioning with multimodal perception data. Then, through an intelligent scheduling model based on historical data pre-training and with minimizing "group survival entropy" as the core optimization objective, it conducts dynamic game simulation and deduction of rescue forces, disaster situation, and environment, and automatically solves for vehicle-task allocation and spatiotemporal path planning schemes that approximate Pareto optimality. Finally, through a robust multimodal communication mechanism that combines intent recognition and channel state assessment, the dynamically generated strategy instructions are reliably and accurately delivered to the target rescue nodes. This achieves a fundamental shift in the scheduling paradigm from "experience-driven" to "model and data-driven." Therefore, based on the global optimization model and real-time closed-loop feedback, it is possible to achieve efficient and scientific allocation of limited rescue resources, thereby determining the globally optimal or near-optimal action sequence under spatiotemporal constraints. Ultimately, this achieves the core objective of significantly shortening the average response time and optimizing resource utilization efficiency while improving the overall success rate of rescue efforts, fundamentally solving the technical problem of low scheduling efficiency in emergency multi-concurrency scenarios. Attached Figure Description

[0038] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0039] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1This is a flowchart illustrating an embodiment of the intelligent dispatching method for emergency rescue vehicles based on BeiDou positioning provided in this application.

[0041] Figure 2 This is a flowchart illustrating Embodiment 2 of the intelligent dispatching method for emergency rescue vehicles based on BeiDou positioning provided in this application;

[0042] Figure 3 This is a flowchart illustrating Embodiment 3 of the intelligent dispatching method for emergency rescue vehicles based on BeiDou positioning provided in this application;

[0043] Figure 4 This is a schematic diagram of the module structure of the intelligent dispatching device for emergency rescue vehicles based on BeiDou positioning, as described in an embodiment of this application.

[0044] Figure 5 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the intelligent dispatching method for emergency rescue vehicles based on BeiDou positioning in the embodiments of this application.

[0045] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0046] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0047] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0048] The main solution in this application's embodiments is:

[0049] Acquire multimodal state perception data transmitted in real time from multiple emergency rescue nodes, wherein the multimodal state perception data consists of BeiDou high-precision positioning information and spatiotemporal tags;

[0050] The multimodal state perception data is input into a preset intelligent scheduling model to generate dynamic vehicle scheduling strategies and collaborative path planning. The intelligent scheduling model is pre-trained based on historical disaster data, the game relationship of rescue resources, and the optimization objective of group survival entropy.

[0051] Based on the dynamic vehicle scheduling strategy and collaborative path planning, scheduling instructions are sent to the corresponding target emergency rescue nodes.

[0052] In this embodiment, the application uses an intelligent dispatching device for emergency rescue vehicles based on BeiDou positioning as the executing entity. For ease of description, it will be referred to as "device" in the following detailed description.

[0053] Because current technology relies heavily on human experience, dispatch centers communicate with vehicles via voice radio and issue commands based on static maps and limited road condition information. This makes it impossible to accurately grasp the precise location, status, and dynamic information of all rescue vehicles in real time, resulting in a lack of data support for decision-making and a high dependence on personal experience in the dispatch process.

[0054] This application provides a solution that adopts a "data-driven dynamic game theory simulation" core architecture. When facing complex emergency scenarios with multiple disaster points occurring simultaneously, resource competition, and a changing environment, it integrates spatiotemporal tags from BeiDou high-precision positioning with multimodal perception data to provide the decision-making model with accurate and real-time global situational input. Then, through an intelligent scheduling model based on historical data pre-training and with minimizing "group survival entropy" as the core optimization objective, it conducts dynamic game simulation and deduction of rescue forces, disaster situation, and environment, automatically solving for vehicle-task allocation and spatiotemporal path planning schemes that approximate Pareto optimality. Finally, through a robust multimodal communication mechanism that combines intent recognition and channel state assessment, dynamically generated strategy instructions are reliably and accurately delivered to the target rescue nodes. This achieves a fundamental shift in the scheduling paradigm from "experience-driven" to "model and data-driven." Therefore, based on the global optimization model and real-time closed-loop feedback, it is possible to achieve efficient and scientific allocation of limited rescue resources, thereby determining the globally optimal or near-optimal action sequence under spatiotemporal constraints. Ultimately, this achieves the core objective of significantly shortening the average response time and optimizing resource utilization efficiency while improving the overall success rate of rescue efforts, fundamentally solving the technical problem of low scheduling efficiency in emergency multi-concurrency scenarios.

[0055] Based on this, this application provides an intelligent dispatching method for emergency rescue vehicles based on BeiDou positioning, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the intelligent dispatching method for emergency rescue vehicles based on BeiDou positioning in this application.

[0056] In this embodiment, the intelligent dispatching method for emergency rescue vehicles based on BeiDou positioning includes steps S10 to S30:

[0057] Step S10: Obtain multimodal state perception data transmitted in real time from multiple emergency rescue nodes, wherein the multimodal state perception data consists of BeiDou high-precision positioning information and spatiotemporal tags;

[0058] It should be noted that emergency rescue nodes include, but are not limited to, emergency rescue vehicles, drones, or individual soldier equipment equipped with BeiDou terminals and various sensors. Multimodal state perception data refers to data integrating multiple types, including visual, infrared, audio, and environmental parameters. BeiDou high-precision positioning information refers to sub-meter or centimeter-level position, velocity, and attitude information obtained through means such as the BeiDou ground-based augmentation system. Spatiotemporal tags refer to the precise timestamps and corresponding three-dimensional geographic coordinates attached to each frame of perception data using BeiDou time synchronization.

[0059] Understandably, this step integrates the positioning and timing capabilities of the BeiDou system to unify scattered and heterogeneous on-site perception information into a standardized spatiotemporal framework. This provides a unique and reliable data foundation for the dispatch center to build a panoramic, synchronous, and quantifiable disaster and resource situation map, solving the problems of information isolation and spatiotemporal inconsistency in traditional methods.

[0060] Step S20: Input the multimodal state perception data into a preset intelligent scheduling model to generate a dynamic vehicle scheduling strategy and collaborative path planning. The intelligent scheduling model is pre-trained based on historical disaster data, the game relationship of rescue resources, and the optimization objective of group survival entropy.

[0061] It should be noted that the intelligent scheduling model is a comprehensive decision-making model based on deep reinforcement learning and game theory. Historical disaster data includes the type, scale, evolution, and handling records of past events. Game theory relationships regarding rescue resources are used to model the dynamic competition and cooperation between different rescue forces and their interactions with the disaster's development under limited resource conditions. The population survival entropy optimization objective is a quantitative objective function, the core of which is to assess and minimize the uncertainty of the overall survival risk faced by the affected population over time.

[0062] Understandably, this step utilizes a pre-trained model to place real-time situational data in a dynamic environment simulating multi-party game for computation. By balancing resource efficiency and risk evolution, the model automatically outputs task allocation and path sequences under a global optimization objective, thereby transforming scheduling decisions from relying on static rules and personal experience into a dynamic, adaptive, and quantitatively-oriented intelligent computation process.

[0063] Step S30: Based on the dynamic vehicle scheduling strategy and collaborative path planning, send scheduling instructions to the corresponding target emergency rescue nodes.

[0064] It should be noted that the dynamic vehicle dispatching strategy specifically includes assigning a target disaster point, task sequence, and priority to each vehicle. Collaborative path planning is a spatiotemporal joint trajectory generated for multiple vehicles, taking into account intersection traffic flow, task handover point locations, and avoiding route conflicts. The target emergency rescue node refers to the specific vehicle or equipment unit specified in the strategy and planning.

[0065] Understandably, this step is crucial for implementing the calculated optimal strategy. The generation and issuance of dispatch instructions transform the abstract scheme output by the intelligent model into concrete, executable action commands, driving the distributed rescue nodes on-site to act in a collaborative and optimized manner, thereby ensuring the systematic nature and execution efficiency of the overall rescue plan. This embodiment provides an intelligent dispatch method for emergency rescue vehicles based on BeiDou positioning, employing a core architecture of "data-driven dynamic game theory simulation." When facing complex emergency scenarios involving multiple disaster points, resource competition, and changing environments, it integrates the spatiotemporal tags of BeiDou high-precision positioning with multimodal perception data to provide the decision-making model with accurate and real-time global situational input. Furthermore, through an intelligent dispatch model pre-trained based on historical data and with minimizing "group survival entropy" as the core optimization objective, it performs dynamic game simulation and deduction of rescue forces, disaster situation, and environment, automatically solving for a vehicle-task allocation and spatiotemporal path planning scheme approaching Pareto optimality. Finally, through a robust multimodal communication mechanism combining intent recognition and channel state assessment, the dynamically generated strategy instructions are reliably and accurately delivered to the target rescue nodes. This achieves a fundamental shift in the scheduling paradigm from "experience-driven" to "model and data-driven." Therefore, based on the global optimization model and real-time closed-loop feedback, it is possible to achieve efficient and scientific allocation of limited rescue resources, thereby determining the globally optimal or near-optimal action sequence under spatiotemporal constraints. Ultimately, this achieves the core objective of significantly shortening the average response time and optimizing resource utilization efficiency while improving the overall success rate of rescue efforts, fundamentally solving the technical problem of low scheduling efficiency in emergency multi-concurrency scenarios.

[0066] In one feasible implementation, the step of inputting the multimodal state perception data into a preset intelligent scheduling model to generate a dynamic vehicle scheduling strategy and cooperative path planning includes:

[0067] Based on the multimodal state perception data, it is input into the dynamic game environment model in the preset intelligent scheduling model;

[0068] Based on the dynamic game environment model, the rescue effectiveness weight of each emergency rescue node for different disaster points and the group survival entropy change rate of each disaster point are calculated. The rescue effectiveness weight is calculated by comprehensively considering the node type, real-time location, equipment status and historical efficiency. The group survival entropy change rate is calculated based on the scale of trapped personnel, environmental threat level and time decay factor.

[0069] With the optimization objective of minimizing the weighted sum of global group survival entropy and total rescue time, dynamic deduction is performed in the dynamic game environment model to obtain the optimal vehicle task allocation matrix and spatiotemporal path sequence, which serve as the dynamic vehicle scheduling strategy and collaborative path planning.

[0070] It should be noted that the dynamic game environment model is a computer simulation environment used to formally describe the dynamic interactions and strategic dependencies among multiple participants, including rescue nodes, disaster evolution, and road networks. The rescue effectiveness weight is a quantitative indicator representing the comprehensive potential of a specific node in handling a specific disaster point. The population survival entropy change rate is a physical information indicator used to measure the rate of deterioration of the overall survival risk of a trapped population at a disaster point per unit time. The optimization objective function is the mathematical criterion upon which the scheduling model is solved, integrating costs from different dimensions through weighted aggregation. The vehicle task allocation matrix is ​​a two-dimensional data structure whose elements identify the assignment relationship between a specific node and a specific disaster point. The spatiotemporal path sequence is a series of time-constrained geographical coordinates planned for each assigned node.

[0071] Understandably, this implementation transforms the scheduling problem into a computable multi-objective optimization problem by constructing a dynamic game environment. By calculating efficiency weights and entropy change rate, the urgency of the disaster and the need for rescue capabilities are quantified. Based on this, deduction and solution can systematically output a scheduling scheme that comprehensively considers efficiency and risk under complex constraints. This shifts the decision-making process from subjective experience-based judgment to objective model calculation, thereby achieving global approximate optimality in resource allocation and path planning under emergency concurrent scenarios.

[0072] For example, refer to Figure 2In a simulated urban earthquake rescue scenario, there are 3 rescue nodes (medical vehicle A, engineering vehicle B, and transport vehicle C) and 2 disaster sites (P1 school and P2 factory). After receiving real-time data, the dynamic game environment model initiates a round of simulation. First, the model calculates the rescue effectiveness weight WA->P1 of node A to disaster site P1 as 0.85 based on the formula WA->P1 = 0.6*(medical equipment coefficient) + 0.3*(BeiDou inverse distance weight from A to P1) - 0.1*(current task load of A). Simultaneously, based on sensor data, it is assessed that there are 50 trapped teachers and students at P1, the building is shaky, the environmental threat level is 0.9, 30 minutes have passed since the earthquake, and the time decay factor is set to 0.1. The calculated group survival entropy change rate of P1 is 50 * 0.9 * exp(-0.1*30) ≈ 2.24. The optimization objective is set to minimize (total survival entropy change rate * 0.7 + total estimated time * 0.3). The model solves iteratively and may output a task assignment matrix of [[A->P1], [B->P2], [C->P1]], and plan a spatiotemporal path sequence for vehicle A: [(t0, location A), (t0+5min, intersection X), (t0+12min, intersection Y), (t0+18min, school P1)].

[0073] In one feasible implementation, the step of acquiring multimodal state perception data transmitted in real time from multiple emergency rescue nodes includes:

[0074] A node agent training framework based on federated causal perception is constructed, wherein each of the emergency rescue nodes acts as a distributed agent;

[0075] Each of the aforementioned intelligent agents trains a local causal reasoning sub-model based on local historical perception data, which is used to identify potential causal relationships between key state features and rescue outcomes;

[0076] Each of the aforementioned intelligent agents uploads the lightweight causal primitives obtained from training to the scheduling center for secure aggregation, generating a global causal knowledge graph;

[0077] Based on the global causal knowledge graph, the intelligent scheduling model is trained through knowledge distillation and reinforcement learning to improve the interpretability and predictive ability of its policy generation.

[0078] It's important to note that the Federated Causal Perception Node Agent Training Framework is a distributed machine learning architecture that coordinates multiple terminals for collaborative model training while protecting local data privacy. Distributed agents refer to edge entities with independent data collection, storage, and computation capabilities, such as emergency rescue vehicles. Local causal inference sub-models are lightweight algorithms deployed on agents to discover causal relationships between variables from local data. Lightweight causal primitives are an abstract, compressed data representation of local causal relationships. Secure aggregation is a cryptographic or differential privacy technique used to protect the data privacy of each participant when aggregating multi-party data. The global causal knowledge graph is a structured knowledge base built by aggregating all primitives, characterizing key causal relationships across the entire domain. Knowledge distillation is a model compression and transfer technique, specifically referring to the process of injecting the causal logic rules inherent in the knowledge graph into an intelligent scheduling model. Reinforcement learning training is a machine learning method that optimizes policies by having agents interact with simulated environments and receive reward signals.

[0079] Understandably, this implementation method, through a federated learning framework, comprehensively utilizes historical experience data distributed across various rescue nodes to uncover causal patterns without centralizing raw data. Constructing a global causal knowledge graph provides interpretable prior knowledge for the scheduling model. Through knowledge distillation and reinforcement learning training incorporating causal constraints, the intelligent scheduling model can be guided to learn decision-making strategies consistent with real-world causal logic, thereby enhancing the reliability of its decision-making basis and its ability to generalize and predict complex disaster scenarios it has never experienced before.

[0080] For example, a node agent (such as an edge computing module inside a fire truck) stores logs of 200 tasks executed over the past year. It analyzes local data using a constraint-based causal discovery algorithm (such as FCI) and finds a strong statistical dependency between "water tank level below 40%" and "decreased firefighting success rate," eliminating confusion caused by "task distance." It preliminarily determines this to be a direct causal relationship, forming a local causal graph primitive (Cause: Water level <40%, Result: Success rate ↓, Confidence: 0.82). This primitive is uploaded after homomorphic encryption. After the dispatch center aggregates similar primitives from 100 nodes, the causal edge weight between node "equipment status" and node "task effectiveness" in the global causal knowledge graph is updated to 0.78. Subsequently, during the training of the main dispatch model, this graph is used for knowledge distillation: when the model decides to send a fire truck with only 35% water tank capacity to a new fire scene, the knowledge distillation loss function generates a large penalty value, prompting the model to adjust its strategy and select another fully equipped vehicle.

[0081] In one feasible implementation, the step of performing knowledge distillation and reinforcement learning training on the intelligent scheduling model based on the global causal knowledge graph includes:

[0082] High-frequency causal rules are extracted from the global causal knowledge graph and transformed into soft constraints for the scheduling strategy.

[0083] During the training process of the game deduction and scheduling model, a policy distillation loss function based on the soft constraint is introduced to guide the model to learn a policy space that conforms to causal priors.

[0084] In a pre-defined simulated disaster environment scenario, a pre-defined near-end policy optimization algorithm is used to conduct adversarial training on a model that incorporates causal constraints, with the success rate and efficiency of rescue as reward signals. This enables the model to stably output robust policies in complex games.

[0085] It should be noted that high-frequency causal rules refer to causal relationship patterns that occur more frequently than a preset threshold, statistically derived from the global causal knowledge graph. Soft constraints are penalty terms introduced into the model optimization objective to encourage, but not force, the model's generated policies to align with causal rules. The policy distillation loss function is a specific loss function used to calculate the difference between the model's current policy and causal prior knowledge. Proximal policy optimization is a reinforcement learning algorithm characterized by ensuring the stability of the training process by limiting the step size of policy updates. In this context, adversarial training refers to the process of stress-testing and optimizing the model under highly challenging and volatile simulated catastrophic environments.

[0086] Understandably, this implementation incorporates statistical causal rules into model training as soft constraints, enabling the data-driven model to absorb human-understandable logical knowledge. The policy distillation loss function guides the model to prioritize exploring policy regions consistent with causal experience within a broad decision space. Combined with adversarial training of proximal policy optimization in a complex simulation environment, the model ultimately learns highly robust scheduling strategies that conform to objective causal laws and adapt to dynamic game dynamics, avoiding the illogical decisions that might arise from purely data-driven approaches.

[0087] For example, a high-frequency rule extracted from the global causal knowledge graph is: "If the base station damage rate in the target area is >60%, then a reconnaissance node that relies on real-time video transmission should be dispatched, increasing its mission failure probability." This rule is transformed into a soft constraint: add a term -η * max(0, base station damage rate - 0.6) * I(node ​​type = reconnaissance) to the model's reward function, where η is the penalty coefficient and I() is the indicator function. In near-end policy optimization (PPO) training, various disaster scenarios are dynamically generated in the simulated environment, such as "a large-scale paralysis of base stations in the eastern part of the city after a magnitude 7 earthquake." The model (agent) may still send reconnaissance drones to the east at the beginning, but due to the aforementioned causal constraint penalty, the round reward is reduced. After multiple policy updates, the model learns that in areas with high base station damage, nodes that do not rely on continuous high-bandwidth communication (such as vehicles carrying offline mapping equipment) should be prioritized, thus making more robust decisions in subsequent similar scenarios.

[0088] In one feasible implementation, the step of sending dispatch instructions to the corresponding target emergency rescue node according to the dynamic vehicle dispatch strategy and cooperative path planning includes:

[0089] The optimal communication link combination is dynamically selected based on the key intent level of the scheduling instruction and the real-time channel state entropy of the target node. The key intent level is comprehensively evaluated by the spatiotemporal sensitivity, action necessity and information complexity of the instruction, and the channel state entropy comprehensively evaluates the uncertainty of link stability, bandwidth and delay.

[0090] For survivability-level commands, transmission is prioritized via the BeiDou short message communication link, and forward error correction coding and duplicate confirmation mechanisms are employed.

[0091] For situational data that is not survivability-level but has high bandwidth requirements, it is transmitted through a backup link pre-configured based on digital twin network simulation. The backup link is a converged communication channel established in advance based on the network twin simulation results.

[0092] It should be noted that the critical intent level is a classification and quantification of the criticality of scheduling instructions, including but not limited to survivability level, critical level, and routine level. Channel state entropy is an information-theoretic metric used to quantify the uncertainty and unreliability of the current state of a communication link. The optimal communication link combination refers to a set of transmission paths selected from all available links to achieve the best balance between reliability and efficiency for transmitting a specific instruction. Forward error correction coding is a coding technique that adds redundant check information before data transmission, used to automatically detect and correct transmission errors at the receiving end. Digital twin network simulation refers to the technique of constructing a virtual mapping synchronized with the physical communication network and extrapolating and predicting the future network state within it. The network twin is the core virtual model of digital twin network simulation.

[0093] Understandably, this implementation achieves adaptive, on-demand allocation of communication resources through a dual assessment of intent level and channel entropy. For survivability-level commands crucial to core operations, the highly reliable BeiDou short message service, combined with enhanced coding, ensures command delivery even under extreme conditions. For large-volume data, the predictive capabilities of network twins are utilized to pre-establish optimized pathways, guaranteeing efficient transmission. This hierarchical and categorized communication strategy, in disaster environments where overall communication resources are limited, maximizes the reliability and timeliness of the command and information flows of the scheduling system.

[0094] For example, refer to Figure 3 The system generates an instruction: "Order medical team M to immediately proceed to latitude and longitude (116.5, 39.8) to provide emergency care; the injured person is experiencing severe bleeding." This instruction is rated as survival-level (spatiotemporal sensitivity = 10 / 10, action necessity = 10 / 10, low information complexity). Simultaneously, the system detects a high channel state entropy (violent fluctuations) for the 5G signal in the target area, while the BeiDou short message link entropy is stable. Based on the strategy, the system selects BeiDou short message as the optimal communication link. Before transmission, the instruction data packet undergoes forward error correction encoding using RS(255, 223) code, adding a 32-byte checksum. This encoded packet is sent three times consecutively. After receiving at least one complete data packet, node M decodes and verifies it, then replies with a "receive confirmation" message via the same link. For another high-definition panoramic mosaic image of the disaster area that needs to be distributed (not survivability-grade, but with high bandwidth requirements), the digital twin network simulation predicted that a certain Ku band of the satellite link would become congested 10 minutes later. Therefore, the system established a converged channel in advance through the SDN controller to transmit the data stream through the aggregated link of "5G (primary) + backup satellite C band (secondary)".

[0095] In one feasible implementation, the step of sending the dispatch instruction to the corresponding target emergency rescue node further includes:

[0096] A unique digital thread identifier is generated for each rescue mission, and data from triggering, decision-making, execution to completion is recorded throughout the entire lifecycle.

[0097] The dynamic vehicle scheduling strategy, collaborative path planning, actual vehicle trajectory and status feedback data, and the final task results are correlated and compared along the digital thread.

[0098] Based on the comparison results, the strategy deviation and comprehensive effectiveness index of this task are calculated and fed back as incremental training data to the game inference scheduling model for online fine-tuning and continuous optimization of the model.

[0099] It should be noted that the digital thread identifier is a unique data index identifier that spans the entire lifecycle of a single task. Lifecycle data includes alarm information, all intermediate decision versions, all instruction and status records, and the final handling report. Strategy deviation is an indicator that quantifies the difference between the actual execution trajectory and the planned path. The comprehensive performance index is a performance score calculated by integrating multiple dimensions such as task completion time, resource consumption, and rescue effectiveness. Incremental training data refers to a small-scale dataset composed of newly generated task data used for further model training.

[0100] Understandably, this implementation method, by constructing a task digital thread, achieves end-to-end traceability and refined review of the effects of each scheduling decision and execution. By comparing planning and execution, the accuracy of the strategy and the unpredictability of the environment can be quantitatively analyzed. The calculated deviation and efficiency index provide a basis for objectively evaluating the performance of the scheduling model. Using these data as incremental feedback inputs to the model for online learning forms a complete closed loop from decision-making to execution to model optimization, enabling the intelligent scheduling system to continuously learn and evolve from actual rescue experience.

[0101] For example, in a rescue mission for a "chemical industrial park leak and fire," the system generated a digital thread ID FIRE-20231027-0845-7a3b9c. Throughout the entire lifecycle, the initial planned path was [Gate, R1, R2, Site]. During actual execution, due to a landslide on road R1, the vehicle's trajectory changed to [Gate, R3, R4, Site]. After the mission, the system calculated the strategy deviation: using a dynamic time warping algorithm, the minimum alignment distance between the actual trajectory point sequence and the planned path sequence was calculated to be 2.5 km, with a total planned path length of 8 km, resulting in a deviation of approximately 31%. The comprehensive efficiency index was calculated as follows: successful evacuation of 20 people * 10 minutes / person = 200, minus the actual time of 120 minutes * 0.5 = 60, and then minus the actual distance of 10 km * 0.2 = 2, yielding an efficiency index of 138. The data set {Scenario characteristics: chemical fire, road collapse, deviation: 0.31, efficiency index: 138} was tagged and then sent to the online fine-tuning module of the scheduling model. The module uses this data as a small batch, with the auxiliary objective of reducing planning deviation in similar "fire + road interruption" scenarios in the future, to perform a gradient descent update on the model parameters, achieving closed-loop optimization.

[0102] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the intelligent dispatching method for emergency rescue vehicles based on Beidou positioning in this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0103] This application also provides an intelligent dispatching device for emergency rescue vehicles based on BeiDou positioning. Please refer to [reference needed]. Figure 4 The BeiDou-based intelligent dispatching device for emergency rescue vehicles includes:

[0104] The acquisition module 10 is used to acquire multimodal state perception data transmitted back in real time by multiple emergency rescue nodes, wherein the multimodal state perception data consists of Beidou high-precision positioning information and spatiotemporal tags;

[0105] The generation module 20 is used to input the multimodal state perception data into a preset intelligent scheduling model to generate dynamic vehicle scheduling strategies and collaborative path planning. The intelligent scheduling model is obtained by pre-training based on historical disaster data, the game relationship of rescue resources, and the optimization objective of group survival entropy.

[0106] The scheduling module 30 is used to send scheduling instructions to the corresponding target emergency rescue nodes according to the dynamic vehicle scheduling strategy and collaborative path planning.

[0107] And / or, the BeiDou-based intelligent dispatching device for emergency rescue vehicles includes:

[0108] The first input module is used to input the multimodal state perception data into the dynamic game environment model in the preset intelligent scheduling model.

[0109] The first calculation module is used to calculate the rescue effectiveness weight of each emergency rescue node for different disaster points, and the group survival entropy change rate of each disaster point, based on the dynamic game environment model. The rescue effectiveness weight is calculated by comprehensively considering the node type, real-time location, equipment status and historical efficiency, and the group survival entropy change rate is calculated based on the scale of trapped personnel, environmental threat level and time decay factor.

[0110] The first deduction module is used to perform dynamic deduction in the dynamic game environment model with the optimization objective of minimizing the weighted sum of global group survival entropy and total rescue time, and solve for the optimal vehicle task allocation matrix and spatiotemporal path sequence, which serve as the dynamic vehicle scheduling strategy and collaborative path planning.

[0111] And / or, the BeiDou-based intelligent dispatching device for emergency rescue vehicles includes:

[0112] The first construction module is used to build a node intelligent agent training framework based on federated causal perception, wherein each of the emergency rescue nodes is a distributed intelligent agent;

[0113] The first training module is used by each of the aforementioned intelligent agents to train a local causal reasoning sub-model based on local historical perception data, in order to identify potential causal relationships between key state features and rescue results.

[0114] The first upload module is used by each of the intelligent agents to upload the lightweight causal primitives obtained from training to the scheduling center for secure aggregation and to generate a global causal knowledge graph.

[0115] The first distillation module is used to perform knowledge distillation and reinforcement learning training on the intelligent scheduling model based on the global causal knowledge graph, so as to improve the interpretability and predictive ability of its strategy generation.

[0116] And / or, the BeiDou-based intelligent dispatching device for emergency rescue vehicles includes:

[0117] The first transformation module is used to extract high-frequency causal rules from the global causal knowledge graph and transform them into soft constraints of the scheduling strategy.

[0118] The first introduction module is used to introduce a policy distillation loss function based on the soft constraint conditions during the training process of the game deduction and scheduling model, so as to guide the model to learn a policy space that conforms to causal priors.

[0119] The first adversarial module is used to train a model that incorporates causal constraints in a pre-set simulated disaster environment scenario. This is achieved by using a pre-set near-end policy optimization algorithm with the success rate and efficiency of rescue as reward signals. The goal is to enable the model to stably output robust strategies in complex games.

[0120] And / or, the BeiDou-based intelligent dispatching device for emergency rescue vehicles includes:

[0121] The first selection module is used to dynamically select the optimal communication link combination based on the key intent level of the scheduling instruction and the real-time channel state entropy of the target node. The key intent level is comprehensively evaluated by the spatiotemporal sensitivity, action necessity and information complexity of the instruction, and the channel state entropy comprehensively evaluates the uncertainty of link stability, bandwidth and delay.

[0122] The first transmission module is used to prioritize the transmission of survivability-level commands via the BeiDou short message communication link, and employs forward error correction coding and duplicate confirmation mechanisms.

[0123] The second transmission module is used to transmit situational data that is not survivable but has high bandwidth requirements through a backup link pre-configured based on digital twin network simulation. The backup link is a converged communication channel established in advance based on the network twin simulation results.

[0124] And / or, the BeiDou-based intelligent dispatching device for emergency rescue vehicles includes:

[0125] The first recording module is used to generate a unique digital thread identifier for each rescue mission and record the entire lifecycle data from triggering, decision-making, execution to completion;

[0126] The first scheduling module is used to associate and compare the dynamic vehicle scheduling strategy with the collaborative path planning, the actual vehicle trajectory and status feedback data, and the final task results along the digital thread.

[0127] The second calculation module is used to calculate the strategy deviation and comprehensive efficiency index of this task based on the comparison results, and feed them back as incremental training data to the game inference scheduling model for online fine-tuning and continuous optimization of the model.

[0128] The intelligent dispatching device for emergency rescue vehicles based on BeiDou positioning provided in this application adopts the intelligent dispatching method for emergency rescue vehicles based on BeiDou positioning in the above embodiments, which can solve the technical problem of low dispatching efficiency of emergency rescue vehicles in emergency multi-concurrency scenarios. Compared with the prior art, the beneficial effects of the intelligent dispatching device for emergency rescue vehicles based on BeiDou positioning provided in this application are the same as the beneficial effects of the intelligent dispatching method for emergency rescue vehicles based on BeiDou positioning provided in the above embodiments, and other technical features in the intelligent dispatching device for emergency rescue vehicles based on BeiDou positioning are the same as the features disclosed in the methods of the above embodiments, and will not be repeated here.

[0129] This application provides an intelligent dispatching device for emergency rescue vehicles based on BeiDou positioning. The intelligent dispatching device for emergency rescue vehicles based on BeiDou positioning includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the intelligent dispatching method for emergency rescue vehicles based on BeiDou positioning in the above embodiment 1.

[0130] The following is for reference. Figure 5 The diagram illustrates a structural schematic of an intelligent dispatching device for emergency rescue vehicles based on BeiDou positioning, suitable for implementing embodiments of this application. The intelligent dispatching device for emergency rescue vehicles based on BeiDou positioning in these embodiments may include, but is not limited to, mobile terminals such as mobile phones, tablets, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PMPs (Portable Media Players), and vehicle-mounted terminals (e.g., vehicle navigation terminals), as well as fixed terminals such as digital televisions and desktop computers. Figure 5 The intelligent dispatching device for emergency rescue vehicles based on BeiDou positioning shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments in this application.

[0131] like Figure 5 As shown, the BeiDou-based intelligent dispatching device for emergency rescue vehicles may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to programs stored in read-only memory (ROM) 1002 or programs loaded from storage device 1003 into random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the BeiDou-based intelligent dispatching device for emergency rescue vehicles. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the BeiDou-based intelligent dispatching equipment for emergency rescue vehicles to exchange data wirelessly or via wired communication with other devices. Although the figure shows an intelligent dispatching equipment for emergency rescue vehicles based on BeiDou positioning with various systems, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems can be implemented alternatively.

[0132] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0133] The intelligent dispatching device for emergency rescue vehicles based on BeiDou positioning provided in this application adopts the intelligent dispatching method for emergency rescue vehicles based on BeiDou positioning in the above embodiments, which can solve the technical problem of low dispatching efficiency of emergency rescue vehicles in emergency multi-concurrency scenarios. Compared with the prior art, the beneficial effects of the intelligent dispatching device for emergency rescue vehicles based on BeiDou positioning provided in this application are the same as the beneficial effects of the intelligent dispatching method for emergency rescue vehicles based on BeiDou positioning provided in the above embodiments, and other technical features in the intelligent dispatching device for emergency rescue vehicles based on BeiDou positioning are the same as the features disclosed in the method of the previous embodiment, and will not be repeated here.

[0134] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0135] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0136] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, which are used to execute the BeiDou positioning-based intelligent dispatching method for emergency rescue vehicles in the above embodiments.

[0137] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0138] The aforementioned computer-readable storage medium may be included in the BeiDou-based intelligent dispatching equipment for emergency rescue vehicles; or it may exist independently and not be installed in the BeiDou-based intelligent dispatching equipment for emergency rescue vehicles.

[0139] The aforementioned computer-readable storage medium carries one or more programs. When the aforementioned one or more programs are executed by the BeiDou-based intelligent dispatching device for emergency rescue vehicles, the BeiDou-based intelligent dispatching device for emergency rescue vehicles acquires multimodal state perception data transmitted back in real time by multiple emergency rescue nodes, wherein the multimodal state perception data consists of BeiDou high-precision positioning information and spatiotemporal tags.

[0140] The multimodal state perception data is input into a preset intelligent scheduling model to generate dynamic vehicle scheduling strategies and collaborative path planning. The intelligent scheduling model is pre-trained based on historical disaster data, the game relationship of rescue resources, and the optimization objective of group survival entropy.

[0141] Based on the dynamic vehicle scheduling strategy and collaborative path planning, scheduling instructions are sent to the corresponding target emergency rescue nodes.

[0142] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0143] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0144] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0145] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described intelligent dispatching method for emergency rescue vehicles based on BeiDou positioning. This addresses the technical problem of low dispatching efficiency for emergency rescue vehicles in multi-concurrency emergency scenarios. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the intelligent dispatching method for emergency rescue vehicles based on BeiDou positioning provided in the above embodiments, and will not be elaborated upon here.

[0146] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described intelligent dispatching method for emergency rescue vehicles based on BeiDou positioning.

[0147] The computer program product provided in this application can solve the technical problem of low dispatch efficiency of emergency rescue vehicles in emergency multi-concurrency scenarios. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the intelligent dispatch method for emergency rescue vehicles based on Beidou positioning provided in the above embodiments, and will not be repeated here.

[0148] All acquisition of signals, information, or actions in this application are carried out in compliance with the relevant data protection laws and policies of the country where the application is located, and with the authorization of the relevant device owner.

[0149] The above description is only a part of the embodiments of this application and does not limit the scope of protection of this application. All equivalent structural transformations made under the technical concept of this application and using the content of this application specification and drawings, or direct / indirect applications in other related technical fields, are included in the scope of protection of this application.

Claims

1. A method for intelligent dispatching of emergency rescue vehicles based on BeiDou positioning, characterized in that, The method includes: Acquire multimodal state perception data transmitted in real time from multiple emergency rescue nodes, wherein the multimodal state perception data consists of BeiDou high-precision positioning information and spatiotemporal tags; The step of acquiring multimodal state perception data transmitted in real time from multiple emergency rescue nodes includes the following: A node agent training framework based on federated causal perception is constructed, wherein each of the emergency rescue nodes acts as a distributed agent; Each of the aforementioned intelligent agents trains a local causal reasoning sub-model based on local historical perception data, which is used to identify potential causal relationships between key state features and rescue outcomes; Each of the aforementioned intelligent agents uploads the lightweight causal primitives obtained from training to the scheduling center for secure aggregation, generating a global causal knowledge graph; Based on the global causal knowledge graph, the intelligent scheduling model is trained by knowledge distillation and reinforcement learning to improve the interpretability and predictive ability of its strategy generation. The multimodal state perception data is input into a preset intelligent scheduling model to generate dynamic vehicle scheduling strategies and collaborative path planning. The intelligent scheduling model is pre-trained based on historical disaster data, the game relationship of rescue resources, and the optimization objective of group survival entropy. The step of inputting the multimodal state perception data into a preset intelligent scheduling model to generate dynamic vehicle scheduling strategies and cooperative path planning includes: Based on the multimodal state perception data, it is input into the dynamic game environment model in the preset intelligent scheduling model; Based on the dynamic game environment model, the rescue effectiveness weight of each emergency rescue node for different disaster points and the group survival entropy change rate of each disaster point are calculated. The rescue effectiveness weight is calculated by comprehensively considering the node type, real-time location, equipment status and historical efficiency. The group survival entropy change rate is calculated based on the scale of trapped personnel, environmental threat level and time decay factor. With the goal of minimizing the weighted sum of global group survival entropy and total rescue time, dynamic deduction is performed in the dynamic game environment model to obtain the optimal vehicle task allocation matrix and spatiotemporal path sequence, which serve as the dynamic vehicle scheduling strategy and collaborative path planning. Based on the dynamic vehicle dispatching strategy and collaborative path planning, dispatching instructions are sent to the corresponding target emergency rescue nodes. The step of sending dispatch instructions to the corresponding target emergency rescue node according to the dynamic vehicle dispatch strategy and cooperative path planning includes: The optimal communication link combination is dynamically selected based on the key intent level of the scheduling instruction and the real-time channel state entropy of the target node. The key intent level is comprehensively evaluated by the spatiotemporal sensitivity, action necessity and information complexity of the instruction, and the channel state entropy comprehensively evaluates the uncertainty of link stability, bandwidth and delay. For survivability-level commands, transmission is prioritized via the BeiDou short message communication link, and forward error correction coding and duplicate confirmation mechanisms are employed. For situational data that is not survivability-level but has high bandwidth requirements, it is transmitted through a backup link pre-configured based on digital twin network simulation. The backup link is a converged communication channel established in advance based on the network twin simulation results.

2. The method as described in claim 1, characterized in that, The steps of performing knowledge distillation and reinforcement learning training on the intelligent scheduling model based on the global causal knowledge graph include: High-frequency causal rules are extracted from the global causal knowledge graph and transformed into soft constraints for the scheduling strategy. During the training process of the dynamic game environment model, a policy distillation loss function based on the soft constraints is introduced to guide the model to learn a policy space that conforms to causal priors. In a pre-defined simulated disaster environment scenario, a pre-defined near-end policy optimization algorithm is used to conduct adversarial training on a model that incorporates causal constraints, with the success rate and efficiency of rescue as reward signals. This enables the model to stably output robust policies in complex games.

3. The method as described in claim 1, characterized in that, Following the step of sending the dispatch instruction to the corresponding target emergency rescue node, the method further includes: A unique digital thread identifier is generated for each rescue mission, and data from triggering, decision-making, execution to completion is recorded throughout the entire lifecycle. The dynamic vehicle scheduling strategy, collaborative path planning, actual vehicle trajectory and status feedback data, and the final task results are correlated and compared along the digital thread. Based on the comparison results, the strategy deviation and comprehensive effectiveness index of this task are calculated, and these are fed back as incremental training data to the dynamic game environment model for online fine-tuning and continuous optimization of the model.

4. An intelligent dispatching device for emergency rescue vehicles based on BeiDou positioning, characterized in that, The device includes: The acquisition module is used to acquire multimodal state perception data transmitted back in real time by multiple emergency rescue nodes, wherein the multimodal state perception data consists of Beidou high-precision positioning information and spatiotemporal tags; The process of acquiring multimodal state perception data transmitted in real time from multiple emergency rescue nodes includes: A node agent training framework based on federated causal perception is constructed, wherein each of the emergency rescue nodes acts as a distributed agent; Each of the aforementioned intelligent agents trains a local causal reasoning sub-model based on local historical perception data, which is used to identify potential causal relationships between key state features and rescue outcomes; Each of the aforementioned intelligent agents uploads the lightweight causal primitives obtained from training to the scheduling center for secure aggregation, generating a global causal knowledge graph; Based on the global causal knowledge graph, the intelligent scheduling model is trained by knowledge distillation and reinforcement learning to improve the interpretability and predictive ability of its strategy generation. The generation module is used to input the multimodal state perception data into a preset intelligent scheduling model to generate dynamic vehicle scheduling strategies and collaborative path planning. The intelligent scheduling model is pre-trained based on historical disaster data, the game relationship of rescue resources, and the optimization objective of group survival entropy. The step of inputting the multimodal state perception data into a preset intelligent scheduling model to generate dynamic vehicle scheduling strategies and cooperative path planning includes: Based on the multimodal state perception data, it is input into the dynamic game environment model in the preset intelligent scheduling model; Based on the dynamic game environment model, the rescue effectiveness weight of each emergency rescue node for different disaster points and the group survival entropy change rate of each disaster point are calculated. The rescue effectiveness weight is calculated by comprehensively considering the node type, real-time location, equipment status and historical efficiency. The group survival entropy change rate is calculated based on the scale of trapped personnel, environmental threat level and time decay factor. With the goal of minimizing the weighted sum of global group survival entropy and total rescue time, dynamic deduction is performed in the dynamic game environment model to obtain the optimal vehicle task allocation matrix and spatiotemporal path sequence, which serve as the dynamic vehicle scheduling strategy and collaborative path planning. The scheduling module is used to send scheduling instructions to the corresponding target emergency rescue nodes according to the dynamic vehicle scheduling strategy and collaborative path planning. The step of sending dispatch instructions to the corresponding target emergency rescue node based on the dynamic vehicle dispatch strategy and collaborative path planning includes: The optimal communication link combination is dynamically selected based on the key intent level of the scheduling instruction and the real-time channel state entropy of the target node. The key intent level is comprehensively evaluated by the spatiotemporal sensitivity, action necessity and information complexity of the instruction, and the channel state entropy comprehensively evaluates the uncertainty of link stability, bandwidth and delay. For survivability-level commands, transmission is prioritized via the BeiDou short message communication link, and forward error correction coding and duplicate confirmation mechanisms are employed. For situational data that is not survivability-level but has high bandwidth requirements, it is transmitted through a backup link pre-configured based on digital twin network simulation. The backup link is a converged communication channel established in advance based on the network twin simulation results.

5. An intelligent dispatching device for emergency rescue vehicles based on BeiDou positioning, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the intelligent dispatching method for emergency rescue vehicles based on BeiDou positioning as described in any one of claims 1 to 3.

6. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the intelligent dispatching method for emergency rescue vehicles based on Beidou positioning as described in any one of claims 1 to 3.

7. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the steps of the intelligent dispatching method for emergency rescue vehicles based on BeiDou positioning as described in any one of claims 1 to 3.

Images