An emergency power intelligent scheduling and path planning method and system

Abstract

This invention discloses an intelligent scheduling and path planning method and system for emergency power supplies, belonging to the field of emergency resource scheduling and management. It includes configuring multiple emergency power supplies as power agent networks and constructing a two-layer data ledger; generating dynamic task contracts based on energy demand information; obtaining global state information from the state chain and negotiating the dynamic task contracts with energy demand information to generate negotiation results; constructing a dynamic energy potential field based on energy demand information and negotiation results; planning movement paths based on the dynamic energy potential field; and updating the state chain with the estimated trajectory and real-time location as new dynamic state information and performing dynamic path correction. By configuring multiple emergency power supplies as power agent networks to form a decentralized network and using a two-layer data ledger, dynamic task contract negotiation, and dynamic energy potential field for scheduling and path planning, this technical solution enables efficient and adaptive scheduling management and optimized allocation of emergency power resources.

Description

Technical Field

[0001] This invention relates to the field of emergency resource scheduling and management, and in particular to an intelligent scheduling and path planning method and system for emergency power supplies. Background Technology

[0002] Emergency power supplies, as a vital strategic resource, provide power assurance for the normal operation of critical infrastructure, commercial centers, and residential areas during emergencies such as natural disasters, major accidents, or power grid failures. They play an irreplaceable role in maintaining social stability and minimizing economic losses. Therefore, establishing an efficient and reliable emergency power dispatch and management system is a crucial component of modern urban emergency management and business continuity planning.

[0003] In related technologies, Chinese invention patent application CN119990652A discloses an emergency resource scheduling and path planning method, including: constructing a dynamic weight evaluation system based on emergency resource scheduling factors, wherein the weight evaluation system can dynamically calculate and adjust the factor weights to determine the most critical resources to be scheduled to the most needed locations in an emergency; constructing a deep reinforcement learning model, using historical emergency response data as a training set, so that the agent can learn and optimize the strategy of selecting the optimal path under different conditions in simulated or historical emergency scenarios.

[0004] The core shortcomings of the aforementioned technologies lie in the rigidity of the evaluation system and the static nature of path planning. Traditional emergency resource allocation methods rely on fixed weighted evaluation systems, which struggle to accurately reflect the dynamic changes in factors such as time urgency, resource demand urgency, traffic congestion, weather impact, disaster severity, affected population size, and road capacity during emergency response. This limits the optimization of resource allocation strategies. Secondly, in terms of path planning, traditional algorithms are often based on static road networks and fixed traffic conditions, making it difficult to cope with unforeseen circumstances such as traffic congestion and road closures that may occur in actual emergency responses. This can lead to suboptimal planned paths, or even failure to reach the target location, thus impacting rescue efficiency. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides an intelligent scheduling and path planning method and system for emergency power supplies. It employs a decentralized network of multiple emergency power supplies as intelligent power agents, and utilizes a two-layer data ledger, dynamic task contract negotiation, and dynamic energy potential field for scheduling and path planning. This approach enables efficient, reliable, and adaptive scheduling management and optimized allocation of emergency power resources.

[0006] The above objectives can be achieved through the following approach:

[0007] An intelligent scheduling and path planning method for emergency power supplies includes configuring multiple emergency power supplies as power agent intelligent agents to form a decentralized power agent intelligent agent network, and constructing a two-layer data ledger containing an identity chain and a state chain, wherein the identity chain is used to record the identity information and historical reputation value of each power agent intelligent agent, and the state chain is used to record the dynamic state information of the power agent intelligent agents in real time; acquiring energy demand information, and generating a dynamic task contract containing task objectives and incentive weights based on the energy demand information; obtaining global state information from the state chain through the power agents intelligent agents, and negotiating the dynamic task contract in combination with locally perceived energy demand information to generate a negotiation result containing role allocation and deployment location; constructing a dynamic energy potential field based on the energy demand information and the deployment location in the negotiation result; planning a movement path for the power agents intelligently based on the dynamic energy potential field; during the movement, updating the state chain with the estimated trajectory and real-time position of the power agents intelligently as new dynamic state information, and performing dynamic path correction based on the information in the state chain.

[0008] Optionally, constructing a two-layer data ledger comprising an identity chain and a state chain includes: generating identity information containing a unique identifier and performance parameters for each power agent, and obtaining its initial historical reputation value; writing the identity information and the historical reputation value into the identity chain; obtaining the power level, location, and operating status of each power agent to form dynamic state information; verifying the consistency of the dynamic state information using a consensus mechanism, and writing the verified dynamic state information into the state chain to form global state information; and encapsulating the identity chain and the state chain to construct a two-layer data ledger.

[0009] Optionally, generating a dynamic task contract containing task objectives and incentive weights includes: collecting and analyzing local energy demand data, identifying demand hotspot areas from the energy demand data; generating task objective codes based on the coverage and power requirements of the demand hotspot areas; calculating incentive weights based on the importance and urgency of the task objective codes, and associating the task objective codes with the incentive weights to form the dynamic task contract.

[0010] Optionally, generating the negotiation result including role allocation and deployment location includes: the power agent obtaining the dynamic task contract and the global state information; each power agent calculating its bidding score by combining its own dynamic state information and the incentive weight of the dynamic task contract; performing a group consensus algorithm on the bidding score among all power agents to determine a unique winning agent, and taking the role undertaken by the winning agent and the corresponding deployment location as the negotiation result.

[0011] Optionally, the construction of the dynamic energy potential field includes: calculating the initial potential energy value of each region in the preset map grid based on the spatial distribution of the energy demand information; calculating the virtual potential field gradient based on the initial potential energy value; and introducing a local potential energy increment in the corresponding region according to the deployment location in the negotiation result to adjust the initial potential energy value, thereby forming the dynamic energy potential field.

[0012] Optionally, the planned movement path includes: calculating the potential field gradient direction of the current position based on the dynamic energy potential field; generating an initial path based on the potential field gradient direction and the target deployment position; acquiring local obstacle information on the real-time movement path through preset sensors, and optimizing the initial path based on the local obstacle information to generate the final movement path.

[0013] Optionally, the dynamic path correction includes: the power agent monitoring its estimated trajectory and the estimated trajectories recorded by other power agents on the state chain to identify potential path conflicts; after identifying the potential path conflict, generating a path correction contract; locally negotiating the path correction contract with the power agents involved in the potential path conflict to generate a correction result, and adjusting its own movement path according to the correction result.

[0014] Optionally, the method further includes: after the power agent completes the task, calculating a task completion score based on the behavioral data recorded in the state chain; updating the historical reputation value of the power agent in the identity chain using the task completion score; and modifying the incentive weight based on the updated historical reputation value.

[0015] Optionally, the method further includes: monitoring the state chain to determine whether there is an abnormal offline agent that has not updated dynamic state information for a predetermined time; after determining that there is an abnormal offline agent, obtaining the unfinished dynamic task contract that it is currently executing; decomposing the unfinished dynamic task contract into a new dynamic task contract and triggering a new round of negotiation to achieve automatic task reallocation.

[0016] Based on the same inventive concept, the present invention also provides an emergency power intelligent scheduling and path planning system, comprising:

[0017] The power agent configuration module is used to configure multiple emergency power supplies as power agents to form a decentralized power agent network and to build a two-layer data ledger containing an identity chain and a state chain.

[0018] The two-tier data ledger management module is used to acquire energy demand information and generate dynamic task contracts containing task objectives and incentive weights based on the energy demand information.

[0019] The dynamic task contract generation module is used to obtain global state information from the state chain through the power agent, and negotiate the dynamic task contract in combination with the locally perceived energy demand information to generate a negotiation result that includes role allocation and deployment location.

[0020] The group negotiation decision-making module is used to construct a dynamic energy potential field based on the energy demand information and the deployment location in the negotiation results.

[0021] The dynamic potential field modeling module is used to plan the movement path of the power intelligent agent based on the dynamic energy potential field.

[0022] The path planning and correction module is used to update the state chain with the predicted trajectory and real-time position of the power agent as new dynamic state information during the movement process, and to perform dynamic path correction based on the information in the state chain.

[0023] Compared with the prior art, the present invention has the following advantages:

[0024] 1. By constructing a decentralized network based on a two-layer data ledger, an immutable identity and reputation profile is established for all participating power agents, and a consensus mechanism ensures the global consistency and credibility of state information. This design solves the problems of opaque information and lack of trust in traditional scheduling systems, providing a solid foundation of trust for subsequent autonomous negotiation and collaborative work, making the entire scheduling management process transparent, auditable, and highly reliable.

[0025] 2. A dynamic task contract with incentive weights and a bidding negotiation mechanism based on group consensus are introduced to transform emergency tasks into standardized, valuable digital contracts, driving power agents to autonomously bid based on their own capabilities and task value. This market-based resource allocation model replaces rigid command-based scheduling, enabling limited emergency power resources to be dynamically and adaptively configured according to the importance and urgency of the task, thus improving resource utilization efficiency and the economic benefits of emergency response.

[0026] 3. A method integrating high-level scheduling decision-making with low-level path planning is proposed. By constructing a dynamic energy potential field, complex task objectives are transformed into a clear navigation gravitational field, guiding the power agent to make macroscopic path selections. Simultaneously, by combining trajectory sharing and local negotiation on the state chain, autonomous obstacle avoidance and collaborative passage among multiple agents are achieved, ensuring high efficiency and safety at the logistics execution level. This makes resource deployment not only scientifically sound in decision-making but also intelligent and smooth in execution.

[0027] 4. Through continuous monitoring of the state chain, an automated fault detection and task redistribution mechanism was established. This mechanism can quickly identify and republish the unfinished tasks of a single power agent when it goes offline abnormally, triggering a new round of allocation, thereby achieving system self-healing and service continuity assurance. This design enhances the robustness of the entire emergency power management system and its ability to complete tasks in complex and harsh environments.

[0028] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 This is a flowchart illustrating an emergency power intelligent scheduling and path planning method according to an embodiment of the present invention.

[0031] Figure 2 This is a schematic diagram illustrating the dynamic task contract generation and incentive weight calculation of an emergency power intelligent scheduling and path planning method and system according to an embodiment of the present invention.

[0032] Figure 3 This is a schematic diagram of the dynamic energy potential field and movement path planning of an emergency power intelligent scheduling and path planning method and system according to an embodiment of the present invention.

[0033] Figure 4 This is a schematic diagram of the structure of an emergency power intelligent scheduling and path planning system according to an embodiment of the present invention. Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0035] Reference Figure 1One embodiment of the present invention proposes an intelligent scheduling and path planning method for emergency power supplies. It adopts a decentralized network by configuring multiple emergency power supplies as power intelligence agents, and performs scheduling and path planning through a two-layer data ledger, dynamic task contract negotiation and dynamic energy potential field. This method can realize efficient, reliable and adaptive scheduling management and optimized configuration of emergency power resources.

[0036] The method described in this embodiment specifically includes:

[0037] Multiple emergency power supplies are configured as power intelligent agents to form a decentralized power intelligent agent network, and a two-layer data ledger containing an identity chain and a state chain is constructed. The identity chain is used to record the identity information and historical reputation value of each power intelligent agent, and the state chain is used to record the dynamic state information of the power intelligent agent in real time.

[0038] Obtain energy demand information and generate dynamic task contracts containing task objectives and incentive weights based on the energy demand information;

[0039] The power agent obtains global state information from the state chain and negotiates the dynamic task contract by combining it with locally perceived energy demand information, generating a negotiation result that includes role allocation and deployment location.

[0040] A dynamic energy potential field is constructed based on the energy demand information and the deployment location in the negotiation results;

[0041] The power agent plans its movement path based on the dynamic energy potential field.

[0042] During the movement, the predicted trajectory and real-time position of the power agent are updated to the state chain as new dynamic state information, and dynamic path correction is performed based on the information in the state chain.

[0043] Specifically, a decentralized emergency power scheduling framework based on multi-agent technology is constructed. The core is to assign an agent identity to each emergency power source and provide a trust foundation and real-time global situational awareness for the network through a two-layer data ledger containing an identity chain and a state chain. Based on this, external energy demand is abstracted into dynamic task contracts with incentive weights, driving the agents of each power source to negotiate in a decentralized manner based on shared state information, determining task executors and deployment locations through market-based bidding. To solve the path planning problem, discrete scheduling decisions are transformed into continuous dynamic energy potential fields, using the potential field gradient to guide agents in macroscopic path planning. Simultaneously, by sharing predicted trajectories on the state chain, a multi-agent collaboration mechanism is established, enabling agents to proactively identify and negotiate to resolve potential path conflicts during movement. This integrates high-level task objectives, group collaboration, and underlying motion control, forming a complete closed-loop autonomous scheduling logic from task generation and decision allocation to path execution and dynamic correction. Through decentralized architecture design, dependence on a central scheduling server is eliminated, avoiding single-point-of-failure risks and improving the robustness and scalability of the entire emergency power system. The introduction of a dual-layer data ledger not only ensures the credibility of each agent's identity and the traceability of its behavior, but also constructs a unified, real-time information sharing platform through a state chain, laying a solid data foundation for autonomous collaboration among agents. Based on a decision-making mechanism of dynamic task contracts and group negotiation, intelligent and market-oriented allocation of emergency resources is achieved, enabling resources to adaptively flow to the highest-value areas, improving scheduling efficiency and accuracy. The combination of dynamic energy potential fields and real-time path correction ensures that the power agent's navigation in complex dynamic environments has both a clear global goal orientation and the micro-adaptability to cope with local obstacles and group conflicts, thereby comprehensively improving the speed, safety, and success rate of emergency response and task execution.

[0044] Optionally, constructing a two-layer data ledger comprising an identity chain and a state chain includes:

[0045] For each power agent, generate identity information containing a unique identifier and performance parameters, obtain its initial historical reputation value, and write the identity information and the historical reputation value into the identity chain;

[0046] Get the power level, location and operating status of each power agent to form dynamic state information;

[0047] A consensus mechanism is used to verify the consistency of the dynamic state information, and the verified dynamic state information is written into the state chain to form global state information;

[0048] The identity chain and the state chain are encapsulated to construct a two-layer data ledger.

[0049] Specifically, each emergency power supply is initialized with a digital identity as a power agent. This process generates a unique and immutable identifier for each power agent, which can be created based on a public key of an asymmetric encryption algorithm. Simultaneously, its physical performance is digitally described, forming a set of performance parameters including rated power, battery capacity, and maximum mobility. These two parts together constitute the core identity information of the power agent. Furthermore, an initial historical reputation value is assigned to each power agent. Subsequently, the identity information, including the unique identifier, performance parameters, and initial historical reputation value, is packaged and recorded in the identity chain after network consensus. The identity chain, acting as a distributed ledger, ensures the permanence and tamper-proof nature of the identity information. Secondly, during operation, each power agent periodically collects its own real-time data through its internal sensors and control units, such as the Battery Management System (BMS) and Global Positioning System (GPS), including remaining power, precise geographical coordinates, and operating mode such as standby, mobile, or power supply status. This data combination forms the dynamic status information of the agent. When a power agent needs to update its state, it broadcasts this dynamic state information to other power agents in the network. Upon receiving this information, other nodes in the network initiate a consensus mechanism. This consensus mechanism consists of rules that all nodes in the network jointly adhere to to ensure data consistency, such as the Practical Byzantine Fault Tolerance (PBFT) algorithm suitable for IoT environments. Through this mechanism, a certain number of nodes cross-validate the received dynamic state information to confirm its source legitimacy, timeliness, and logical rationality. Only when more than a preset threshold of agents reach a consensus and confirm the validity of the state information will it be approved for writing into a new block of the state chain. The preset threshold is two-thirds of the total number of agents. All verified dynamic state information on the state chain converges into a globally shared, real-time updated state information. A complete two-layer data ledger is constructed by logically encapsulating the identity chain and the state chain. Specifically, each record in the state chain contains a unique identifier of the power agent that generated the state information. Any query for dynamic status information can be traced back to the corresponding trusted static identity information in the identity chain through this identifier, thereby binding dynamic behavior with trusted entities and forming a decentralized data management foundation with a clear structure and reliable data.

[0050] For example, consider a network containing 10 emergency power supplies. Taking emergency power supply EPS001 as an example, its public key and private key are generated using an elliptic curve cryptography algorithm such as secp256k1, stored only within the power supply's intelligent body. This public key serves as the unique identifier for EPS001. Simultaneously, we need to define a set of physical performance parameters for EPS001, such as a rated power of 10kW, a battery capacity of 50kWh, a maximum moving speed of 5m / s, and an initial historical reputation value of 0.5 (range 0-1). This information is packaged and, through network consensus, using a Practical Byzantine Fault Tolerance (PBFT) algorithm, requires more than 2 / 3 of the nodes to reach agreement. Once 7 nodes in the network agree on the genesis block, it is recorded in the identity chain. Subsequently, EPS001 periodically collects its own data through the Battery Management System (BMS) and the Global Positioning System (GPS). For example, at T=5 minutes, the collected data includes remaining power of 45kWh, geographical coordinates, and the operating status of "Powering On". This data is combined into dynamic status information and broadcast to the network. Other nodes in the network, such as EPS002 and EPS003, will cross-validate this information upon receipt. If more than a preset threshold, such as 5 out of 7 nodes, confirms the validity of the state information, the information is written into a new block in the state chain. Each record in the state chain contains the unique public key of EPS001, thus being associated with the identity chain and ensuring data traceability and security.

[0051] Optionally, generating a dynamic task contract that includes task objectives and incentive weights includes:

[0052] Collect and analyze local energy demand data, and identify demand hotspots from the energy demand data;

[0053] Based on the coverage and power requirements of the aforementioned hotspot areas, generate the task target code;

[0054] Incentive weights are calculated based on the importance and urgency of the task target code, and the task target code is associated with the incentive weights to form the dynamic task contract.

[0055] Specifically, by accessing multiple information sources such as the city's emergency command center, power dispatch network, or IoT sensors, local energy demand data is continuously collected and analyzed. This data includes power outage reports, user assistance requests, and power consumption warnings for critical facilities. Through real-time analysis of this geospatially labeled data, such as using the density-based spatial clustering algorithm DBSCAN or defining geographic grids for demand density statistics, areas with highly concentrated energy demand—i.e., demand hotspots—are automatically identified. For each identified demand hotspot, a task target code is generated based on its specific energy consumption characteristics. The task target code is a structured data object that clearly defines the core requirements of the task, such as the central geographic coordinates of the area, the radius or polygon range to be covered, and the minimum power requirement to ensure the basic operation of the area. For example, for a hospital's demand hotspot, the task target code would include the hospital's precise location, the critical building range to be covered, and the power kilowatts required to maintain core medical equipment. To quantify the priority of the task, the system calculates a corresponding incentive weight. This incentive weight comprehensively considers the importance and urgency of the task. Specifically, its calculation follows the formula:

[0056] ;

[0057] in, This represents the total incentive weight of the dynamic task contract. This is a normalized value for task importance, determined based on the nature of the demand hotspot area. For example, critical infrastructure such as hospitals and communication base stations will be assigned a higher initial importance value, while ordinary residential areas will have a relatively lower importance value. Subsequently, a function is used to map it to a standard range such as 0 to 1. The normalized task urgency value is dynamically calculated based on the duration of the power outage, the estimated restoration time, or the urgency of the request submission time. For example, an area that has been without power for several hours will have a higher urgency value than an area that has just submitted a request; this value is also normalized. α and β are preset weighting coefficients, with α plus β equaling 1, used to adjust the relative proportion of importance and urgency in the final incentive weight calculation. This can be used to prioritize key objectives or respond to the most urgent needs based on macro-strategies. The generated task target code is associated with the calculated incentive weights, and the two are encapsulated into an immutable digital contract, i.e., a dynamic task contract. This contract is then published to the power agent network for all power agents to access and use as the basis for subsequent negotiation and decision-making. The dynamic task contract generation process is shown in Figure 2. By identifying geospatial demand hotspots, normalizing task importance and urgency, and weighting the incentive weights, a quantitative basis is provided for agent bidding.

[0058] For example, by analyzing data from the city's emergency command center, a hospital area was identified as having a high concentration of energy demand due to a power outage. The system generates a task objective code for this area, including the central geographic coordinates, a coverage radius of 200m, and a minimum power requirement of 5kW. To quantify the task's priority, corresponding incentive weights are calculated. The hospital's importance is initially set to 0.8, and the power outage duration is 2 hours. First, the importance needs to be normalized. Linear normalization is used to map [0,1] to [0,1]. Secondly, the urgency is normalized using a sigmoid function: Where k is the steepness factor (e.g., k=1), t is the power outage duration of 2 hours, and t_0 is a reference time of 1 hour. The calculation yields... If α = 0.6 and β = 0.4, then the total incentive weight... Finally, the task objective code and incentive weights are encapsulated into a dynamic task contract and published to the network for all power agents to access.

[0059] Optionally, the generation of the negotiation result, which includes role assignment and deployment location, includes:

[0060] The power agent acquires the dynamic task contract and the global state information;

[0061] Each power agent calculates its bidding score by combining its own dynamic state information and the incentive weight of the dynamic task contract.

[0062] A group consensus algorithm is executed among all power agents based on the bidding scores to determine a unique winning agent, and the role and corresponding deployment location of the winning agent are used as the negotiation result.

[0063] Specifically, when a new dynamic task contract is published to the decentralized power agent network, all power agents in a standby or allocable state will actively acquire this contract. Simultaneously, each power agent obtains global state information, including the latest battery level, location, and operational status of all other agents, by accessing the state chain. Secondly, after acquiring the task information, each power agent enters a localized decision-making and evaluation phase. It combines its own dynamic state information, such as its current battery level and location, with the incentive weights and task objectives contained in the dynamic task contract to calculate its bidding score for the task. This bidding score is a comprehensive evaluation metric, quantified by the following formula:

[0064] ;

[0065] In this formula, It is the final bid score of the i-th power agent. It is the incentive weight of the dynamic task contract, obtained directly from the contract, and represents the value of the task. The energy matching score is calculated based on the power available to the power agent, the power required for the task, and the expected service duration. The more available power, the higher the score. It is the distance cost score, which is calculated based on the distance between its current location and the deployment location of the mission target. The closer the distance, the higher the score. It is a reputation score, which is directly derived from the historical reputation value recorded in the identity chain. The higher the reputation, the higher the score. These are preset normalized weighting coefficients used to balance the importance of energy, distance, and reputation in the bidding process; their sum is 1. Through this formula, each power agent transforms its multi-dimensional capabilities and states into a single, comparable bidding score. To determine a unique task executor among all competitors, the network executes a consensus algorithm on the bidding scores submitted by all agents, with each agent broadcasting its calculated score to the entire network. After receiving the bidding scores from all other participants, all agents independently execute a deterministic selection algorithm locally, such as simply selecting the agent with the highest score as the winner. Since all agents have the same input data—all bidding scores—and follow the same selection rules, they will deterministically reach a unique consensus, jointly confirming the same winning agent. This winning agent will be assigned the role of task executor, and its deployment location will be the location specified in the dynamic task contract. These two pieces of information together constitute the final negotiation result.

[0066] For example, after the dynamic task contract is published, the power agent network needs to negotiate to determine the specific task executor. EPS001, EPS002, and EPS003 obtain the aforementioned dynamic task contract, and their respective status information is as follows: EPS001 has a power of 45kWh, position 1, and reputation value of 0.6; EPS002 has a power of 30kWh, position 2, and reputation value of 0.8; EPS003 has a power of 55kWh, position 3, and reputation value of 0.7. Each power agent needs to calculate its own bidding score. First, the energy matching score is calculated. If the task requires 5kW of power for 2 hours, then 10kWh of electricity is needed. Using a linear scaling factor: Since the score cannot exceed 1, we take 1. Since the score cannot exceed 1, we take 1. Since the score cannot exceed 1, we take 1. Next, we calculate the distance cost score. Using an inverse proportional function Where k is the distance sensitivity parameter, with a default value of k=1000. Calculate the distance: .get Preset Then the bid score is: Since EPS002 scored the highest, it was assigned the role of Task Executor, and its deployment location is [location to be filled in].

[0067] Optionally, constructing the dynamic energy potential field includes: calculating the initial potential energy value of each region in the preset map grid based on the spatial distribution of the energy demand information;

[0068] The virtual potential field gradient is calculated based on the initial potential energy value;

[0069] Based on the deployment location in the negotiation results, a local potential energy increment is introduced in the corresponding area to adjust the initial potential energy value, thereby forming the dynamic energy potential field.

[0070] Specifically, the entire geographical area of ​​the emergency response is pre-gridized on a digital map, discretizing the continuous two-dimensional space into independent grid cells, each with a unique coordinate identifier. Next, based on the spatial distribution characteristics of the acquired energy demand information, an initial potential energy value is calculated for each grid cell. Specifically, the total energy demand or density falling within each grid cell is statistically analyzed and mapped to the cell's initial potential energy value using a pre-defined function. In this model, potential energy is defined as a scalar field representing the attractiveness of demand; areas with higher demand have correspondingly higher initial potential energy values, thus forming an initial potential field reflecting the global energy demand distribution. Subsequently, based on this initial potential energy value distribution, a virtual potential field gradient covering the entire map can be calculated. This gradient points in the direction of the fastest potential energy growth at any point, i.e., the direction of the fastest demand growth. After completing the construction of the basic potential field, the target deployment location of the unique winning agent is extracted based on the negotiation results output by the group negotiation decision module. To guide the winning agent precisely to this specific target, the initial potential energy value of the grid cell corresponding to the deployment location and its neighboring region is adjusted by introducing a local potential energy increment. This increment is numerically greater than the peak potential energy of the surrounding region, and its function is to make the target deployment location a strong attracting pole in the entire potential field. The resulting dynamic energy potential field can be expressed as:

[0071] ;

[0072] In this formula, p represents any point on the map. U(p) is the final dynamic potential energy value at that point. It is the initial potential energy value at that location, determined by global energy demand information, and is a function of the demand density of the region where that point is located. This is a local potential energy increment function introduced at the negotiated deployment location. This function reaches its maximum value when p equals the negotiated deployment location and decays rapidly with increasing distance from the negotiated deployment location, forming a potential energy peak centered on the deployment location. Through such superposition and adjustment, a dynamic energy potential field is constructed that reflects both the global demand situation and highlights the current highest priority task objective.

[0073] For example, to guide the power agent to the target location effectively, a dynamic energy potential field needs to be constructed. The 500m x 500m area surrounding the hospital is divided into a 10m x 10m grid, and the energy demand density within a certain grid cell (i,j) is 0.2kW / m². The demand density is mapped to the initial potential energy value using a linear scaling factor:

[0074] Where k is a proportionality constant, such as 10, then The grid cell corresponding to the deployment location. Introducing local potential energy increments in the region and its vicinity using a Gaussian function: , where A is the peak amplitude (e.g., A=100). It is the standard deviation (e.g.) ). At the deployment location At a location 10m away from the deployment position ≈93.9. The final dynamic energy potential field is U(p) = .

[0075] Optionally, the planned movement path includes:

[0076] The potential field gradient direction at the current position is calculated based on the dynamic energy potential field.

[0077] An initial path is generated based on the potential field gradient direction and the target deployment location;

[0078] The system acquires local obstacle information on the real-time travel path using preset sensors, and optimizes the initial path based on this local obstacle information to generate the final travel path.

[0079] Specifically, the winning agent, based on the dynamic energy potential field constructed by the dynamic potential field modeling module, calculates the potential field gradient direction at its current position using numerical methods. This gradient is a vector pointing in the direction of the fastest increase in potential energy value within the dynamic energy potential field, physically equivalent to the direction of the maximum attractive force acting on the power agent. This attractive force encompasses both the pull of the global demand distribution and a strong direction towards the specific deployment location. Next, the power agent uses this potential field gradient direction as its basic navigation vector, combined with its final target deployment location, to generate an initial path through iterative calculation. This process can be viewed as a gradient ascent process; at each step, the power agent chooses to move a small distance along the current gradient direction, repeating this process continuously. All movement steps connected together constitute the idealized initial path from the starting point to the deployment location. This path macroscopically guarantees the strategy of moving towards the highest-value target. Next, as the power agent travels along this initial path, it uses its onboard sensors, such as LiDAR, high-definition cameras, or ultrasonic sensor arrays, to perform real-time, high-frequency scanning of the physical environment ahead and around the path. This allows it to acquire information on local obstacles such as sudden traffic congestion, temporary roadblocks, or rubble not marked on the map. Once a local obstacle is detected, the path planning algorithm immediately performs dynamic optimization of the initial path. This optimization process can introduce the concept of repulsive force; each obstacle generates a virtual repulsive force field around it, the magnitude of which is inversely proportional to the distance between the power agent and the obstacle. The power agent's next movement decision will be determined by the vector sum of the attractive and repulsive forces, and its motion vector can be represented as:

[0080] ;

[0081] in, It is the motion vector of the power supply agent in the next moment. It is an attraction vector calculated based on the gradient of the dynamic energy potential field, and its direction points in the direction of the fastest increase in potential energy, guiding the agent toward the target. It is the total repulsion vector generated by the detected local obstacles, and its direction is away from the obstacles, used to avoid collisions. and This is a preset gain coefficient used to adjust the relative weights of attractive and repulsive forces. By calculating this vector sum in real time, a smooth curve is generated that bypasses obstacles and reverts to the original target. This continuously dynamically optimized trajectory is the final movement path. The dynamic energy potential field and path planning effect are shown in Figure 3. The initial potential field reflects the global demand, the target position forms a potential energy peak, and the agent moves along the gradient direction and avoids obstacles, achieving precise deployment.

[0082] For example, the winning power agent needs to plan its movement path based on the dynamic energy potential field. EPS002 starts planning its path from its current position, first calculating the gradient direction of the potential field at that position. The gradient of the potential field in the x-direction is dU / dx, and the gradient in the y-direction is dU / dy. The gradient can be calculated numerically, such as using the finite difference method. The calculated gradient direction vector is (0.1, -0.2). EPS002 uses this gradient direction as the basic navigation vector to iteratively generate an initial path, moving a small distance, for example, 1m, along the gradient direction at each step until it approaches the deployment position. During this initial path movement, EPS002 scans the environment ahead using a lidar and detects an obstacle 5m away, which generates a repulsive force field. The magnitude of the repulsive force is inversely proportional to the distance: ,in It is the repulsive force coefficient. If the direction of the repulsive force vector is away from the obstacle, the magnitude is 10 / 5 = 2. Then the next motion vector EPS002 moves along this vector direction, avoiding the obstacle.

[0083] Optionally, the dynamic path correction includes:

[0084] The power agent monitors its predicted trajectory and the predicted trajectories recorded by other power agents on the state chain to identify potential path conflicts.

[0085] After identifying the potential path conflicts, a path correction contract is generated;

[0086] The power agent involved in the potential path conflict negotiates the path correction contract locally, generates a correction result, and adjusts its own movement path according to the correction result.

[0087] Specifically, based on the power agent currently performing a movement task, it continuously reads and monitors the estimated trajectory information of all other moving power agents from the state chain. This estimated trajectory information consists of a series of future time points and their corresponding predicted location coordinates uploaded to the state chain by each power agent after planning its movement path. The power agent performs spatiotemporal cross-analysis of its own estimated trajectory with all other estimated trajectories obtained from the state chain. When it predicts that at a future time point, its predicted position will be less than the distance to another power agent's predicted position, a potential path conflict is identified. Next, after identifying a potential path conflict, the power agent that initiated the monitoring immediately generates a path correction contract. This contract is a standardized data structure containing unique identifiers of both conflicting parties, the predicted time and geographical coordinates of the conflict, and the initiator's task priority. This contract is then sent to the other power agent involved in the potential path conflict. Subsequently, the two parties receiving the path correction contract enter a local, point-to-point negotiation phase. They do not need to broadcast across the entire network; communication only occurs between the conflicting parties. The negotiated decision is based on a set of pre-defined, deterministic rules to achieve rapid decision-making. For example, it compares the incentive weights of the dynamic task contracts being executed by both parties, with the party with the lower incentive weight assuming the avoidance responsibility; or, if the weights are equal, it compares the historical reputation values ​​of both parties, with the party with the lower reputation value avoiding the conflict. The result of the negotiation, i.e., which party adjusts its path, is the correction result. Finally, based on the correction result generated by the negotiation, the agent assuming the avoidance responsibility immediately adjusts its movement path, for example, by temporarily reducing speed, pausing and waiting, or slightly deviating from the original path to avoid the conflict point. The adjusted new movement path generates a new predicted trajectory, which is immediately updated in the state chain by the agent, thus completing the dynamic path correction.

[0088] For example, to avoid collisions between multiple power agents during movement, dynamic path correction is required. When EPS002 detects that the predicted trajectory of EPS004 may conflict with its own predicted trajectory within 2 minutes, and the distance is less than 5 meters, EPS002 generates a path correction contract, which includes the identities of both parties, the time and coordinates of the conflict, and its own task priority. After receiving the contract, EPS004 negotiates with EPS002. Assuming the negotiation rule is based on comparing task priorities, and EPS004's task priority is lower, EPS004 assumes the responsibility of avoidance. EPS004 temporarily reduces its speed, waits for 1 minute to avoid the conflict point, and then resumes its original speed, thereby avoiding the potential collision.

[0089] Optionally, the method further includes:

[0090] After the power agent completes the task, the task completion score is calculated based on the behavioral data recorded in the state chain.

[0091] The task completion score is used to update the historical reputation value of the power agent in the identity chain;

[0092] The incentive weights are modified based on the updated historical reputation values.

[0093] Specifically, after a power agent completes the work specified in a dynamic task contract, such as providing power for a specified duration at a target deployment location, a task evaluation procedure is initiated. This procedure accesses the state chain to retrieve and analyze all behavioral data recorded by the power agent during the task execution. This data constitutes a complete digital footprint of its behavior, including whether it arrived at the deployment location on time, whether the power output during the power supply process was stable and met requirements, and whether it effectively adhered to path coordination rules. Based on this objectively recorded behavioral data, a comprehensive task completion score is calculated using a multi-dimensional evaluation model. This score can be quantified in the following forms:

[0094] ;

[0095] in, This represents the final task completion score. f is a preset scoring function. It is a timeliness indicator for tasks, obtained by comparing the actual arrival time recorded in the state chain with the deployment time required by the contract. It is a quality indicator of the task, which is used to evaluate its stability and accuracy by comparing the actual power supply curve recorded in the state chain with the power value required by the contract. These are compliance indicators for the task, determined by examining its behavior records on the state chain to identify violations such as improper handling of path conflicts. After quantification and weighted aggregation, these indicators generate a standardized task completion score. Next, this calculated task completion score is used to update the historical reputation value recorded for the power agent in its identity chain. This update process is not a simple replacement but employs a smooth update mechanism to reflect its long-term performance; the update formula can be expressed as:

[0096] ;

[0097] in, This is the updated historical reputation value. It is the old historical reputation value stored in the identity chain. This is the score for completing this task. This is an update factor between 0 and 1, used to control the impact of a single task's performance on overall reputation. The updated historical reputation value is packaged and written into the identity chain, achieving a permanent update to the power agent's reputation profile. Based on the completion status of the current task and the updated historical reputation value, the future incentive weight generation strategy is dynamically modified. Specifically, the system analyzes the relationship between task completion scores and initial incentive weights. If tasks with high incentive weights are frequently failed by agents with low reputation values, their base incentive weights are further increased when generating dynamic task contracts for similar tasks in the future to attract agents with higher reputation and stronger capabilities to participate in the bidding. Conversely, if certain tasks are consistently completed successfully, their incentive weights are appropriately adjusted to optimize the allocation of incentive resources.

[0098] For example, after the power agent completes its task, it needs to be evaluated and its reputation value updated. When EPS002 completes its power supply task, the system evaluates its behavioral data. Assume EPS002 arrives on time, provides stable power, but experiences a minor path conflict. Then the on-time performance... Punctuality, quality Stable power and compliance Minor conflict. The scoring function is linearly weighted: ,in ,but Update the historical reputation value, assuming γ = 0.1. ,but The system will Write it into the identity chain to complete the reputation update.

[0099] Optionally, the method further includes:

[0100] By monitoring the state chain, it can be determined whether there are any abnormal offline intelligent agents that have not updated their dynamic state information for a predetermined period of time.

[0101] After determining the existence of the abnormal offline agent, obtain the unfinished dynamic task contract that it is currently executing;

[0102] The incomplete dynamic task contract is decomposed into a new dynamic task contract and a new round of negotiation is triggered to achieve automatic task redistribution.

[0103] Specifically, each power agent or dedicated monitoring node in the network continuously performs health checks by monitoring the state chain. It iterates through the latest dynamic state information of all power agents on the state chain, paying particular attention to the timestamps accompanying each entry. A reasonable time threshold, such as several minutes, is pre-set, representing the normal maximum interval for a power agent to update its state. If the latest timestamp of a power agent's record on the state chain exceeds this pre-set time threshold, the agent is marked as an abnormal offline agent. Secondly, once an abnormal offline agent is identified through the above monitoring, the subsequent fault-tolerant processing procedure is immediately initiated. This involves querying the unfinished dynamic task contracts undertaken by the abnormal offline agent before it went offline, which are recorded as being executed on the state chain. By accessing the state chain and possible contract repositories, key information such as the goal and remaining workload of the unfinished task is precisely obtained. For example, if a task is to supply power for two hours and the agent goes offline after one hour, then the unfinished dynamic task contract is the remaining one hour of power supply task. Next, this incomplete dynamic task contract is processed by decomposing it or directly transforming it into a new dynamic task contract. Its deployment location remains unchanged, but the incentive weight is dynamically adjusted upwards based on the increasing urgency. This newly generated dynamic task contract is immediately published to the power agent network, automatically triggering a new round of group negotiation and decision-making, achieving seamless and automatic task redistribution.

[0104] For example, to ensure system reliability, it is necessary to handle abnormal situations and automatically reassign tasks. Suppose EPS005 goes offline after 1 hour of power supply, and the latest timestamp of EPS005 on the state chain is more than 5 minutes from the current time. EPS005 is marked as an abnormal offline agent. A query of the unfinished task contracts undertaken by EPS005 reveals that there is 1 hour of remaining power supply time. A new contract requiring 1 hour of power supply is generated, with the deployment location remaining the same, but the incentive weight is increased by 10%. This new contract is published to the network, triggering a new round of negotiation and decision-making, thereby achieving automatic task reassignment and ensuring the continuity of energy supply.

[0105] Reference Figure 4 Based on the same inventive concept, the present invention also provides an emergency power intelligent scheduling and path planning method system, the system comprising:

[0106] The power agent configuration module is used to configure multiple emergency power supplies as power agents to form a decentralized power agent network and to build a two-layer data ledger containing an identity chain and a state chain.

[0107] The two-tier data ledger management module is used to acquire energy demand information and generate dynamic task contracts containing task objectives and incentive weights based on the energy demand information.

[0108] The dynamic task contract generation module is used to obtain global state information from the state chain through the power agent, and negotiate the dynamic task contract in combination with the locally perceived energy demand information to generate a negotiation result that includes role allocation and deployment location.

[0109] The group negotiation decision-making module is used to construct a dynamic energy potential field based on the energy demand information and the deployment location in the negotiation results.

[0110] The dynamic potential field modeling module is used to plan the movement path of the power intelligent agent based on the dynamic energy potential field.

[0111] The path planning and correction module is used to update the state chain with the predicted trajectory and real-time position of the power agent as new dynamic state information during the movement process, and to perform dynamic path correction based on the information in the state chain.

[0112] It should be noted that the electrical connections between the various units described above do not necessarily represent direct or indirect connections. Any indirect connection method can be applied to the embodiments of the present invention as long as it achieves the purpose of the present invention. The above descriptions are merely exemplary embodiments of the present invention and should not be construed as limiting the scope of the present invention.

[0113] All equivalent changes and modifications made in accordance with the teachings of this invention are still within the scope of this invention. Those skilled in the art will readily conceive of other embodiments of this invention upon considering the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this invention that follow the general principles of this invention and include common knowledge or conventional techniques in the art not described herein.

Claims

1. An intelligent scheduling and path planning method for emergency power supplies, characterized in that, The method includes: Multiple emergency power supplies are configured as power intelligent agents to form a decentralized power intelligent agent network, and a two-layer data ledger containing an identity chain and a state chain is constructed. The identity chain is used to record the identity information and historical reputation value of each power intelligent agent, and the state chain is used to record the dynamic state information of the power intelligent agent in real time. The process involves acquiring energy demand information and generating a dynamic task contract containing task target codes and incentive weights. This generation includes: collecting and analyzing local energy demand data to identify demand hotspot areas; generating task target codes based on the coverage and power requirements of the demand hotspot areas; and calculating incentive weights based on the importance and urgency of the task target codes. ； in, This represents the total incentive weight of the dynamic task contract. This is the normalized value for task importance. This represents the normalized value of task urgency, where α and β are preset weighting coefficients. The task target code is then associated with the incentive weight to form the dynamic task contract; The power agent obtains global state information from the state chain and negotiates the dynamic task contract by combining it with locally perceived energy demand information, generating a negotiation result that includes role allocation and deployment location. Generating the negotiation result includes: the power agent obtaining the dynamic task contract and the global state information; each power agent calculating its bidding score by combining its own dynamic state information and the incentive weight of the dynamic task contract. ； in, It is the final bid score of the i-th power agent. This is the incentive weight of the dynamic task contract. It is the energy matching score. It is the distance cost score. It's a reputation score. , , These are preset weighting coefficients; A group consensus algorithm is executed among all power agents based on the bidding scores to determine a unique winning agent, and the role and corresponding deployment location of the winning agent are used as the negotiation result. A dynamic energy potential field is constructed based on the energy demand information and the deployment location in the negotiation results. The construction of the dynamic energy potential field includes: calculating the initial potential energy value of each region in a preset map grid based on the spatial distribution of the energy demand information; calculating the virtual potential field gradient based on the initial potential energy value; and introducing a local potential energy increment in the corresponding region according to the deployment location in the negotiation results to adjust the initial potential energy value, thereby forming the dynamic energy potential field. ； Where p represents any point on the map, and U(p) is the final dynamic potential energy value of that point. It is the initial potential energy value at that location, determined by global energy demand information. It is the local potential energy increment function introduced at the deployment location determined through negotiation; The power agent plans its movement path based on the dynamic energy potential field. During the movement, the predicted trajectory and real-time position of the power agent are updated to the state chain as new dynamic state information, and dynamic path correction is performed based on the information in the state chain.

2. The emergency power intelligent scheduling and path planning method according to claim 1, characterized in that, The construction of a two-layer data ledger comprising an identity chain and a state chain includes: For each power agent, generate identity information containing a unique identifier and performance parameters, obtain its initial historical reputation value, and write the identity information and the historical reputation value into the identity chain; Get the power level, location and operating status of each power agent to form dynamic state information; A consensus mechanism is used to verify the consistency of the dynamic state information, and the verified dynamic state information is written into the state chain to form global state information; The identity chain and the state chain are encapsulated to construct a two-layer data ledger.

3. The emergency power intelligent scheduling and path planning method according to claim 1, characterized in that, The planned movement path includes: The potential field gradient direction at the current position is calculated based on the dynamic energy potential field. An initial path is generated based on the potential field gradient direction and the target deployment location; The system acquires local obstacle information on the real-time travel path using preset sensors, and optimizes the initial path based on this local obstacle information to generate the final travel path.

4. The emergency power intelligent scheduling and path planning method according to claim 1, characterized in that, The dynamic path correction includes: The power agent monitors its predicted trajectory and the predicted trajectories recorded by other power agents on the state chain to identify potential path conflicts. After identifying the potential path conflicts, a path correction contract is generated; The power agent involved in the potential path conflict negotiates the path correction contract locally, generates a correction result, and adjusts its own movement path according to the correction result.

5. The emergency power intelligent scheduling and path planning method according to claim 1, characterized in that, The method further includes: After the power agent completes the task, the task completion score is calculated based on the behavioral data recorded in the state chain. The task completion score is used to update the historical reputation value of the power agent in the identity chain; The incentive weights are modified based on the updated historical reputation values.

6. The emergency power intelligent scheduling and path planning method according to claim 1, characterized in that, The method further includes: By monitoring the state chain, it can be determined whether there are any abnormal offline intelligent agents that have not updated their dynamic state information for a predetermined period of time. After determining the existence of the abnormal offline agent, obtain the unfinished dynamic task contract that it is currently executing; The incomplete dynamic task contract is decomposed into a new dynamic task contract and a new round of negotiation is triggered to achieve automatic task redistribution.

7. An emergency power intelligent dispatching and path planning system, applied to the emergency power intelligent dispatching and path planning method as described in any one of claims 1-6, characterized in that, The system includes: The power agent configuration module is used to configure multiple emergency power supplies as power agents to form a decentralized power agent network and to build a two-layer data ledger containing an identity chain and a state chain. A two-layer data ledger management module is used to acquire energy demand information and generate dynamic task contracts containing task target codes and incentive weights based on the energy demand information. The generation of dynamic task contracts containing task target codes and incentive weights includes: collecting and analyzing local energy demand data, identifying demand hotspot areas from the energy demand data; generating task target codes based on the coverage and power requirements of the demand hotspot areas; calculating incentive weights based on the importance and urgency of the task target codes, and associating the task target codes with the incentive weights to form the dynamic task contract. A dynamic task contract generation module is used to obtain global state information from the state chain through the power agent and negotiate the dynamic task contract in combination with locally perceived energy demand information to generate a negotiation result including role allocation and deployment location. The generation of the negotiation result including role allocation and deployment location includes: the power agent obtaining the dynamic task contract and the global state information; each power agent calculating its bidding score by combining its own dynamic state information and the incentive weight of the dynamic task contract; and performing a group consensus algorithm on the bidding score among all power agents to determine a unique winning agent, and taking the role undertaken by the winning agent and the corresponding deployment location as the negotiation result. A group negotiation decision module is used to construct a dynamic energy potential field based on the energy demand information and the deployment location in the negotiation result. The construction of the dynamic energy potential field includes: calculating the initial potential energy value of each region in a preset map grid based on the spatial distribution of the energy demand information; calculating the virtual potential field gradient based on the initial potential energy value; and introducing a local potential energy increment in the corresponding region according to the deployment location in the negotiation result to adjust the initial potential energy value, thereby forming the dynamic energy potential field. The dynamic potential field modeling module is used to plan the movement path of the power intelligent agent based on the dynamic energy potential field. The path planning and correction module is used to update the state chain with the predicted trajectory and real-time position of the power agent as new dynamic state information during the movement process, and to perform dynamic path correction based on the information in the state chain.

Images