Distribution network distributed self-healing evaluation and decision-making method, system, equipment and medium

Abstract

The invention discloses a distributed self-healing evaluation and decision-making method, system and device for a power distribution network and a medium. The method comprises the following steps: acquiring local state information of an intelligent agent; based on the local state information of the intelligent agent, performing distributed state evaluation by adopting a consistency algorithm to obtain a global state evaluation result; performing fault diagnosis and system isolation based on a global state evaluation result to obtain an optimal isolation scheme; based on the optimal isolation scheme, regional coordination and global optimization are carried out through an intelligent agent communication and negotiation mechanism and a distributed collaborative decision-making mechanism, and an optimal power supply reconstruction scheme is obtained; according to the method, the high efficiency, the reliability and the adaptability of the self-healing process of the power distribution network are ensured.

Description

Technical Field

[0001] This invention relates to the field of distribution network self-healing technology, and in particular to a method, system, device and medium for evaluating and making decisions on distributed self-healing in distribution networks. Background Technology

[0002] As power systems evolve towards intelligence and distributed systems, distribution network self-healing technology has become a key technology for ensuring power supply reliability. Traditional distribution network self-healing systems mostly adopt a centralized architecture, with a central control center collecting information, analyzing faults, and formulating recovery strategies. However, this centralized architecture faces several challenges: First, as the scale of the distribution network expands, the central processing unit faces excessive computational burden; second, centralized decision-making relies on the reliability of the communication network, and communication interruptions will affect the entire system's functionality; third, it is difficult to adapt to dynamic topology changes in the distribution network and the stochastic characteristics of distributed energy resources; and fourth, there is a lack of an effective self-healing process evaluation mechanism, making it impossible to guarantee the optimality and reliability of the self-healing results. Summary of the Invention

[0003] In view of the aforementioned existing problems, the present invention is proposed.

[0004] Therefore, this invention provides a method, system, device, and medium for distributed self-healing evaluation and decision-making in distribution networks to address the problems of high computational burden, slow response speed, poor adaptability, and lack of effective evaluation mechanisms in existing distribution network self-healing systems.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0006] In a first aspect, the present invention provides a method for evaluating and making decisions on distributed self-healing in power distribution networks, comprising:

[0007] Obtain local state information of the agent;

[0008] Based on the local state information of the agent, a consensus algorithm is used to perform distributed state evaluation to obtain the global state evaluation result;

[0009] Based on the global state assessment results, fault diagnosis and system isolation are performed to obtain the optimal isolation scheme;

[0010] Based on the optimal isolation scheme, the optimal power supply reconfiguration scheme is obtained by regional coordination and global optimization through intelligent agent communication and negotiation mechanism and distributed collaborative decision-making mechanism.

[0011] Execute the optimal power reconfiguration scheme, obtain and evaluate the self-healing results.

[0012] As a preferred embodiment of the distributed self-healing evaluation and decision-making method for distribution networks described in this invention, the step of obtaining the global state evaluation result includes:

[0013] Obtain local state information of each agent;

[0014] Based on the acquired data, each agent performs a preliminary estimate of the initial state based on its own data.

[0015] Set the neighbor set, weighting coefficients, convergence threshold, and maximum number of iterations;

[0016] The state estimates of each agent are updated synchronously using an improved consensus algorithm;

[0017] The weighting coefficients are dynamically adjusted based on the state differences between the agents.

[0018] Determine if the convergence condition has been met; if not, return to continue iterating.

[0019] If convergence is achieved, the global state evaluation result is output.

[0020] The beneficial effects of this preferred technical solution are that it effectively integrates the local information of each agent by using an improved consensus algorithm, dynamically adjusts the weighting coefficients to adapt to state differences, and achieves fast and robust global state assessment, thereby improving the accuracy and efficiency of power distribution network fault diagnosis and self-healing decision-making.

[0021] As a preferred embodiment of the distributed self-healing evaluation and decision-making method for distribution networks described in this invention, the step of obtaining the optimal isolation scheme includes:

[0022] Based on the state assessment results, a multi-feature fusion model is used to perform fault detection on multiple device-level agents to obtain fault detection results.

[0023] Integrate fault detection results from multiple device-level intelligent agents to locate the faulty section;

[0024] Based on the fault location results, the minimum cut set algorithm is applied to obtain the optimal isolation scheme.

[0025] As a preferred embodiment of the distributed self-healing evaluation and decision-making method for distribution networks described in this invention, the agent communication and negotiation mechanism includes:

[0026] Determine the communication mode;

[0027] Define a standard message format;

[0028] Design a reliable transmission protocol, calculate the probability of successful transmission, and optimize the retransmission strategy;

[0029] Introduce an event-triggered communication mechanism to optimize resource allocation;

[0030] By introducing a reputation mechanism and setting a priority arbitration mechanism, a negotiation process based on the contract network is constructed to facilitate information exchange and task negotiation.

[0031] As a preferred embodiment of the distributed self-healing evaluation and decision-making method for distribution networks described in this invention, the distributed collaborative decision-making mechanism includes:

[0032] The self-healing decision-making problem is modeled as a multi-agent cooperative game.

[0033] Design the utility function for the intelligent agent;

[0034] Each agent generates a set of candidate policies based on local information;

[0035] All agents perform a global evaluation of the policy combination;

[0036] The agent selects the optimal combination of strategies through a voting mechanism;

[0037] Introducing a heuristic search strategy to accelerate convergence;

[0038] The large problem is decomposed into multiple sub-regions, and the coordination between the sub-regions is carried out through the Lagrange relaxation algorithm.

[0039] By alternating the update strategy and multiplier, the optimal power reconfiguration scheme is finally obtained.

[0040] The beneficial effects of this preferred technical solution are that by modeling the self-healing decision problem as a multi-agent cooperative game and employing strategies such as agent utility function design, global evaluation, voting mechanism, heuristic search and Lagrange relaxation method, the optimal power supply reconfiguration scheme can be explored and determined efficiently, significantly improving the recovery speed and global optimization performance of the distribution network after a fault.

[0041] As a preferred embodiment of the distributed self-healing evaluation and decision-making method for distribution networks described in this invention, the step of acquiring and evaluating the self-healing results includes:

[0042] Define process evaluation dimensions, including response speed, collaboration efficiency, and resource utilization;

[0043] Calculate response speed indicators, collaborative efficiency indicators, and resource utilization rate indicators;

[0044] The process performance score is obtained by weighted combination;

[0045] Define the dimensions for evaluating the results, including load recovery rate, network reliability, and economy;

[0046] Calculate load recovery rate, network reliability, and economic indicators;

[0047] The performance score of the result is obtained by weighted combination;

[0048] The evaluation results are updated using a distributed consensus algorithm.

[0049] As a preferred embodiment of the distributed self-healing evaluation and decision-making method for distribution networks described in this invention, the improved consensus algorithm is expressed as follows:

[0050]

[0051] Where, x i (k) represents the state estimate of agent i in the kth iteration; N i Denotes the set of neighbors of agent i; a ij The weighting coefficients are represented by α, the acceleration factor is represented by w. j This represents the confidence weight of agent j.

[0052] Secondly, the present invention provides a distributed self-healing evaluation and decision-making system for power distribution networks, comprising:

[0053] The acquisition module is used to acquire local state information of the agent;

[0054] The distributed state evaluation module is used to perform distributed state evaluation based on the agent's local state information and a consensus algorithm to obtain the global state evaluation result.

[0055] The isolation scheme generation module is used to perform fault diagnosis and system isolation based on the global state assessment results, and obtain the optimal isolation scheme.

[0056] The distributed collaborative decision-making and optimization module is used to obtain the optimal power supply reconfiguration scheme by performing regional coordination and global optimization based on the optimal isolation scheme through intelligent agent communication and negotiation mechanism and distributed collaborative decision-making mechanism.

[0057] The self-healing execution and evaluation module is used to execute the optimal power reconfiguration scheme and obtain and evaluate the self-healing results.

[0058] Thirdly, the present invention provides an electronic device, comprising:

[0059] Memory, used to store programs;

[0060] A processor is configured to execute the computer-executable instructions, which, when executed by the processor, implement the steps of the distributed self-healing evaluation and decision-making method for the power distribution network.

[0061] Fourthly, the present invention provides a computer-readable storage medium, comprising: when the program is executed by a processor, the steps of implementing the distributed self-healing evaluation and decision-making method for power distribution networks.

[0062] The beneficial effects of this invention are as follows: This invention designs a three-layer architecture multi-agent distributed system, realizing hierarchical distributed control for self-healing decision-making; it solves the problem of global state perception under local information through a distributed state evaluation method based on consensus algorithms; it realizes global optimization decision-making in a distributed environment through a multi-agent collaborative decision-making mechanism based on game theory; and it supports real-time monitoring and evaluation of the self-healing process by constructing a multi-dimensional self-healing evaluation system, providing new ideas and methods for the development of distribution network self-healing technology, which is of great significance for improving the reliability and intelligence level of distribution network operation. Attached Figure Description

[0063] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0064] Figure 1 This is a basic flowchart illustrating a distributed self-healing evaluation and decision-making method for power distribution networks, provided in one embodiment of the present invention.

[0065] Figure 2 This is a complete flowchart illustrating a distributed self-healing evaluation and decision-making method for power distribution networks, provided as an embodiment of the present invention. Detailed Implementation

[0066] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0067] Example 1, referring to Figure 1 As an embodiment of the present invention, a method for evaluating and making decisions on distributed self-healing in power distribution networks is provided, comprising:

[0068] S100: Acquire local state information of the agent;

[0069] S200: Based on the local state information of the agent, a consensus algorithm is used to perform distributed state evaluation to obtain the global state evaluation result;

[0070] S300: Based on the global state assessment results, perform fault diagnosis and system isolation to obtain the optimal isolation scheme;

[0071] S400: Based on the optimal isolation scheme, it achieves regional coordination and global optimization through intelligent agent communication and negotiation mechanisms and distributed collaborative decision-making mechanisms to obtain the optimal power reconfiguration scheme;

[0072] S500: Executes the optimal power reconfiguration scheme, obtains and evaluates the self-healing results.

[0073] It should be noted that distribution network self-healing systems face numerous challenges during operation, including rapid and accurate fault detection and location, effective isolation of fault areas to prevent escalation of impact, optimized power reconfiguration to restore power supply to affected areas, and ensuring communication efficiency and decision-making speed throughout the process. With the expansion of power grid scale and the increase in complex fault scenarios, the limitations of traditional centralized control methods in terms of computational load, response speed, and adaptability become increasingly apparent. Distributed architectures, by constructing multi-level (device-level, feeder-level, and region-level) intelligent agent networks, effectively combine local information processing with global collaborative optimization. This not only distributes the computational load and improves response speed but also enhances the system's scalability and robustness, which is crucial for achieving efficient and reliable distribution network self-healing.

[0074] Therefore, to address the problems of high computational burden, slow response speed, poor adaptability, and lack of effective evaluation mechanisms in existing distribution network self-healing systems, a distributed self-healing evaluation and decision-making system for distribution networks is constructed. By rationally dividing the responsibilities of agents and designing an efficient communication and negotiation mechanism, global state assessment and collaborative decision-making based on local information are achieved. The system adopts a hierarchical distributed architecture, including device-level, feeder-level, and region-level agents. Distributed state assessment is achieved through a consensus algorithm, and collaborative decision-making is achieved through a game theory-based negotiation mechanism. Finally, a multi-dimensional self-healing evaluation system is constructed to ensure the efficiency, reliability, and adaptability of the distribution network self-healing process.

[0075] Example 2, refer to Figure 2 As one embodiment of the present invention, based on the previous embodiment, a distributed self-healing evaluation and decision-making method for distribution networks is provided, including:

[0076] In this embodiment, the distributed self-healing system architecture of the distribution network adopts a three-layer structure, including device-level agents (ELA), feeder-level agents (FLA), and region-level agents (RLA). The basic model of the agents is represented as follows:

[0077] AG i =S i ,P i D i C i E i

[0078] In the formula, AG iS represents the i-th agent; i P represents a state set, containing the current operating state of the agent. i D represents the perception set, describing the information acquisition capability of an intelligent agent; i C represents the decision set, containing the decision actions that the agent can perform; i E represents a communication set, defining the communication capabilities and protocols of the agents; i This represents the evaluation set, which includes evaluation metrics and methods for the agents.

[0079] The device-level intelligent agent is responsible for device status monitoring and basic control, and its state set is represented as follows:

[0080]

[0081] In the formula, V i Indicates voltage; I i T represents current; i Indicates equipment temperature; SW i Indicates the switch status; ST i Indicates the operating status of the equipment; AL i This indicates an alarm message.

[0082] The feeder-level agent is responsible for feeder state evaluation and local decision-making. Its state set is as follows:

[0083]

[0084] In the formula, TS i Represents the topology state; LS i Indicates load status; PS i Indicates the power flow state; FS i Indicates fault status; RS i Indicates the refactoring state.

[0085] The regional-level agent is responsible for regional coordination and global optimization, and its state set is as follows:

[0086]

[0087] In the formula, NS i Indicates network status; RS i Indicates resource status; CS i Indicates a coordinated state; PS i Indicates the planning state; ES i Indicates the evaluation status.

[0088] A hierarchical communication network is formed among the various intelligent agents, employing a publish / subscribe-based communication mechanism to ensure the real-time and reliable exchange of information. The communication process is represented as follows:

[0089] Mij =ID s ID r ,T,C,D,P

[0090] In the formula, M ij This represents a message from agent i to agent j; ID s Indicates the sender identifier; ID r T represents the receiver identifier; C represents the timestamp; D represents the message category; and P represents the message priority.

[0091] In this embodiment of the application, the local state information obtained by each agent through the sensor network in step S100 is represented as follows:

[0092] x i (0)=f i (m i )

[0093] In the formula, x i (0) represents the initial state estimate of agent i; f i (·) denotes the local state estimation function; m i Represents the intelligent agent m i The measurement dataset.

[0094] In this embodiment of the application, step S200 uses a consensus algorithm for distributed state evaluation, including the following steps:

[0095] The agent updates its state estimate using an iterative consensus algorithm:

[0096]

[0097] In the formula, x i (k) represents the state estimate of agent i in the kth iteration; N i Denotes the set of neighbors of agent i; a ij Denotes the weighting coefficients, satisfying

[0098] To improve the convergence speed and robustness of the consensus algorithm, an acceleration factor and confidence weight are introduced:

[0099]

[0100] In the formula, α represents the acceleration factor, used to adjust the convergence speed; w j The confidence weight represents the confidence level of agent j, reflecting the reliability of its estimate.

[0101] To address the dynamic nature of network topology changes, an adaptive weighting coefficient is designed:

[0102]

[0103] In the formula, β ij γ represents the similarity between agents i and j; γ is an adjustment parameter that controls the sensitivity of the similarity.

[0104] The convergence condition for distributed state evaluation is:

[0105] max i,j |x i (k)-x j (k)|<ε

[0106] In the formula, ε is the preset convergence threshold.

[0107] In this embodiment of the application, step S300 involves fault diagnosis and system isolation, including the following steps:

[0108] For fault condition assessment, a fault diagnosis model based on multi-feature fusion is designed:

[0109]

[0110] In the formula, F i g(·) represents the fault assessment result; g(·) represents the fault diagnosis function. Represents the j-th type of feature; θ i Indicates diagnostic parameters.

[0111] Multi-feature fusion employs an improved Dempster-Shafer evidence theory, introducing credibility weights:

[0112]

[0113] In the formula, m(Θ) represents the base probability assignment after fusion; m j (Θ) represents the basic probability assignment for the j-th source of evidence; w j d represents the weight of the j-th evidence source; j λ represents the degree of conflict of evidence sources; λ is the adjustment parameter.

[0114] The evidence fusion rule is defined as follows:

[0115]

[0116] In the formula, m 1,2 (A) represents the degree of confidence in event A after the fusion of two sources of evidence; K represents the conflict factor.

[0117] To mitigate the impact of conflicting evidence, a correction term is introduced:

[0118]

[0119] In the formula, δ is the conflict allocation coefficient.

[0120] In this embodiment of the application, the multi-feature fusion in step S300 includes calculating the basic probability assignment (BPA) and its confidence weight for each feature, correcting the basic probability assignment of each feature based on the confidence weight, and fusing all the corrected basic probability assignments using the improved Dempster-Shafer (DS) evidence theory combination rule to obtain a globally consistent fault assessment result.

[0121] In an optional implementation, step S300, multi-feature fusion, includes defining the local state estimates provided by each agent and their corresponding prior probabilities, calculating the posterior probabilities through a Bayesian network model, and fusing all local state estimates using conditional dependencies to obtain a globally consistent fault assessment result.

[0122] In another optional implementation, step S300 involves multi-feature fusion to define the local state estimates provided by each agent and their corresponding membership functions, calculate the fuzzy value of each feature based on these membership functions, and fuse all local state estimates through fuzzy inference rules to obtain a globally consistent fault assessment result.

[0123] It should be noted that this invention employs an improved Dempster-Shafer (DS) evidence theory for multi-feature fusion and fault assessment. Compared to Bayesian networks and fuzzy logic, its main advantage lies in its ability to more effectively handle uncertainty and incomplete information. DS evidence theory adjusts the trust levels of different data sources by introducing credibility weights, thus providing more robust fusion results in conflicting or uncertain data. While Bayesian networks excel at handling probabilistic dependencies, they may require complex model structure adjustments and substantial prior knowledge when dealing with highly conflicting or uncertain data, leading to increased computational complexity and less stable results. Furthermore, although fuzzy logic can naturally handle fuzziness and subjective judgment, its effectiveness is highly dependent on the design of membership functions and rule bases, and improper design can easily introduce biases. In contrast, the improved DS evidence theory not only retains its ability to handle uncertainty but also enhances the distinguishability of information from different sources by introducing credibility weights, enabling the system to achieve efficient and accurate fault diagnosis and assessment even in complex environments.

[0124] In this embodiment of the application, the agent communication and negotiation mechanism in step S400 includes the following steps:

[0125] The communication protocol adopts a publish / subscribe model and supports both point-to-point communication and broadcast communication.

[0126] The message format is defined as follows:

[0127] MSG = Header, Payload, Trailer

[0128] Header = ID src ID dst ,TS,Type,Len,Pri

[0129] In the formula, ID src Indicates the source address; ID dst The target address is represented by TS; the timestamp is represented by Type; the message type is represented by Len; and the priority is represented by Pri.

[0130] To improve communication reliability, a reliable transmission protocol based on the ACK / NACK mechanism was designed:

[0131] P succ =1-[1-(1-P e ) L ] N

[0132] In the formula, P succ P represents the probability of successful transmission; e L represents the channel error rate; L represents the message length; and N represents the maximum number of retransmissions.

[0133] To reduce communication overhead, an event-triggered communication mechanism is introduced:

[0134]

[0135] In the formula, x i (t) represents the state of agent i at time t; x i (t k ) indicates the state at the time of the last communication; δ i This is the trigger threshold.

[0136] The intelligent agent negotiation adopts a contract-based negotiation mechanism, comprising three phases: task publication, bidding, and awarding. In the task publication phase, the initiator broadcasts task information to potential participants:

[0137] Task=ID,Type,Content,Deadline,Utility

[0138] During the bidding phase, participants evaluate the task and submit bids:

[0139] Bid = ID task ID bidder Price, Capability, Commitment

[0140] During the awarding phase, the initiator evaluates the bids and selects the best bid:

[0141] b * =argmax b∈B U(b)

[0142] U(b)=α1u price (b)+α2u cap (b)+α3u com (b)

[0143] In the formula, U(b) represents the utility of bid b; u price (b) indicates price utility; u cap (b) indicates capability utility; u com (b) represents the commitment utility; α1, α2, and α3 are weighting coefficients.

[0144] To improve negotiation efficiency, a reputation mechanism is introduced:

[0145] Rep i (t+1)=βRep i (t)+(1-β)Perf i (t)

[0146] In the formula, Rep i denoted by , where represents the reputation value of agent i at time t; Perf represents the performance evaluation; and β is the smoothing coefficient.

[0147] Conflict resolution during the negotiation process will be conducted using a priority-based arbitration mechanism.

[0148]

[0149] In the formula, p i Indicates the priority of agent i; Indicates the basic priority; Indicates task-related priority; Indicates time-related priority; γ1, γ2, and γ3 are weighting coefficients.

[0150] In this embodiment, the distributed collaborative decision-making mechanism in step S400 is the core of realizing the self-healing of the distribution network. This invention designs a multi-agent collaborative decision-making method based on game theory. This method models the self-healing decision problem as a multi-agent cooperative game, and achieves the globally optimal decision by designing appropriate utility functions and negotiation mechanisms.

[0151] The decision game model is defined as:

[0152] G = N,S i U i

[0153] In the formula, G represents the game model; N represents the set of agents participating in the game; S iU represents the policy set of agent i; i Let i represent the utility function of agent i.

[0154] The design of the agent utility function comprehensively considers power restoration benefits, operating costs, and network constraints.

[0155] U i (s i ,s -i )=ω1R i (s i ,s -i )-ω2C i (s i )-ω3P i (s i ,s -i )

[0156] In the formula, U i s represents the utility of agent i; i s represents the policy of agent i; -i R represents the policy combination of other agents; i Represents the recovery benefit function; C i P represents the operating cost function; i This represents the constraint penalty function; ω1, ω2, and ω3 are weight coefficients.

[0157] The recovery benefit function takes into account both load importance and recovery time.

[0158]

[0159] In the formula, L i w represents the set of loads that agent i is responsible for; j p represents the importance weight of load j; j t represents the power of load j; j λ represents the recovery time of load j; λ is the time decay coefficient.

[0160] The operating cost function includes switching operation costs and network losses:

[0161]

[0162] In the formula, SW i c represents the set of switches controlled by agent i; k Indicates the operating cost of switch k; o k Indicates the number of operations performed on switch k; B i R represents the set of lines managed by agent i; l I represents the resistance of line l; l η represents the current in line l; η is the loss cost coefficient.

[0163] The constraint penalty function is used to ensure that network constraints are satisfied:

[0164]

[0165] In the formula, V represents the maximum allowable current of line l; ref Indicates the reference voltage; ΔV max μ1 and μ2 represent the maximum permissible voltage deviation; μ1 and μ2 are penalty coefficients.

[0166] Multi-agent collaborative decision-making employs a negotiation-based game theory approach, comprising three stages: proposal generation, proposal evaluation, and consensus decision-making. In the proposal generation stage, each agent generates candidate strategies based on local information.

[0167]

[0168] In the formula, This represents the set of candidate policies generated by agent i; Let K represent the k-th candidate strategy; K is the number of candidate strategies.

[0169] During the proposal evaluation phase, the agent evaluates the combination of strategies:

[0170]

[0171] In the formula, E(s) represents the global evaluation value of the strategy combination s; U i (s) represents the utility of agent i under policy combination s.

[0172] In the consensus decision-making phase, the agents select the optimal strategy combination through a voting mechanism:

[0173]

[0174] In the formula, s represents the final chosen strategy combination; S c Represents the set of candidate strategy combinations; v i This represents the vote value of agent i for policy combination s.

[0175] To improve decision-making efficiency, a heuristic search strategy is introduced:

[0176] s t+1 =s t +Δs t

[0177]

[0178] In the formula, s t Δs represents the strategy combination in the t-th iteration. t Indicates the amount of strategy adjustment; Indicates the evaluation gradient; r t Represents the random disturbance term; α t β t γ t These are control parameters.

[0179] To handle large-scale decision-making problems, a hierarchical decomposition strategy is adopted:

[0180] s = s1, s2, ..., s M

[0181] In the formula, s m Let represent the decision variable for the m-th sub-region. Coordination between sub-regions is achieved using the Lagrange relaxation method:

[0182]

[0183] In the formula, L(s,λ) represents the Lagrangian function; E m (s m ) represents the evaluation function for subregion m; g j (s) represents the j-th coupling constraint; λ j This represents the corresponding Lagrange multiplier.

[0184] The final decision-making process is achieved by alternating updates to the strategy and multipliers:

[0185]

[0186] In the formula, ρ k This is the step size parameter.

[0187] In this embodiment of the application, the evaluation of the self-healing result in step S500 includes two aspects: process evaluation and result evaluation, as detailed below:

[0188] The evaluation indicators for the self-healing process include response speed, collaborative efficiency, and resource utilization:

[0189]

[0190] In the formula, E proc Indicates process evaluation indicators; F resp E represents the response speed indicator. coor E represents the collaborative efficiency index; res Indicators representing resource utilization efficiency; These are the weighting coefficients.

[0191] The response speed index is defined as:

[0192] E resp =exp(-τ) r / τ0)

[0193] In the formula, τ r τ represents the actual response time; τ0 is the reference time constant.

[0194] Collaboration efficiency metrics take into account communication overhead and negotiation rounds:

[0195]

[0196] In the formula, C represents the amount of communication data; C max The maximum allowed communication volume; N represents the negotiation round; N max The maximum allowed number of rounds; α c β c These are the weighting coefficients.

[0197] Resource utilization rate indicators reflect the rationality of resource allocation:

[0198]

[0199] In the formula, r i This represents the actual usage of the i-th type of resource; w represents the maximum available quantity of the i-th type of resource; i This represents the weight of resource importance.

[0200] The evaluation metrics for self-healing results include load recovery rate, network reliability, and economy:

[0201]

[0202] In the formula, E resu E represents the evaluation index of the results. load E represents the load recovery rate index; reli E represents a network reliability metric. econ Indicates economic indicators; These are the weighting coefficients.

[0203] The load recovery rate index is defined as:

[0204]

[0205] In the formula, L represents the load set; This represents the power after load i is restored; Indicates the power of load i before the fault; w i This represents the weighting of load importance.

[0206] Network reliability metrics comprehensively consider both structural reliability and operational reliability:

[0207] E reli =β1E stru +β2E oper

[0208] In the formula, E reli E represents the structural reliability index. oper This represents the operational reliability index; β1 and β2 are weighting coefficients.

[0209] Structural reliability metrics are based on network topology analysis:

[0210]

[0211] In the formula, a ij This represents the connectivity between nodes i and j; n represents the number of nodes.

[0212] Operational reliability metrics take into account equipment load rate and redundancy:

[0213]

[0214] In the formula, m represents the number of devices; L i This indicates the load on device i; This indicates the maximum load capacity of device i.

[0215] Economic indicators take into account both operating costs and loss costs:

[0216]

[0217] In the formula, C op Indicates operating costs; C loss Indicates loss cost; C ref For reference cost.

[0218] To achieve distributed evaluation, a distributed evaluation algorithm based on a consensus mechanism was designed:

[0219]

[0220] In the formula, This represents the evaluation result of agent i in the k-th iteration; a ij This represents the level of trust between agents i and j.

[0221] In the embodiments of this application, such as Figure 2 As shown, the distributed self-healing process of the distribution network includes five stages: fault detection, fault location, system isolation, power supply reconfiguration, and service restoration. The self-healing process is completed collaboratively by a multi-agent system, with each agent participating in the decision-making and execution of its respective stage according to its responsibilities.

[0222] During the fault detection phase, the device-level intelligent agent collects fault information and makes a preliminary judgment:

[0223] FD i =Loc i Type i ,TSi ,Sev i

[0224] In the formula, FD i This represents the fault detection result of agent i; Loc i Indicates the location of the fault; Type i Indicates the fault type; TS i Indicates the time of failure; Sev i Indicates the severity of the fault.

[0225] During the fault location phase, the feeder-level agent coordinates the detection results of multiple device-level agents to accurately locate the faulty section:

[0226] FL = Sec, Type, Time, Conf

[0227] In the formula, FL represents the fault location result; Sec represents the fault section; Type represents the fault type; Time represents the fault occurrence time; and Conf represents the location confidence level.

[0228] During the system isolation phase, the agents collaboratively determine the isolation scheme:

[0229] IS=SW open ,SW veri Area

[0230] In the formula, IS represents the isolation scheme; SW open Indicates the set of switches that need to be disconnected; SW veri This represents the set of switches that need to be verified; Area represents the isolation region.

[0231] Intelligent agents use a graph-theoretic minimum cut set algorithm to determine the optimal isolation scheme:

[0232]

[0233] stG(V,E\SW open (Does not include faulty sections)

[0234] In the formula, Ω represents the set of all switches that satisfy the isolation condition; G(V,E) represents the network topology graph; V represents the set of nodes; and E represents the set of edges.

[0235] During the power supply reconfiguration phase, regional-level agents and feeder-level agents collaborate to determine the reconfiguration plan. The reconfiguration problem is modeled as a multi-objective optimization problem:

[0236] min F(x)=f1(x),f2(x),…,f m (x)

[0237] stg i(x)≤0,i=1,2,…,p

[0238] h j (x)=0,j=1,2,…,q

[0239] In the formula, x represents the reconstruction scheme; f i (x) represents the i-th objective function; g i (x) denotes an inequality constraint; h j (x) represents the equality constraint.

[0240] The reconfiguration objectives include maximizing load recovery rate, minimizing network loss, and minimizing the number of switching operations.

[0241]

[0242] In the formula, L rest L represents the set of loads whose power has been restored; B represents the set of lines; SW represents the set of switches; x k and These represent the target state and initial state of switch k, respectively.

[0243] Constraints include radial operating constraints, voltage constraints, and current constraints.

[0244] g1(x): The network topology is radial.

[0245]

[0246] The multi-objective optimization problem is solved using an improved NSGA-II algorithm, combined with distributed computing to accelerate the solution process:

[0247] Pop t+1 =Select(Combine(Pop) t Offspring t ))

[0248]

[0249] In the formula, Pop t Denotes the t-th generation population; Offspring t Rank represents the t-th generation descendant; i CD represents the non-dominant ranking rank of individual i. i Distance indicates the degree of crowding.

[0250] During the service recovery phase, execute the reconstruction plan and verify the recovery effect:

[0251] SR = SW close ,Seq,Veri

[0252] In the formula, SR represents the service recovery plan; SW close This represents the set of switches that need to be closed; Seq represents the closing sequence; Veri represents the verification measure.

[0253] Optimization of switching operation sequence takes into account transient effects and recovery benefits:

[0254]

[0255] In the formula, Impact(SW) t ) represents the transient effect of switching operation; Benefit(SW) t ) represents the recovery benefit brought about by the switching operation; α(t) and β(t) are time-varying weighting coefficients.

[0256] In this embodiment, the multi-objective optimization problem is handled by assigning the initial population to different computing nodes, with each node independently performing genetic operations such as selection, crossover, and mutation to generate a new offspring population. The optimal individual or population information is periodically exchanged between the nodes. The merged population is sorted and selected using the non-dominated sorting mechanism and crowding distance calculation of the improved NSGA-II algorithm to quickly find a uniformly distributed and high-quality Pareto front solution set.

[0257] In one alternative implementation, the multi-objective optimization problem is addressed by assigning an initial particle swarm to different computing nodes, with each node independently performing particle velocity and position update operations. Based on their respective local and global best positions, the nodes periodically exchange the position information of the optimal particles, update the particle velocity and position using the rules of particle swarm optimization, and quickly find a uniformly distributed and high-quality Pareto front solution set.

[0258] In another alternative implementation, the multi-objective optimization problem is handled by decomposing the original multi-objective optimization problem into multiple single-objective sub-problems, assigning these sub-problems to different computing nodes for parallel processing, with each node independently performing genetic operations, optimizing based on the objective function of its own sub-problem, periodically exchanging optimal individuals or local information between adjacent nodes, dynamically adjusting the weight vector as needed, and quickly finding a uniformly distributed and high-quality Pareto front solution set.

[0259] It should be noted that this invention employs an improved NSGA-II algorithm combined with distributed computing to accelerate the solution process, exhibiting significant advantages over distributed particle swarm optimization and decompositional multi-objective evolutionary algorithms when dealing with multi-objective optimization problems. NSGA-II effectively maintains population diversity through non-dominated sorting and crowding distance mechanisms, ensuring uniform distribution on complex Pareto fronts. In contrast, while distributed particle swarm optimization is easily parallelized, it is prone to getting trapped in local optima when dealing with high-dimensional multi-objective optimization problems and struggles to guarantee solution diversity. Compared to decompositional multi-objective evolutionary algorithms, which rely on precise weight vector design to achieve effective subproblem decomposition, NSGA-II does not require complex weight vector settings, reducing the need for human intervention and better adapting to different types of multi-objective optimization problems. The improved NSGA-II further enhances search efficiency and convergence speed through distributed computing, making it more robust and flexible in large-scale, complex environments.

[0260] Example 4, referring to Tables 1-3, is an embodiment of the present invention. This embodiment provides a method and system for assisting deep brain stimulation surgery and treatment based on digital twin technology. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through specific implementation methods and implementation effects.

[0261] The specific details of this embodiment are as follows:

[0262] To verify system performance, this invention constructs a simulation platform incorporating the IEEE 33-node system. Simulation experiments are conducted in the MATLAB environment, configured with a multi-agent simulation environment including 30 device-level agents, 10 feeder-level agents, and 3 region-level agents.

[0263] Table 1. Performance Comparison of Different Self-Healing Methods

[0264] index Centralized approach Distributed approach Hierarchical method Method of the present invention Calculation time (s) 12.5 5.2 7.8 3.6 Load recovery rate (%) 85.6 78.3 87.2 92.5 Communication overhead (KB) 356.8 128.5 245.6 152.3 Number of operations 15 21 14 12 System losses (kW) 245.6 312.8 238.5 225.7

[0265] Table 2 System response time (s) under different fault scenarios

[0266] Fault type Single Fault Multi-point failure Cascade Fault Extreme failure Centralized approach 8.5 15.3 25.7 42.8 Distributed approach 3.2 7.5 18.6 35.2 Hierarchical method 4.8 9.2 15.8 28.6 Method of the present invention 2.5 5.8 10.2 18.5

[0267] Table 3. Algorithm Scalability Analysis for Networks of Different Sizes

[0268] Network scale Calculation time (s) Memory usage (MB) Communication overhead (KB) Convergence cycles 33 nodes 2.5 45.6 125.8 15 69 nodes 4.8 78.2 195.6 22 123 nodes 8.6 128.5 285.3 28 200 nodes 15.4 215.8 425.6 35

[0269] Experimental results demonstrate that, compared with traditional methods, the proposed method exhibits significant advantages in computational efficiency, load recovery rate, and communication efficiency. Particularly in scenarios involving expanded network scale and complex faults, the proposed method demonstrates better scalability and robustness. In a 33-node system, the computation time of the proposed method is reduced by 71.2% compared to the centralized method, while the load recovery rate is improved by 6.9%. In multi-point fault scenarios, the proposed method reduces response time by 37.0% compared to the hierarchical method, while communication overhead is reduced by 38.0%. These results fully validate the effectiveness and advancement of the proposed multi-agent collaborative distributed self-healing evaluation and decision-making system for distribution networks.

[0270] Example 4 is an embodiment of the present invention. This embodiment differs from the first embodiment in that it provides a distributed self-healing evaluation and decision-making system for power distribution networks.

[0271] It should be noted that the technical solution of the distributed self-healing evaluation and decision-making system for distribution networks is based on the same concept as the technical solution of the distributed self-healing evaluation and decision-making method for distribution networks described above. For details not described in detail in the technical solution of the distributed self-healing evaluation and decision-making system for distribution networks in this embodiment, please refer to the description of the technical solution of the distributed self-healing evaluation and decision-making method for distribution networks described above.

[0272] This embodiment of a distributed self-healing evaluation and decision-making system for a distribution network includes:

[0273] The acquisition module is used to acquire local state information of the agent;

[0274] The distributed state evaluation module is used to perform distributed state evaluation based on the agent's local state information and a consensus algorithm to obtain the global state evaluation result.

[0275] The isolation scheme generation module is used to perform fault diagnosis and system isolation based on the global state assessment results, and obtain the optimal isolation scheme.

[0276] The distributed collaborative decision-making and optimization module is used to obtain the optimal power supply reconfiguration scheme by performing regional coordination and global optimization based on the optimal isolation scheme through intelligent agent communication and negotiation mechanism and distributed collaborative decision-making mechanism.

[0277] The self-healing execution and evaluation module is used to execute the optimal power reconfiguration scheme and obtain and evaluate the self-healing results.

[0278] This embodiment also provides an electronic device applicable to a distributed self-healing evaluation and decision-making method for power distribution networks, including:

[0279] The system includes a memory and a processor. The memory stores computer-executable instructions, and the processor executes these instructions to implement a distributed self-healing evaluation and decision-making method for a power distribution network as proposed in the above embodiments.

[0280] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements a distributed self-healing evaluation and decision-making method for a distribution network as proposed in the above embodiments.

[0281] The storage medium proposed in this embodiment and the method for implementing a distributed self-healing evaluation and decision-making system for power distribution networks proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0282] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0283] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for evaluating and making decisions on distributed self-healing in power distribution networks, characterized in that, include: Obtain local state information of the agent; Based on the local state information of the agent, a consensus algorithm is used to perform distributed state evaluation to obtain the global state evaluation result; Based on the global state assessment results, fault diagnosis and system isolation are performed to obtain the optimal isolation scheme; Based on the optimal isolation scheme, the optimal power supply reconfiguration scheme is obtained by regional coordination and global optimization through intelligent agent communication and negotiation mechanism and distributed collaborative decision-making mechanism. Execute the optimal power reconfiguration scheme, obtain and evaluate the self-healing results.

2. The distributed self-healing evaluation and decision-making method for distribution networks as described in claim 1, characterized in that: The obtained global state evaluation result includes: Obtain local state information of each agent; Based on the acquired data, each agent performs a preliminary estimate of the initial state based on its own data. Set the neighbor set, weighting coefficients, convergence threshold, and maximum number of iterations; The state estimates of each agent are updated synchronously using an improved consensus algorithm; The weighting coefficients are dynamically adjusted based on the state differences between the agents. Determine if the convergence condition has been met; if not, return to continue iterating. If convergence is achieved, the global state evaluation result is output.

3. The distributed self-healing evaluation and decision-making method for distribution networks as described in claim 1 or 2, characterized in that: The process of obtaining the optimal isolation scheme includes: Based on the state assessment results, a multi-feature fusion model is used to perform fault detection on multiple device-level agents to obtain fault detection results. Integrate fault detection results from multiple device-level intelligent agents to locate the faulty section; Based on the fault location results, the minimum cut set algorithm is applied to obtain the optimal isolation scheme.

4. The distributed self-healing evaluation and decision-making method for distribution networks as described in claim 3, characterized in that: The agent communication and negotiation mechanism includes: Determine the communication mode; Define a standard message format; Design a reliable transmission protocol, calculate the probability of successful transmission, and optimize the retransmission strategy; Introduce an event-triggered communication mechanism to optimize resource allocation; By introducing a reputation mechanism and setting a priority arbitration mechanism, a negotiation process based on the contract network is constructed to facilitate information exchange and task negotiation.

5. The distributed self-healing evaluation and decision-making method for distribution networks as described in claim 4, characterized in that: The distributed collaborative decision-making mechanism includes: The self-healing decision-making problem is modeled as a multi-agent cooperative game. Design the utility function for the intelligent agent; Each agent generates a set of candidate policies based on local information; All agents perform a global evaluation of the policy combination; The agent selects the optimal combination of strategies through a voting mechanism; Introducing a heuristic search strategy to accelerate convergence; The large problem is decomposed into multiple sub-regions, and the coordination between the sub-regions is carried out through the Lagrange relaxation algorithm. By alternating the update strategy and multiplier, the optimal power reconfiguration scheme is finally obtained.

6. The distributed self-healing evaluation and decision-making method for distribution networks as described in claim 5, characterized in that: The acquisition and evaluation of self-healing results includes: Define process evaluation dimensions, including response speed, collaboration efficiency, and resource utilization; Calculate response speed indicators, collaborative efficiency indicators, and resource utilization rate indicators; The process performance score is obtained by weighted combination; Define the dimensions for evaluating the results, including load recovery rate, network reliability, and economy; Calculate load recovery rate, network reliability, and economic indicators; The performance score of the result is obtained by weighted combination; The evaluation results are updated using a distributed consensus algorithm.

7. The distributed self-healing evaluation and decision-making method for distribution networks as described in claim 6, characterized in that: The improved consensus algorithm is expressed as follows: Where, x i (k) represents the state estimate of agent i in the kth iteration; N i Denotes the set of neighbors of agent i; a ij The weighting coefficients are represented by α, the acceleration factor is represented by w. j This represents the confidence weight of agent j.

8. A distributed self-healing evaluation and decision-making system for distribution networks, using the method described in any one of claims 1-7, characterized in that, include: The acquisition module is used to acquire local state information of the agent; The distributed state evaluation module is used to perform distributed state evaluation based on the agent's local state information and a consensus algorithm to obtain the global state evaluation result. The isolation scheme generation module is used to perform fault diagnosis and system isolation based on the global state assessment results, and obtain the optimal isolation scheme. The distributed collaborative decision-making and optimization module is used to obtain the optimal power supply reconfiguration scheme by performing regional coordination and global optimization based on the optimal isolation scheme through intelligent agent communication and negotiation mechanism and distributed collaborative decision-making mechanism. The self-healing execution and evaluation module is used to execute the optimal power reconfiguration scheme and obtain and evaluate the self-healing results.

9. An electronic device, characterized in that, include: Memory, used to store programs; A processor for loading the program to perform the steps of the method as claimed in any one of claims 1-7.

10. A computer-readable storage medium storing a program, characterized in that, When the program is executed by a processor, it implements the steps of the method as described in any one of claims 1-7.

Images