Heterogeneous multi-agent system structure reconstruction method, equipment, medium and product

Abstract

The invention discloses a heterogeneous multi-agent system structure reconstruction method and device, a medium and a product, and relates to the technical field of multi-agent systems and reliability, and the method comprises the steps: building a first target heterogeneous multi-agent network model when a running target heterogeneous multi-agent system is affected by disturbance and has a fault entity; randomly generating a plurality of reconstruction schemes according to the fault node and the entity object constraint; performing iterative optimization on the plurality of reconstruction schemes according to an elastic objective function and an optimization algorithm to obtain an optimal reconstruction scheme; wherein the elasticity objective function is used for calculating elasticity corresponding to each reconstruction scheme; and finally, according to the optimal reconstruction scheme, reconstructing the structure of the target heterogeneous multi-agent system. By means of the scheme, the structure of the artificial heterogeneous multi-agent system can be reconstructed, so that the system can still keep the basic functions after being disturbed, and self-repairing of the system is achieved.

Description

Technical Field

[0001] The present application relates to the field of multi-agent systems and reliability technology, and in particular to a method, device, medium and product for reconstructing the structure of a heterogeneous multi-agent system. Background Art

[0002] Multi-agent systems are a typical type of complex system, encompassing various systems in nature and human society, such as biological swarms, robotic swarms, and weapon swarms. Each agent in the system possesses independent perception, decision-making, and execution capabilities, and there is a rich interplay of information and collaboration among them. To achieve system efficiency and modularity, the overall functionality of a multi-agent system can often be decomposed into multiple sub-functions, including perception, decision-making, and execution, with these sub-functions allocated based on the capabilities and characteristics of each agent. For example, multiple sensors collect information from different locations or for different objects; multiple decision makers process the relevant data and make local control decisions based on local or global objectives; and multiple actuators execute the instructions they receive. The entire system achieves information sharing and coordinated control through a network, enabling efficient overall system management and mission execution. Such multi-agent systems are defined as heterogeneous multi-agent systems (HMS).

[0003] Artificial heterogeneous multi-agent systems need to be able to self-adjust and optimize based on environmental changes or internal states, thereby reconstructing a new system structure that both meets current needs and maintains good performance. Reconfiguration capability characterizes the adaptive and self-repair mechanisms of a multi-agent system, while system resilience, the ability to maintain basic functionality after perturbations, is closely related. However, heterogeneous multi-agent systems are complex and diverse, and a universally applicable reconfiguration method is currently lacking. Summary of the Invention

[0004] The purpose of this application is to provide a method, device, medium and product for reconstructing the structure of a heterogeneous multi-agent system, which can be used to reconstruct the structure of an artificial heterogeneous multi-agent system so that the system can still maintain its basic functions after being disturbed and achieve system self-repair.

[0005] To achieve the above objectives, this application provides the following solutions:

[0006] In a first aspect, the present application provides a method for reconstructing a heterogeneous multi-agent system structure, comprising:

[0007] When the running target heterogeneous multi-agent system is affected by a disturbance and a faulty node occurs, a first target heterogeneous multi-agent network model is constructed; the first target heterogeneous multi-agent network model includes nodes, edges and meta-paths; the nodes represent entities in the target heterogeneous multi-agent system; the nodes include target nodes, perception nodes, decision nodes and execution nodes; the edges represent the relationship between the nodes; the meta-path is an information transmission path with the perception node as the starting point and the execution node as the end point, and is composed of the perception node, decision node and execution node and the corresponding edges; the perception node is used to perceive the relevant information of the target node.

[0008] Several reconstruction schemes are randomly generated based on the faulty nodes and entity object constraints; each of the reconstruction schemes is a set of reconstruction support class nodes used to replace all faulty nodes; in the reconstruction scheme, one reconstruction support class node replaces one faulty node, and different reconstruction support nodes replace different faulty nodes; the entity object constraints are used to screen out reconstruction support class nodes of the same type as the faulty nodes from the first target heterogeneous multi-agent network model.

[0009] According to the elasticity objective function and the optimization algorithm, several reconstruction schemes are iteratively optimized to obtain the optimal reconstruction scheme; the elasticity objective function is used to calculate the elasticity corresponding to the reconstruction scheme; the elasticity is the ratio of the comprehensive performance evaluation index corresponding to the reconstruction scheme to the total reconstruction time; the comprehensive performance evaluation index is the sum of the performance evaluation indexes of the second target heterogeneous multi-agent network model; the second target heterogeneous multi-agent network model is a network model obtained by replacing the corresponding meta-path in the first target heterogeneous multi-agent network model with the reconstruction path; the corresponding meta-path is the meta-path passing through a certain fault node; the reconstruction path is a new meta-path obtained by replacing the certain fault node in the corresponding meta-path with a reconstruction support class node according to the reconstruction scheme; the performance evaluation index of the second target heterogeneous multi-agent network model is calculated based on the operating efficiency of each meta-path and reconstruction path in the second target heterogeneous multi-agent network model; the total reconstruction time is the time from the start of reconstruction to the completion of the startup of the reconstruction support class nodes corresponding to all fault nodes.

[0010] According to the optimal reconstruction scheme, the structure of the running target heterogeneous multi-agent system is reconstructed.

[0011] In a second aspect, the present application provides a computer device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-mentioned heterogeneous multi-agent system structure reconstruction method.

[0012] In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-mentioned method for reconstructing the structure of a heterogeneous multi-agent system.

[0013] In a fourth aspect, the present application provides a computer program product, comprising a computer program, which, when executed by a processor, implements the steps of the above-mentioned heterogeneous multi-agent system structure reconstruction method.

[0014] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0015] The present application provides a method, device, medium and product for reconstructing the structure of a heterogeneous multi-agent system. When the running target heterogeneous multi-agent system is affected by a disturbance and a faulty entity occurs, a first target heterogeneous multi-agent network model is constructed; the first target heterogeneous multi-agent network model includes nodes, edges and meta-paths; a number of reconstruction schemes are randomly generated according to the constraints of the faulty nodes and entity objects; then, according to the elasticity objective function and the optimization algorithm, the several reconstruction schemes are iteratively optimized to obtain the optimal reconstruction scheme; wherein the elasticity objective function is used to calculate the elasticity corresponding to each reconstruction scheme, and the elasticity is the ratio of the comprehensive performance evaluation index corresponding to the reconstruction scheme to the total reconstruction time; the comprehensive performance evaluation index is the sum of the performance evaluation indexes of each second target heterogeneous multi-agent network model; the second target The heterogeneous multi-agent network model is the network model obtained by replacing the corresponding meta-path in the first target heterogeneous multi-agent network model using a reconstruction path; the corresponding meta-path is the meta-path passing through a certain faulty node; the reconstruction path is the new meta-path obtained by replacing the certain faulty node in the corresponding meta-path with a reconstruction support node according to the reconstruction scheme; the performance evaluation index of the second target heterogeneous multi-agent network model is calculated based on the operating efficiency of each meta-path and the reconstruction path in the second target heterogeneous multi-agent network model; the total reconstruction time is the time from the start of reconstruction to the completion of the startup of the reconstruction support nodes corresponding to all faulty nodes; finally, according to the optimal reconstruction scheme, the structure of the running target heterogeneous multi-agent system is reconstructed. Through the above scheme, the structure of the artificial heterogeneous multi-agent system can be reconstructed, so that the system can still maintain its basic functions after being disturbed, and achieve system self-repair. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0017] Figure 1 A flowchart of a method for reconfiguring a heterogeneous multi-agent system structure provided in one embodiment of the present application;

[0018] Figure 2 A schematic flow chart of a method for reconfiguring a heterogeneous multi-agent system structure provided in another embodiment of the present application;

[0019] Figure 3 A schematic diagram of a reconfigurable operating loop model for a heterogeneous multi-agent system according to an embodiment of the present application;

[0020] Figure 4 A flowchart of establishing performance evaluation indicators for a heterogeneous multi-agent network model provided in one embodiment of the present application;

[0021] Figure 5 A flowchart for determining entity object constraints, time constraints, and elasticity objective functions provided in one embodiment of the present application;

[0022] Figure 6 A schematic diagram of a reconstruction support entity coding mode provided in an embodiment of the present application;

[0023] Figure 7 A flowchart of performing genetic operations on an initial population based on selection, crossover, and mutation operators provided in one embodiment of the present application;

[0024] Figure 8 A schematic diagram of a reconstruction support entity decoding mode provided in an embodiment of the present application;

[0025] Figure 9 A schematic diagram of the system structure of an emergency response system provided in one embodiment of the present application;

[0026] Figure 10 A schematic diagram of a heterogeneous multi-agent system structure reconstruction device provided in one embodiment of the present application;

[0027] Figure 11 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0028] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0029] In nature and human society, numerous multi-agent systems are able to self-adjust and optimize based on environmental changes and internal states. In particular, they dynamically adjust the roles, functions, and interactions of individual agents to reconstruct a new system structure that both meets current needs and offers superior performance. Reconfigurability characterizes the adaptive and self-repairing mechanisms of multi-agent systems, while the core of system resilience is the ability of a system to maintain its basic functionality after perturbations. The two are closely linked. Through reconfiguration, systems can quickly recover from adverse conditions, thereby reducing vulnerability and enhancing resilience. However, research on the reconfiguration of heterogeneous multi-agent systems presents numerous difficulties and challenges. The most significant is the complex structure of heterogeneous multi-agent systems, resulting in a large reconfiguration decision space. Existing methods for reconfiguring heterogeneous multi-agent systems suffer from limited effectiveness, low analytical efficiency, and poor interpretability and applicability.

[0030] Heterogeneous networks are a special form of complex network characterized by nodes and edges with varying attributes or types. Therefore, heterogeneous multi-agent systems can be constructed to abstract and quantify the characteristics and interactions of different elements in the system. The heterogeneity of network nodes and edges can be studied to reveal the behavior and functionality of heterogeneous multi-agent systems. The present invention aims to provide a highly interpretable and applicable method for reconstructing heterogeneous multi-agent systems.

[0031] For artificial multi-agent systems such as weapon systems, robot swarms, and emergency response systems, reconfiguration is a design decision. This application, inspired by biological multi-agent systems, proposes a heuristic-based structural reconfiguration method for heterogeneous multi-agent systems. This method implements a reconfiguration mechanism for most artificial heterogeneous multi-agent systems, ensuring that they maintain basic operational functionality and performance even after being disturbed or attacked.

[0032] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0033] In an exemplary embodiment, Figure 1 As shown, a method for reconstructing the structure of a heterogeneous multi-agent system is provided. The method is executed by a computer device, specifically a computer device such as a terminal or a server, or a terminal and a server. In an embodiment of the present application, the method includes the following steps S11 to S14. Among them:

[0034] Step S11: When the running target heterogeneous multi-agent system is affected by a disturbance and a faulty node occurs, a first target heterogeneous multi-agent network model is constructed; the first target heterogeneous multi-agent network model includes nodes, edges and meta-paths; the nodes represent entities in the target heterogeneous multi-agent system; the nodes include target nodes, perception nodes, decision nodes and execution nodes; the edges represent the relationship between the nodes; the meta-path is an information transmission path with the perception node as the starting point and the execution node as the end point, and is composed of the perception node, decision node and execution node and the corresponding edges; the perception node is used to perceive the relevant information of the target node.

[0035] In this application, a loop formed by a meta-path, a target node corresponding to the meta-path, and an edge connecting the meta-path and the target node is called a running loop.

[0036] Step S12: Randomly generate several reconstruction schemes based on the faulty nodes and entity object constraints; each of the reconstruction schemes is a set of reconstruction support class nodes used to replace all faulty nodes; in the reconstruction scheme, one reconstruction support class node replaces one faulty node, and different reconstruction support nodes replace different faulty nodes; the entity object constraints are used to screen out reconstruction support class nodes of the same type as the faulty nodes from the first target heterogeneous multi-agent network model.

[0037] Step S13: According to the elastic objective function and the optimization algorithm, several reconstruction schemes are iteratively optimized to obtain the optimal reconstruction scheme; the elastic objective function is used to calculate the elasticity value corresponding to the reconstruction scheme; the elasticity value is the ratio of the comprehensive performance evaluation index corresponding to the reconstruction scheme to the total reconstruction time; the comprehensive performance evaluation index is the sum of the performance evaluation indexes of the second target heterogeneous multi-agent network model; the second target heterogeneous multi-agent network model is a network model obtained by replacing the corresponding meta-path in the first target heterogeneous multi-agent network model with the reconstruction path; the corresponding meta-path is a meta-path passing through a certain fault node; the reconstruction path is a new meta-path obtained by replacing the certain fault node in the corresponding meta-path with a reconstruction support class node according to the reconstruction scheme; the performance evaluation index of the second target heterogeneous multi-agent network model is calculated based on the operating efficiency of each meta-path and reconstruction path in the second target heterogeneous multi-agent network model; the total reconstruction time is the time from the start of reconstruction to the completion of the startup of the reconstruction support class nodes corresponding to all fault nodes.

[0038] The replacement of all failed nodes starts simultaneously.

[0039] Step S14: Reconstruct the structure of the running target heterogeneous multi-agent system according to the optimal reconstruction scheme.

[0040] In an exemplary embodiment, the elastic objective function is:

[0041]

[0042] Among them, Re h represents the elasticity value corresponding to the reconstruction scheme h; P h ′ represents the comprehensive performance evaluation index corresponding to the reconstruction scheme h, P ij represents the performance evaluation index of the second target heterogeneous multi-agent network model obtained after replacing the faulty node i with the reconstruction support class node j, U represents the set of faulty nodes in the first target heterogeneous multi-agent network model, and R h represents the set of reconstruction support class nodes in the reconstruction scheme h; T rh Represents the total reconstruction time corresponding to the reconstruction scheme h.

[0043] In an exemplary embodiment, the performance evaluation index P of the second target heterogeneous multi-agent network model is ij The calculation formula is as follows:

[0044]

[0045] Among them, N T represents the number of target nodes in the set T of target nodes in the second target heterogeneous multi-agent network model; t represents the operating efficiency of the target node t, which is determined based on the operating efficiencies of the meta-paths and reconstructed paths corresponding to the target node t; g t It represents the number of meta-paths and reconstructed paths corresponding to the target node t, and is also the number of running loops whose starting and ending points are the target node t.

[0046] That is, the performance of the entire second-objective heterogeneous multi-agent network model is expressed as the number of efficiency-weighted running loops, which is equal to the sum of the number of meta-paths and the number of reconstructed paths.

[0047] In an exemplary embodiment, the operating efficiency v of the target node t is t The determination process is:

[0048] Step S21: determining the time factors of the meta-paths and the reconstructed paths corresponding to the target node t according to the load factors and load capacity factors of the nodes in the meta-paths and the reconstructed paths corresponding to the target node t.

[0049] Step S22: determining the operating efficiency of each of the meta-paths and the reconstructed path corresponding to the target node t according to the time factor of each of the meta-paths and the reconstructed path corresponding to the target node t; the operating efficiency is inversely proportional to the time factor.

[0050] Step S23: The minimum value of the operating efficiency of each of the meta-paths and the reconstructed path corresponding to the target node t is determined as the operating efficiency v of the target node t. t .

[0051] In an exemplary embodiment, step S21 specifically includes:

[0052] Step S211: Calculate the load factors of the nodes in the meta-path and the reconstructed path according to the load factor calculation formula; the load factor calculation formula is:

[0053]

[0054] Among them, SI k represents the load factor of node k; f t k It represents the number of meta-paths and reconstructed paths corresponding to the target node t that pass through node k, and the number of running loops that pass through node k (node ​​k is a node other than the target node) and whose starting and ending points are the target node t.

[0055] Since the operation loop actually represents the task operation process of an entity in a multi-agent system, the operation load of an entity is related to the number of meta-paths passing through the entity. Therefore, referring to the concept of node betweenness centrality (which can measure the information transmission ability of a node and is defined as the frequency of the shortest path passing through the node), this application proposes a node load factor, which is defined as the frequency of meta-paths passing through the node. Based on the node load factor, this application further proposes a node load weight, see step S212.

[0056] Step S212: Calculate the load weight of the node according to the load weight calculation formula and the load factor of the node; the load weight calculation formula is:

[0057]

[0058] Among them, ω k represents the load weight of node k; CI k It represents the load capacity factor of node k, which indicates the ability of node k to bear the task load and is determined by the inherent performance of the node.

[0059] In some embodiments, when the load weight of the node at the initial moment is 1, it means that the load capacity of the node at the initial moment just meets the initial task load of the node.

[0060] Step S213: Sum the load weights of all nodes in the meta-path or the reconstructed path to obtain a time factor of the meta-path or the reconstructed path. The time factor of the meta-path or the reconstructed path is also the time factor of the corresponding running ring.

[0061] For a node, the more meta-paths that pass through it, the heavier its load is. As a result, the running time of the meta-path increases due to the influence of the load. Therefore, the running time of the meta-path is proportional to the load weight. i The total running time is expressed by the time factor:

[0062]

[0063] Among them, ε i Represents the metapath MP i time factor.

[0064] For a target node t, the completion time of the task is determined by the meta-path with the longest running time corresponding to the target node t. At the same time, the running efficiency is inversely proportional to the running time. Therefore, the running efficiency of the target node t is expressed as:

[0065]

[0066] in, Indicates the maximum time factor among the time factors of each meta-path (including the reconstructed path) corresponding to the target node t.

[0067] In an exemplary embodiment, the total reconstruction time T corresponding to the reconstruction scheme h is rh The calculation formula is as follows:

[0068] T rh =Max(t ij +t cj ),i∈U,j∈R h ;

[0069] Among them, t ij represents the allocation time to replace the faulty node i with the corresponding reconstruction support node j, t cj Represents the startup time of the reconstruction support class node j; Max(·) represents the maximum. Total reconstruction time T rh Represents the total time to replace all faulty nodes with corresponding reconstruction support class nodes.

[0070] In an exemplary embodiment, the optimization algorithm is a genetic algorithm; and iteratively optimizing the first number of reconstruction schemes according to the elastic objective function and the optimization algorithm to obtain the optimal reconstruction scheme specifically includes:

[0071] Step S131: performing a crossover operation on individuals in a population corresponding to the last number of iterations according to a crossover operator to obtain individuals after the crossover operation; the population includes a plurality of individuals; the individuals represent reconstruction schemes.

[0072] Step S132: performing a mutation operation on the individuals after the crossover operation according to the mutation operator to obtain individuals after the mutation operation, and establishing a population corresponding to the current number of iterations based on the individuals after the mutation operation.

[0073] Step S133: Calculate the elasticity value of each individual in the population corresponding to the current number of iterations according to the elasticity objective function.

[0074] Step S134: Determine whether the current number of iterations reaches the preset number of iterations.

[0075] Step S135: If the judgment result is no, then determine the updated population after the current number of iterations based on the elasticity value and selection operator of each individual in the population corresponding to the current number of iterations, and update the population corresponding to the previous number of iterations to the updated population after the current number of iterations, add one to the current number of iterations, and return to step S135 to perform a mutation operation on the individuals after the crossover operation according to the mutation operator to obtain the individuals after the mutation operation.

[0076] Step S136: If the judgment result is yes, the individual with the largest elasticity value in the population corresponding to the current number of iterations is determined as the optimal reconstruction solution.

[0077] In another exemplary embodiment of the present application, Figure 2 As shown in Figure 2, the heterogeneous multi-agent system structure reconstruction method can also be implemented through the following steps:

[0078] Step S101: Establish a reconfigurable heterogeneous network model of the multi-agent system (i.e., a heterogeneous multi-agent network model, referred to as a heterogeneous multi-agent network model, network model, or network) according to the multi-agent system structure.

[0079] The network model includes the node type, the number of nodes, and the edges connecting the nodes.

[0080] Step S102: Using the task load as the weight, establish the performance evaluation index of the heterogeneous multi-agent network model based on the number of running loops.

[0081] Step S103: Determine entity object constraints, time constraints, and elasticity objective functions for network reconstruction.

[0082] Step S104: randomly generate a number of initial solutions (individuals) that meet the constraints based on the reconstructed supporting entity coding mode to form an initial population.

[0083] Step S105: performing genetic operations on the initial population based on the reconstruction support entity (ie, reconstruction support class node) selection, crossover, and mutation operators to generate the next generation population.

[0084] Step S106: After several rounds of iteration, the solution with the best elasticity in the population is output.

[0085] Step S107: Generate a reconstruction solution with optimal flexibility based on the reconstruction support entity decoding mode.

[0086] The following is an introduction using an emergency response system as an example.

[0087] First, a corresponding network model is constructed based on the emergency response system. The cyclical process formed by the three stages or functions of perception, decision-making, and execution is a fundamental cyclical process that is ubiquitous in automation, system artificial intelligence, and even biological systems. The emergency response system includes various types of sensors, decision-making bodies, and execution organizations. The relationship between sensors, decision-making bodies, execution organizations, and response targets (such as fires) can be represented by an operational loop model. Furthermore, functional redundancy is considered for each entity in the emergency response system to support functional replacement by redundant entities after a failure. Figure 3 This is a reconfigurable operating ring model proposed in the embodiment of the present application, which is suitable for modeling the reconstruction mechanism of heterogeneous multi-agent systems such as emergency response systems.

[0088] Based on the reconfigurable operating ring model, the embodiment of the present invention provides the elements of a heterogeneous multi-agent network model of an emergency response system, including the following:

[0089] 1) Represents a heterogeneous multi-agent network model, including node set V, edge set E, and point type mapping Edge type mapping ψ, reconstruction mapping δ.

[0090] 2) The point set V of the heterogeneous multi-agent network model based on the reconfigurable operation ring includes five types of nodes: perception, decision, execution, goal, and reconstruction support, that is, V = T∪S∪D∪A∪R = {v1,v2,v3,...,v N}.in, is the set of perception nodes;

[0091] is the set of decision nodes;

[0092] A collection of execution class nodes;

[0093] To reconstruct the support class nodes;

[0094] is the set of target nodes.

[0095] Use N T =|T|、N S =|S|、N D =|D|、N A =|A|、N R =|R| to represent the number of target nodes, perception nodes, decision nodes, execution nodes, and reconstruction support nodes in the heterogeneous multi-agent network model, then the number of nodes in the heterogeneous multi-agent network model N=N S +N D +N A +N R +N T .

[0096] 3) The edge set E={e1,e2,e3,...,e j}, which contains l T→S , l S→D , l S→S , l D→S , l D→D , l D→A , l A→T , l T→R , l R→D , l S→R , l R→A , l D→R , l R→T Thirteen other types of edges.

[0097] 4) Point type mapping of heterogeneous multi-agent network model based on reconfigurable operation ring It is a point set to point type set I={i T ,i S ,i D ,i A ,i R}, the specific expression is

[0098] 5) The edge type mapping ψ of the heterogeneous multi-agent network model based on the reconfigurable running ring is a mapping function from an edge set to an edge class set L. The specific expression is ψ=E→L, ψ(e)∈L, edge class set L={l T→S ,l S→D ,l S→S ,l D→S ,l D→D ,l D→A ,l A→T ,l T→R ,l R→D ,l S→R ,lR→A ,l D→R ,l R→T}.

[0099] 6) The reconstruction mapping δ of the heterogeneous multi-agent network model based on the reconfigurable running ring indicates that the reconstruction point set can be mapped to the three point type sets H = {i S ,i D ,i A}, the expression is δ=R→H, δ(v R )∈H.

[0100] In some embodiments, the reconstruction support class node itself can also be a perception class node, a decision class node, and an execution class node in the system. For example, the perception class node a1 in the system is also a reconstruction support class node. When a2, which is also a perception class node, fails, a1 replaces a2 to implement the function of a2. At this time, a1 takes into account the functional implementation of both a1 and a2.

[0101] In this application, the reconstruction path is proposed based on the heterogeneous network meta-path, which is as follows:

[0102] Definition of meta path:

[0103]

[0104] Refactor path definition:

[0105]

[0106] Among them, MP represents a set of network edges, which represents a meta-path in the network, v1∈S, v u ∈D,v u+1 ∈A; RP represents the node v containing the reconstruction support i A meta-path.

[0107] Based on the establishment of the network model of the emergency response system, a network model performance evaluation index is constructed to provide evaluation parameters for the heuristic-based reconstruction optimization process. Figure 4 , Figure 4 This is a flowchart of a specific implementation of step S102 provided in this embodiment. Step S102 may include the following steps:

[0108] Step S201: Establishing a network node load factor index.

[0109] Step S202: establishing a network node load weight index based on the network node load factor index.

[0110] Step S203: Based on the network node load weight index, a total running time factor index of the network meta-path (ie, the time factor of the meta-path in the above embodiment) is established.

[0111] Step S204: Based on the total running time factor index of the network element path, a network node running efficiency index (ie, the running efficiency of the target node in the above embodiment) is established.

[0112] Step S205: Based on the network node operation efficiency index, establish an efficiency-weighted network performance evaluation index.

[0113] Based on the performance evaluation indicators of the heterogeneous multi-agent network model, we can determine the entity object constraints and time constraints, and determine the elastic objective function of network reconstruction according to the fluctuation of the network performance indicators in the actual operation process. Figure 5 , Figure 5 This is a flowchart of a specific implementation of step S103 provided in an embodiment of the present application. Step S103 may include the following steps:

[0114] Step S301: Determine entity object constraints for network reconstruction.

[0115] Step S302: Determine the time constraint of network reconfiguration.

[0116] Step S303: Determine the elasticity objective function of network reconstruction.

[0117] The entity object constraint means that the reconstruction support class node must be of the same type as the fault node, which can be expressed by the following function:

[0118]

[0119] Among them, i∈R, j∈U, U represents the set of fault node numbers after damage, R represents the set of reconstruction support node numbers, δ(·), They are reconstruction mapping and node type mapping respectively.

[0120] The time constraint means that since the reconstruction scheme starts at the same time, that is, the replacement of all failed nodes is started at the same time, the total reconstruction time is equal to the longest reconstruction time. The function expression is shown in the above embodiment.

[0121] The elastic objective function is used to evaluate the reconstruction effect of heterogeneous multi-agent network models after being disturbed or attacked.

[0122] After determining the entity object constraints, time constraints, and elastic objective function of network reconstruction, the reconstruction problem can be solved based on a heuristic algorithm, as follows:

[0123] First, based on the reconstruction support entity encoding mode, a number of initial individuals that meet the entity object constraints are randomly generated to form the initial population. Figure 6 , Figure 6 Schematic diagram of the reconstruction support entity encoding mode provided in an embodiment of the present application. Specifically, assuming there are N reconstruction support nodes and M fault nodes, and each node has a number, each fault node requires a reconstruction support node to re-form a complete running ring element path (i.e., reconstruction path). According to the reconstruction support node numbers assigned to the fault nodes, they are arranged in sequence to form a coding sequence. If the fault node does not have a corresponding reconstruction support node assigned, the code is 0.

[0124] After randomly generating a number of initial individuals that meet the constraints based on the reconstruction support entity encoding mode to form the initial population, the initial population is genetically operated based on the elastic objective function and selection, crossover and mutation operators to generate the next generation population. Figure 7 , Figure 7 The flowchart of step S105 provided in the embodiment of the present application may include the following steps:

[0125] Step S501: Selecting a solution in the population based on the elastic objective function. Specifically, the solution in the population is selected based on the elasticity value of each solution in the population and a selection operator, where the elasticity value of each solution is calculated by the elastic objective function.

[0126] Step S502: performing a crossover operation on the selected solution to obtain a solution after the crossover operation.

[0127] Step S503: performing a mutation operation on the solution after the crossover operation to obtain a solution after the mutation operation.

[0128] Furthermore, the selection operator, crossover operator, and mutation operator may be a hybrid selection operator, a partial match crossover operator, and a mutation operator. It is understood that the selection of the above genetic operation operators is only an example provided in the embodiments of this application, and those skilled in the art may select other operators to perform genetic operations on the population according to actual conditions.

[0129] Repeat the above genetic operation, and after several rounds of iterations, determine the solution with the best elasticity from the population obtained in the last iteration.

[0130] After determining the most flexible solution, the most flexible reconstruction solution can be generated based on the corresponding reconstruction support entity decoding mode. Figure 8 , Figure 8The schematic diagram of the reconstruction scheme with the best elasticity based on the reconstruction support entity decoding mode provided in the embodiment of the present application. Specifically, the decoding process mainly obtains the reconstruction support class node corresponding to each fault node according to the reconstruction support class node coding sequence, forms a reconstruction scheme, and adjusts the heterogeneous multi-agent system. At the same time, Figure 8 As shown, the reconstruction time (t ij +t c ), sample the heterogeneous multi-agent network model according to the reconstruction time, and perform performance evaluation to obtain the performance value (P ij In this embodiment, it is assumed that the startup time of all reconstruction support class nodes is the same, so it is uniformly represented by tc. Of course, in other scenarios, the startup time of different reconstruction support class nodes may be different.

[0131] To help understand the solution of this application, let's use a case study to further illustrate this. Consider a city's fire emergency response system, which has four operational loops. Each loop contains ten sensing entities, such as temperature sensors or cameras, one decision-making entity, namely the emergency command center, and one execution entity, namely an on-duty fire brigade. The execution entity has four reconfiguration support entities, namely four standby fire brigades. The operational loops in this fire emergency response system operate as follows: When a fire occurs, the fire is the target entity. Information about the fire is sensed by the temperature sensor or camera (sensing entity) and transmitted to the emergency command center (decision-making entity). The emergency command center then directs the on-duty fire brigade (execution entity) to carry out firefighting tasks. If the on-duty fire brigade is unable to execute the operational loop due to casualties, it is considered an execution entity failure and needs to be reconfigured from the standby fire brigade or another on-duty fire brigade in the operational loop. A reconfiguration plan is then randomly generated, where the failed execution entity is reconfigured by a specific entity, which can be the standby fire brigade or another on-duty fire brigade in the operational loop, for a total of seven execution entities. The performance of the fire emergency response system after reconstruction according to the reconstruction scheme is calculated, namely the number of efficiency-weighted operation cycles (performance evaluation indicator). The elastic objective function value is calculated based on the total reconstruction time. The reconstruction scheme with the largest elastic objective function value (i.e., the optimal) is selected for reconstruction, which means that a fire brigade is assigned to replace the injured fire brigade to execute the operation cycle.

[0132] In another case, a multi-agent system of an emergency response system includes 9 perception entities, 9 decision entities, 9 execution entities and 9 target entities. The specific system structure is as follows: Figure 9 As shown. Figure 9In the equation, Si represents the sensing entity (sensing node), Di represents the decision-making entity (decision-making node), Ai represents the execution entity (execution node), and Ti represents the target entity (target node). Initially, the multi-agent system has 9 target entities and 37 operation loops. The load weight of each entity is 1. The operation efficiency of each target entity and the number of operation loops are shown in Table 1.

[0133] Table 1 Number of running loops and running efficiency of target entities

[0134] Target Entity T1 T2 T3 T4 T5 T6 T7 T8 T9 Number of running rings 6 11 2 3 3 4 4 3 1 Operational efficiency 0.25 0.2 0.33 0.25 0.33 0.2 0.2 0.25 0.33

[0135] The final performance of the multi-agent system was 0.978.

[0136] Based on the same inventive concept, the embodiments of the present application also provide a heterogeneous multi-agent system structure reconstruction device for implementing the heterogeneous multi-agent system structure reconstruction method involved above. The implementation solution provided by this device is similar to the implementation solution described in the above method. Therefore, the specific limitations in one or more embodiments of the heterogeneous multi-agent system structure reconstruction device provided below can be found in the above-mentioned limitations on the heterogeneous multi-agent system structure reconstruction method, and will not be repeated here.

[0137] In an exemplary embodiment, Figure 10 As shown, a heterogeneous multi-agent system structure reconstruction device 700 is provided, comprising:

[0138] The network construction module 701 constructs a reconfigurable heterogeneous multi-agent network model (RHMN) for a specific system structure, which is used to characterize the static and dynamic operation characteristics of the system structure.

[0139] The performance evaluation module 702 is used to evaluate the network operation capability characteristics and provide a target for reconstruction.

[0140] The reconstruction solution generating module 703 is used to generate a reconstruction solution based on a heuristic algorithm.

[0141] The network adjustment module 704 is used to control the readjustment of the multi-agent network model structure.

[0142] The present application also provides an application scenario, which applies the above-mentioned heterogeneous multi-agent system structure reconstruction method. Specifically: the heterogeneous multi-agent system structure reconstruction method provided in this embodiment can be applied in the heterogeneous multi-agent system structure reconstruction scenario. The heterogeneous multi-agent system structure reconstruction scenario includes a disturbance impact confirmation link and a heterogeneous multi-agent system structure reconstruction link; in the disturbance impact confirmation link, the fault nodes affected by the disturbance are determined, and in the heterogeneous multi-agent system structure reconstruction link, the heterogeneous multi-agent system structure is reconstructed based on the determined fault nodes. The heterogeneous multi-agent system structure reconstruction method provided in this embodiment belongs to the heterogeneous multi-agent system structure reconstruction link. Specifically, a first target heterogeneous multi-agent network model is constructed, and several reconstruction schemes are randomly generated according to the fault nodes and entity object constraints. According to the elastic objective function and the optimization algorithm, several reconstruction schemes are iteratively optimized to obtain the optimal reconstruction scheme, and then the structure of the target multi-agent system is reconstructed.

[0143] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 11 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O) and a communication interface. The processor, memory and input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used for related data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for reconstructing the structure of a heterogeneous multi-agent system is implemented.

[0144] Those skilled in the art will understand that Figure 11 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present application and does not constitute a limitation on the computer device to which the solution of the present application is applied. A specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement. In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps of the above-mentioned method embodiments when executing the computer program.

[0145] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0146] In an exemplary embodiment, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0147] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0148] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0149] The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processors involved in the various embodiments provided herein may include, but are not limited to, general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic units, data processing logic units based on quantum computing, and the like.

[0150] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0151] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above examples is only intended to help understand the method and core concept of this application. At the same time, for those skilled in the art, based on the concept of this application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for reconstructing the structure of a heterogeneous multi-agent system, characterized in that: The heterogeneous multi-agent system structure reconstruction method comprises: When the running target heterogeneous multi-agent system is affected by a disturbance and a faulty node appears, a first target heterogeneous multi-agent network model is constructed; the first target heterogeneous multi-agent network model includes nodes, edges and meta-paths; the nodes represent entities in the target heterogeneous multi-agent system; the nodes include target nodes, perception nodes, decision nodes and execution nodes; the edges represent the relationship between the nodes; the meta-path is an information transmission path consisting of perception nodes, decision nodes and execution nodes and corresponding edges, with the perception node as the starting point and the execution node as the end point; the perception node is used to perceive the relevant information of the target node; A plurality of reconstruction schemes are randomly generated according to the faulty nodes and entity object constraints; each of the reconstruction schemes is a set of reconstruction support class nodes used to replace all faulty nodes; in the reconstruction schemes, one reconstruction support class node replaces one faulty node, and different reconstruction support nodes replace different faulty nodes; the entity object constraints are used to filter out reconstruction support class nodes of the same type as the faulty nodes from the first target heterogeneous multi-agent network model; According to the elastic objective function and the optimization algorithm, several reconstruction schemes are iteratively optimized to obtain the optimal reconstruction scheme; the elastic objective function is used to calculate the elasticity corresponding to the reconstruction scheme; the elasticity is the ratio of the comprehensive performance evaluation index corresponding to the reconstruction scheme to the total reconstruction time; the comprehensive performance evaluation index is the sum of the performance evaluation indexes of the second target heterogeneous multi-agent network model; the second target heterogeneous multi-agent network model is a network model obtained by replacing the corresponding meta-path in the first target heterogeneous multi-agent network model with the reconstruction path; the corresponding meta-path is a meta-path passing through a certain fault node; the reconstruction path is a new meta-path obtained by replacing the certain fault node in the corresponding meta-path with a reconstruction support class node according to the reconstruction scheme; the performance evaluation index of the second target heterogeneous multi-agent network model is calculated based on the operating efficiency of each meta-path and the reconstruction path in the second target heterogeneous multi-agent network model; the total reconstruction time is the time from the start of reconstruction to the completion of the startup of the reconstruction support class nodes corresponding to all fault nodes; According to the optimal reconstruction scheme, the structure of the running target heterogeneous multi-agent system is reconstructed.

2. The method for reconstructing the structure of a heterogeneous multi-agent system according to claim 1, characterized in that: The elastic objective function is: Among them, Re h represents the elasticity corresponding to the reconstruction scheme h; P h ′ represents the comprehensive performance evaluation index corresponding to the reconstruction scheme h, P ij represents the performance evaluation index of the second target heterogeneous multi-agent network model obtained after replacing the faulty node i with the reconstruction support node j, U represents the set of faulty nodes in the first target heterogeneous multi-agent network model, and R h represents the set of reconstruction support class nodes in the reconstruction scheme h; T rh Represents the total reconstruction time corresponding to the reconstruction scheme h.

3. The method for reconstructing the structure of a heterogeneous multi-agent system according to claim 2, characterized in that: The performance evaluation index P of the second target heterogeneous multi-agent network model ij The calculation formula is as follows: Among them, N T represents the number of the target nodes in the set T of the target nodes in the second target heterogeneous multi-agent network model; v t represents the operating efficiency of the target node t, and the operating efficiency of the target node t is determined according to the operating efficiencies of each meta-path and the reconstructed path corresponding to the target node t; g t Represents the number of meta-paths and reconstructed paths corresponding to the target node t.

4. The method for reconstructing the structure of a heterogeneous multi-agent system according to claim 3, characterized in that: The operating efficiency v of the target node t t The determination process is: Determine the time factor of each of the meta-paths and the reconstructed path corresponding to the target node t according to the load factor and the load capacity factor of the node in each of the meta-paths and the reconstructed path corresponding to the target node t; Determining the operation efficiency of each of the meta-paths and the reconstructed path corresponding to the target node t according to the time factor of each of the meta-paths and the reconstructed path corresponding to the target node t; the operation efficiency is inversely proportional to the time factor; The minimum value of the operating efficiencies of the meta-paths and the reconstructed paths corresponding to the target node t is determined as the operating efficiency v of the target node t. t .

5. The method for reconstructing the structure of a heterogeneous multi-agent system according to claim 4, characterized in that: The determining, according to the load factors and load capacity factors of the nodes in the meta-paths and the reconstructed paths corresponding to the target node t, the time factors of the meta-paths and the reconstructed paths corresponding to the target node t specifically includes: According to the load factor calculation formula, the load factors of the nodes in the meta-path and the reconstructed path are calculated; the load factor calculation formula is: Among them, SI k represents the load factor of node k; Indicates the number of nodes k passed through in the meta-path corresponding to the target node t and the reconstructed path; The load weight of the node is calculated according to the load weight calculation formula and the load factor of the node; the load weight calculation formula is: Among them, ω k represents the load weight of node k; CI k represents the load capacity factor of node k; The load weights of all nodes in the meta-path or the reconstructed path are summed to obtain a time factor of the meta-path or the reconstructed path.

6. The method for reconstructing the structure of a heterogeneous multi-agent system according to claim 2, characterized in that: The total reconstruction time T corresponding to the reconstruction scheme h rh The calculation formula is as follows: T rh =Max(t ij +t cj ),i∈U,j∈R h ; Among them, t ij represents the time to replace the faulty node i with the corresponding reconstruction support node j, t cj represents the startup time of the reconstruction support class node j; Max(·) means taking the maximum value.

7. The method for reconstructing the structure of a heterogeneous multi-agent system according to claim 1, characterized in that: The optimization algorithm is a genetic algorithm; according to the elastic objective function and the optimization algorithm, the first number of reconstruction schemes are iteratively optimized to obtain the optimal reconstruction scheme, specifically including: According to the crossover operator, a crossover operation is performed on individuals in a population corresponding to the previous iteration number to obtain individuals after the crossover operation; the population includes a plurality of individuals; the individuals are reconstruction schemes; According to the mutation operator, a mutation operation is performed on the individuals after the crossover operation to obtain individuals after the mutation operation, and a population corresponding to the current number of iterations is established according to the individuals after the mutation operation; According to the elastic objective function, calculate the elasticity value of each individual in the population corresponding to the current number of iterations; Determine whether the current number of iterations reaches the preset number of iterations; If the judgment result is no, then according to the elasticity value and selection operator of each individual in the population corresponding to the current iteration number, determine the population updated by the current iteration number, and update the population corresponding to the previous iteration number to the population updated by the current iteration number, add one to the current iteration number, return to the step, perform mutation operation on the individuals after the crossover operation according to the mutation operator, and obtain the individuals after the mutation operation; If the judgment result is yes, the individual with the largest elasticity value in the population corresponding to the current number of iterations is determined as the optimal reconstruction solution.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the heterogeneous multi-agent system structure reconstruction method described in any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for reconstructing the structure of a heterogeneous multi-agent system described in any one of claims 1 to 7 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for reconstructing the structure of a heterogeneous multi-agent system described in any one of claims 1 to 7 is implemented.

Images