Unmanned cluster toughness evaluation method based on task network efficiency

Abstract

The invention provides an unmanned cluster toughness evaluation method based on task network efficiency, and belongs to the technical field of unmanned clusters. The method comprises the following steps that a task network model of an unmanned cluster is constructed, the task network model comprises a topological network structure sequentially composed of a detection layer, a decision layer, an execution layer and a task layer, the topological network structure comprises different types of nodes, all the nodes are in communication connection, and all the nodes have multiple association modes; designing a node failure strategy and a node recovery strategy according to the topology network structure; according to the node failure strategy and the node recovery strategy, an unmanned cluster performance index is formulated, and according to the dynamic change of the unmanned cluster performance index, an unmanned cluster toughness index is formulated; and carrying out toughness evaluation on the unmanned cluster by using the unmanned cluster toughness index. According to the method, the performance of the unmanned cluster can be effectively recovered, and the toughness of the task network model is remarkably improved.

Description

Technical Field

[0001] This application relates to the field of unmanned cluster technology, and particularly to a method for evaluating the resilience of an unmanned cluster based on task network efficiency. Background Art

[0002] Task network efficiency is a key indicator for measuring the performance of an unmanned cluster task network, directly reflecting the information collaboration ability and rapid response ability among units within the unmanned cluster. A highly efficient unmanned cluster task network can ensure the rapid transmission of instructions and information among multiple unmanned units, thereby significantly enhancing the overall task effectiveness of the unmanned cluster. Evaluating the resilience of an unmanned cluster by studying the task network efficiency of the unmanned cluster not only helps to enhance the stability and reliability of the unmanned cluster in complex environments but also reflects the efficient autonomous collaborative planning ability and excellent dynamic adaptation ability of the unmanned cluster.

[0003] The resilience of an unmanned cluster refers to the ability of the unmanned cluster to comprehensively utilize the cluster absorption ability to resist internal failures and external abnormal factor interferences under a specific task background, and to use adaptive technologies to reduce the impacts caused by these interferences, and to dynamically reorganize the structure of the unmanned cluster through the cluster recovery ability to improve the task effectiveness of the unmanned cluster, ensuring that tasks in different situations can be successfully completed.

[0004] As a comprehensive indicator of an unmanned cluster, resilience reflects the ability of the unmanned cluster to resist, absorb, adapt to, and recover from interferences. Currently, the quantitative evaluation methods for the resilience of unmanned clusters in the academic community are mainly divided into three categories: resilience evaluation methods based on index analysis, resilience evaluation methods based on simulation, and resilience evaluation methods based on performance curves. Although these resilience evaluation methods all have their own advantages, they all have certain problems, which are specifically as follows:

[0005] First, for the resilience evaluation method based on index analysis, a complete set of index systems is constructed, standardized basic indexes are established, and the aggregation of indexes is realized. The resilience of the unmanned cluster is measured by obtaining the final top-level evaluation value. However, this resilience evaluation method often needs to be combined with other resilience evaluation methods and cannot be evaluated alone;

[0006] Second, for the resilience evaluation method based on simulation, within the framework of a task simulation system, by capturing the data generated during the simulation process and performing similar statistical analysis and evaluation calculations on these data, the resilience value of the task system is deduced. Although this resilience evaluation method can theoretically clearly reflect the task dynamic evolution process of the unmanned cluster, in practical applications, there are problems such as high complexity of the simulation system, difficult solution, and poor interpretability of the internal working mechanism.

[0007] Finally, the resilience evaluation method based on the performance curve originated from the resilience triangle model proposed by Bruno and introduced the performance curve of the integral model to evaluate resilience. This is the most commonly used method for evaluating the resilience of unmanned clusters in the academic community. However, this resilience evaluation method is too complex in data acquisition and processing, resulting in high computational costs and low accuracy, limiting its scope of application.

[0008] Therefore, it is necessary to propose a solution to improve one or more problems existing in the above related technical solutions.

[0009] It should be noted that the information disclosed in the above background art section is only used to enhance the understanding of the background of the present application, and thus may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0010] An embodiment of the present application provides a method for evaluating the resilience of an unmanned cluster based on the efficiency of a task network, including the following steps:

[0011] Construct a task network model of the unmanned cluster, where the task network model includes a topological network structure successively composed of a detection layer, a decision layer, an execution layer, and a task layer; the topological network structure includes different types of nodes, all the nodes are communicatively connected, and there are multiple association modes between all the nodes;

[0012] According to the topological network structure, design node failure strategies and node recovery strategies;

[0013] According to the node failure strategies and the node recovery strategies, formulate performance indicators of the unmanned cluster, and according to the dynamic changes of the performance indicators of the unmanned cluster, formulate resilience indicators of the unmanned cluster;

[0014] Use the resilience indicators of the unmanned cluster to evaluate the resilience of the unmanned cluster.

[0015] In an exemplary embodiment of the present application, the types of the nodes include: the detection layer includes multiple sensing nodes, the decision layer includes multiple decision nodes, the execution layer includes multiple execution nodes, and the task layer includes multiple target nodes;

[0016] The set of all the sensing nodes is represented as: S = {S 1 , S 2 , …, S i , …, S j , …, S I}, S I represents the S I th sensing node, and I represents the number of sensing nodes;

[0017] The set of all the decision nodes is represented as: D = {D1 , D 2 , …, D i , …, D j , …, D J}, D J represents the D J -th decision node, and J represents the number of decision nodes;

[0018] The set of all the execution nodes is represented as: W = {W 1 , W 2 , …, W i , …, W j , …, W N}, W N represents the W N -th execution node, and N represents the number of execution nodes;

[0019] The set of all the target nodes is represented as: T = {T 1 , T 2 , …, T i , …, T j , …, T M}, T M represents the T M -th target node, and M represents the number of target nodes.

[0020] In an exemplary embodiment of the present application, the multiple association patterns include:

[0021] T i → S i , indicating that the S i -th sensing node detects the T i -th target node and obtains relevant information from the T i -th target node;

[0022] S i → S j , indicating that the S i -th sensing node exchanges the obtained relevant information with the S j -th sensing node;

[0023] S i → D i , indicating that the S i -th sensing node transmits the obtained relevant information to the D i -th decision node;

[0024] D i → D j , indicating that the D i -th decision node obtains from the D jReceive task instructions at each of the decision nodes to implement the decision node D i and the decision node D j for decision coordination between them;

[0025] D i →S i indicates that the D-th i decision node transmits the task instruction to the S-th i sensing node, causing the S-th i sensing node to adjust the detection task;

[0026] D i →W i indicates that the D-th i decision node transmits the task instruction to the W-th i execution node, causing the W-th i execution node to execute a specific task;

[0027] W i →T i indicates that the W-th i execution node affects the T-th i target node according to the task instruction.

[0028] In an exemplary embodiment of the present application, the detection layer is represented by G S the decision layer is represented by G D the execution layer is represented by G W and the task layer is represented by G T ;

[0029] All the association patterns between the detection layer and the decision layer are represented by E SD all the association patterns between the decision layer and the execution layer are represented by E DW all the association patterns between the execution layer and the task layer are represented by E WT and all the association patterns between the task layer and the detection layer are represented by F TS ;

[0030] The expression of the task network model is:

[0031]

[0032] where GA N represents the task network model, G TS represents the detection and sensing network pointing from the task layer to the detection layer; G SS represents the information sharing network between the detection layers; G SDRepresents an information transfer network pointing from the detection layer to the decision-making layer; G DS Represents an information feedback network pointing from the decision-making layer to the detection layer; G DD Represents an information fusion network between decision-making layers; G DW Represents a command network pointing from the decision-making layer to the execution layer; G WT Represents a task assignment network pointing from the execution layer to the task layer.

[0033] In an exemplary embodiment of the present application, the node failure strategy includes a random failure strategy, a maximum degree failure strategy, and a maximum betweenness failure strategy; the maximum degree failure strategy and the maximum betweenness failure strategy are collectively referred to as the deliberate failure strategy;

[0034] The random failure strategy is used to simulate the situation of node failure in the task network model due to its own internal faults and external environmental interferences; the deliberate failure strategy is used to characterize the overall performance change of the task network model when affected by various deliberate failure factors;

[0035] The node recovery strategy includes a random recovery strategy, a maximum degree recovery strategy, and a maximum betweenness recovery strategy;

[0036] The random recovery strategy means that at any moment, a failure node set is constructed according to the failure information of all the nodes stored in the task network model. The failure node set includes multiple failure nodes, and random numbers with an exponential distribution are generated for all the failure nodes; if the random number corresponding to the failure node is less than a preset simulation parameter, it is considered that the failure node completes recovery at this moment;

[0037] The maximum degree recovery strategy means that the degrees generated by all the failure nodes in the failure node set are sorted in descending order of nodes, and the failure node corresponding to the maximum degree is selected for recovery; if the failure node is successfully recovered, other failure nodes are continued to be recovered according to this recovery step; if the failure node is not successfully recovered, the recovery process at this moment ends;

[0038] The maximum betweenness recovery strategy means that the betweennesses generated by all the failure nodes in the failure node set are sorted in descending order of nodes, and the failure node corresponding to the maximum betweenness is selected for recovery; if the failure node is successfully recovered, other failure nodes are continued to be recovered according to this recovery step; if the failure node is not successfully recovered, the recovery process at this moment ends.

[0039] In an exemplary embodiment of the present application, the steps of constructing the random failure strategy include:

[0040] Using the exponential distribution sampling method to probabilistically characterize the failure situation of all the nodes within a certain period of time;

[0041] Among them, the expression of the probability characterization is:

[0042]

[0043] Among them, F(t; λ) represents the probability that a node operates normally from the initial moment to the failure moment, t represents the t-th moment within the task cycle, and λ represents the failure rate of the node in the task network model;

[0044] Within a certain period of time, all the nodes operating normally generate first random numbers within [0, 1], and the first random numbers are represented by random 1 to represent;

[0045] Compare all the first random numbers with F(t; λ) respectively. If random 1 <F(t; λ), then the node is the failure node, and extract the relevant information of all the associated patterns of each failure node from the task network model respectively, to be used after the node recovers;

[0046] The steps for constructing the maximum degree failure strategy include:

[0047] Use the exponential distribution sampling method to perform probability characterization on the failure situations of all the nodes within a certain period of time;

[0048] Within a certain period of time, all the nodes operating normally generate second random numbers within [0, 1], and the second random numbers are represented by random 2 to represent;

[0049] Compare all the second random numbers with F(t; λ) respectively. If random 2 <F(t; λ), then the node is the failure node, and extract the relevant information of all the associated patterns of each failure node from the task network model respectively, to be used after the node recovers;

[0050] Select the node corresponding to the maximum degree among the remaining nodes, and iteratively execute the above two steps until the generated second random number is greater than or equal to F(t; λ), and the failure process of the node ends;

[0051] Among them, the degree represents the number of links between a certain node and all other nodes in the task network model;

[0052] The steps for constructing the maximum betweenness failure strategy include:

[0053] Use the exponential distribution sampling method to probabilistically characterize the failure conditions of all the nodes within a certain period of time;

[0054] Within a certain period of time, all the nodes operating normally generate a third random number within [0, 1], and the third random number is represented by random 3 to represent;

[0055] Compare all the third random numbers with F(t; λ) respectively. If random 3 <F(t; λ), then the node is a failed node, and extract the relevant information of all the associated patterns of each failed node from the task network model respectively, to be used after the node recovers;

[0056] Select the node corresponding to the largest betweenness centrality among the remaining nodes, and iteratively execute the above two steps until the generated third random number is greater than or equal to F(t; λ), at which point the node failure process ends;

[0057] Among them, the betweenness centrality represents the proportion of the number of paths passing through the node in all the shortest paths in the task network to the total number of shortest paths.

[0058] In an exemplary embodiment of the present application, the step of designing a node failure strategy and a node recovery strategy according to the topological network structure includes:

[0059] Determine the initial states of all the nodes in the task network model, initialize the network parameters of the task network model, and set the seed of the random number generator according to the design strategy;

[0060] At any moment, for all the nodes of the task network model, generate the node failure sequence and the node recovery sequence of all the nodes respectively according to the probability characterization of the design strategy;

[0061] Select all the failed nodes and all the recovered nodes, update the states of all the failed nodes and all the recovered nodes according to the probability characterization, and record the relevant information of all the failed nodes and all the recovered nodes.

[0062] In an exemplary embodiment of the present application, the step of formulating an unmanned cluster performance function according to the node failure strategy and the node recovery strategy, and formulating an unmanned cluster resilience index according to the dynamic change of the unmanned cluster performance function includes:

[0063] Select appropriate node failure strategy and node recovery strategy, and design relevant simulation experiments;

[0064] In each simulation experiment cycle, calculate the initial task network global efficiency according to the initial states of all the nodes in the task network model;

[0065] Calculate the task network global efficiency according to the node failure strategy and the node recovery strategy at each moment in the simulation experiment cycle;

[0066] Sum up the task network global efficiencies at all moments in the simulation experiment cycle, divide the obtained total efficiency by the initial task network global efficiency, and then multiply by time to obtain the unmanned cluster resilience index.

[0067] In an exemplary embodiment of the present application, the expression of the task network global efficiency is:

[0068]

[0069] where NE on (t) represents the task network global efficiency, N represents the number of effective closed loops of all target nodes in the task network model; N r represents the number of target nodes in the task network model, t represents the t-th moment in the task cycle, t n represents the serial number parameter of the target node, represents the length of the shortest effective closed loop passing through the target node.

[0070] In an exemplary embodiment of the present application, the expression of the unmanned cluster resilience index is:

[0071]

[0072] where R represents the unmanned cluster resilience index, t 0 represents the initial moment, t f represents the moment of recovery to stability, P(t) represents the performance of the unmanned cluster at the t-th moment in the task cycle, P(t 0 ) represents the initial performance of the unmanned cluster.

[0073] Beneficial effects:

[0074] The present application provides an unmanned cluster resilience evaluation method based on task network efficiency, which has at least the following beneficial effects:

[0075] (1) The present application realizes a complete evaluation system by constructing a task network model, a node failure strategy, a node recovery strategy, a task network global efficiency, and an unmanned cluster resilience index, and can independently evaluate the resilience of an unmanned cluster;

[0076] (2) Through the complete and independent resilience evaluation system of this application, the dynamic evolution process of the tasks of the unmanned cluster can be clearly reflected, and it is not difficult to simulate using the random failure strategy and the deliberate failure strategy, which is convenient for solving.

[0077] (3) In terms of the acquisition and processing of data in the unmanned cluster resilience evaluation method proposed in this application, compared with the current resilience evaluation method based on the performance curve, the calculation cost is lower and the accuracy is higher. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] The accompanying drawings herein are incorporated into the specification and form a part of the specification, showing embodiments consistent with the present application, and are used together with the specification to explain the principles of the present application. Obviously, the accompanying drawings in the following description are only some embodiments of the present application, and those of ordinary skill in the art can obtain other drawings based on these drawings without creative efforts.

[0079] Figure 1 Schematic diagram showing the steps of a method for evaluating the resilience of an unmanned cluster based on task network efficiency in an exemplary embodiment of the present application;

[0080] Figure 2 Schematic diagram showing the structure of a task network model in an exemplary embodiment of the present application;

[0081] Figure 3 Schematic diagram showing the classification of the entire evaluation system for evaluating the resilience of an unmanned cluster proposed in an exemplary embodiment of the present application;

[0082] Figure 4 Schematic diagram of a curve showing the change in the resilience performance of a task network model based on a node failure strategy and a node recovery strategy in an exemplary embodiment of the present application;

[0083] Figure 5 Schematic diagram showing the comparison of different failure - recovery strategies in a simulation experiment of an exemplary embodiment of the present application;

[0084] Figure 6 Schematic diagram showing the comparison of different maximum - degree failure - recovery strategies in a simulation experiment of an exemplary embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0085] Example embodiments will now be described more fully with reference to the accompanying drawings. However, the example embodiments can be implemented in various forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this application will be more complete and comprehensive, and will fully convey the concept of the example embodiments to those skilled in the art. The features, structures, or characteristics described can be combined in any suitable manner in one or more embodiments.

[0086] In addition, the attached drawings are only schematic diagrams of the present application and are not necessarily drawn to scale. The same reference numerals in the drawings represent the same or similar parts, and thus repeated descriptions thereof will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor devices and / or microcontroller devices.

[0087] This exemplary embodiment provides a method for evaluating the resilience of an unmanned cluster based on task network efficiency, as Figure 1 shown, the evaluation method may include the following steps:

[0088] Step S101: Construct a task network model of the unmanned cluster, where the task network model includes a topological network structure successively composed of a detection layer, a decision layer, an execution layer, and a task layer; the topological network structure includes different types of nodes, all nodes are communicatively connected, and there are multiple association modes between all nodes;

[0089] Step S102: Design node failure strategies and node recovery strategies according to the topological network structure.

[0090] Step S103: Formulate performance indicators of the unmanned cluster according to the node failure strategies and node recovery strategies, and formulate resilience indicators of the unmanned cluster according to the dynamic changes of the unmanned cluster performance indicators.

[0091] Step S104: Use the resilience indicators of the unmanned cluster to evaluate the resilience of the unmanned cluster.

[0092] The embodiment of the present application proposes a method for evaluating the resilience of an unmanned cluster based on task network efficiency, which has at least the following beneficial effects:

[0093] (1) The present application realizes a complete evaluation system through constructing a task network model, node failure strategies, node recovery strategies, the global efficiency of the task network, and the resilience indicators of the unmanned cluster, and can independently evaluate the resilience of the unmanned cluster;

[0094] (2) Through the complete and independent resilience evaluation system, the present application can clearly reflect the task dynamic evolution process of the unmanned cluster, and it is not difficult to simulate using random failure strategies and deliberate failure strategies, which is convenient for solving.

[0095] (3) Compared with the current resilience evaluation method based on performance curves, the method for evaluating the resilience of the unmanned cluster proposed in the present application has a lower computational cost and higher accuracy in data acquisition and processing.

[0096] Next, a method for evaluating the resilience of an unmanned cluster based on the efficiency of a task network proposed in this exemplary embodiment will be described in more detail.

[0097] In step S101 of this embodiment, as Figure 2 shown, a task network model of the unmanned cluster is constructed. Based on the Observation-Orientation-Decision-Action (OODA) loop theory, that is, the OODA loop theory, the task network model includes a topological network composed of a detection layer, a decision-making layer, an execution layer, and a task layer. The more types of OODA loops that the task network model of the unmanned cluster can construct for a single target node, the more diverse task means the unmanned cluster has when affecting the target node, thereby reflecting that the unmanned cluster has a relatively strong task execution ability for a certain target node.

[0098] As a large-scale complex system, the relationships between each unit inside the unmanned cluster are complex and diverse. In this embodiment, each unit in the unmanned cluster is regarded as a different node, and various association modes of interconnection, interoperability, and interaction between units are regarded as the connections between nodes, thereby obtaining a multi-level, heterogeneous, and directed task network topological structure of the unmanned cluster.

[0099] The task network model includes a topological network structure successively composed of a detection layer, a decision-making layer, an execution layer, and a task layer. Among them, the topological network structure includes different types of nodes, all nodes are communicatively connected, and there are various association modes between all nodes.

[0100] Furthermore, in this embodiment, the types of nodes are specifically as follows:

[0101] The detection layer contains multiple sensing nodes, and the set of all sensing nodes is represented as: S = {S 1 , S 2 , …, S i , …, S j , …, S I}, S I represents the S I th sensing node, I represents the number of sensing nodes, and S represents the set of sensing nodes.

[0102] The decision-making layer contains multiple decision-making nodes, and the set of all decision-making nodes is represented as: D = {D 1 , D 2 , …, D i , …, D j , …, D J}, D J represents the D JThere are N decision nodes, J represents the number of decision nodes, and D represents the set of decision nodes.

[0103] The execution layer includes multiple execution nodes, and the set of all execution nodes is represented as: W = {W 1 , W 2 , …, W i , …, W j , …, W N}, where W N represents the W N -th execution node, N represents the number of execution nodes, and W represents the set of execution nodes.

[0104] The task layer includes multiple target nodes, and the set of all target nodes is represented as: T = {T 1 , T 2 , …, T i , …, T j , …, T M}, where T M represents the T M -th target node, M represents the number of target nodes, and T represents the set of target nodes.

[0105] Furthermore, in this embodiment, all association patterns include:

[0106] T i → S i , indicating that the S i -th sensing node detects the T i -th target node and obtains relevant information from the T i -th target node;

[0107] S i → S j , indicating that the S i -th sensing node exchanges the obtained relevant information with the S j -th sensing node;

[0108] S i → D i , indicating that the S i -th sensing node transmits the obtained relevant information to the D i -th decision node;

[0109] D i → D j , indicating that the D i -th decision node receives a task instruction from the D j -th decision node, realizing decision collaboration between decision node D i and decision node D j ;

[0110] D i →S i indicates that the D i th decision node transmits the task instruction to the S i th sensing node, enabling the S i th sensing node to adjust the detection task;

[0111] D i →W i indicates that the D i th decision node transmits the task instruction to the W i th execution node, enabling the W i th execution node to execute the specific task;

[0112] W i →T i indicates that the W i th execution node affects the T i th target node according to the task instruction.

[0113] The association patterns among all nodes in the unmanned cluster are diverse and can be adjusted according to the specific requirements of the task and the changes in the task environment to ensure the efficiency of information transmission and the reliability of task completion. For example, according to different demand situations, among all sensing nodes, all decision nodes, and all execution nodes, some nodes can use wireless communication to improve the mobility and flexibility of the unmanned cluster; another part of the nodes can rely on wired connections to obtain more stable and secure data transmission.

[0114] Furthermore, in this embodiment, considering that the multi-level coupled network of the unmanned cluster has a large number of heterogeneous nodes and various heterogeneous relationships, using the hypernetwork model can effectively and clearly represent the structural characteristics of the task network model of the unmanned cluster. According to the association patterns of different layers of the unmanned cluster network, there is no direct communication relationship between the target nodes in the task layer and the execution nodes in the execution layer of the unmanned cluster. Therefore, the corresponding intra-layer connection relationship is an empty set. Using hypernetwork theory, in this embodiment, high-level abstractions are made for the nodes and association patterns in each layer of the unmanned cluster network, and thus the following can be obtained:

[0115] The detection layer is represented by G S ; the decision layer is represented by G D ; the execution layer is represented by G W ; the task layer is represented by G T . All association patterns between the detection layer and the decision layer are represented by E SD ; all association patterns between the decision layer and the execution layer are represented by E DW ; all association patterns between the execution layer and the task layer are represented by EW T; all association patterns between the task layer and the detection layer are represented by ETS representation

[0116] Furthermore, the expression of the task network model is as follows:

[0117]

[0118] where, G AN represents the task network model, and G TS represents the detection and perception network pointing from the target layer to the detection layer; G SS represents the information sharing network between the detection layers; G SD represents the information transfer network pointing from the detection layer to the decision-making layer; G DS represents the information feedback network pointing from the decision-making layer to the detection layer; G DD represents the information fusion network between the decision-making layers; G DW represents the command network pointing from the decision-making layer to the execution layer; G WT represents the task assignment network pointing from the execution layer to the target layer.

[0119] In step S102 of this embodiment, as Figure 3 shown, according to the topological structure of the task network model, a node failure strategy and a node recovery strategy are designed. Step S102 of this embodiment may include the following sub-steps:

[0120] Sub-step S1021: Determine the initial states of all nodes in the task network model, initialize the network parameters of the task network model, and set the seed of the random number generator according to the design strategy.

[0121] Sub-step S1022: At any moment, for all nodes of the task network model, generate the node failure sequence and node recovery sequence of all nodes respectively according to the probability representation of the design strategy.

[0122] Sub-step S1023: Select all failed nodes and all recovered nodes, update the states of all failed nodes and all recovered nodes according to the probability representation, and record the relevant information of all failed nodes and all recovered nodes.

[0123] Furthermore, the node failure strategy includes a random failure strategy and a deliberate failure strategy, where the deliberate failure strategy includes a maximum degree failure strategy and a maximum betweenness failure strategy.

[0124] The random failure strategy can be used to simulate the situation where node failures in the task network model are caused by its own internal faults and external environmental interferences. The deliberate failure strategy can be used to characterize the overall performance change of the task network model when it is affected by various deliberate failure factors.

[0125] Furthermore, the steps for constructing the random failure strategy include:

[0126] First, use the exponential distribution sampling method to probabilistically characterize the failure situations of all nodes within a certain period of time;

[0127] Among them, the expression for probabilistic characterization is:

[0128]

[0129] Among them, F(t; λ) represents the probability that a node operates normally from the initial moment to the failure moment, t represents the t-th moment within the task cycle, and λ represents the failure rate of nodes in the task network model.

[0130] Secondly, within a certain period of time, all nodes operating normally generate a first random number within [0, 1], and the first random number is represented by random 1 to represent.

[0131] Finally, compare all the first random numbers with F(t; λ) respectively. If random 1 < F(t; λ), then the node is a failed node, and relevant information of all associated patterns of each failed node is extracted from the task network model respectively for use after the node recovers.

[0132] The steps for constructing the maximum degree failure strategy include:

[0133] First, use the exponential distribution sampling method to probabilistically characterize the failure situations of all nodes within a certain period of time;

[0134] Within a certain period of time, all nodes operating normally generate a second random number within [0, 1], and the second random number is represented by random 2 to represent;

[0135] Compare all the second random numbers with F(t; λ) respectively. If random 2 < F(t; λ), then the node is a failed node, and relevant information of all associated patterns of each failed node is extracted from the task network model respectively for use after the node recovers;

[0136] Select the node corresponding to the maximum degree among the remaining nodes, and iteratively execute the above two steps until the generated second random number is greater than or equal to F(t; λ), at which point the node failure process ends.

[0137] Here, the degree represents the number of connections of a certain node in the task network model to all other nodes.

[0138] Furthermore, the steps for constructing the maximum betweenness failure strategy include:

[0139] Use the exponential distribution sampling method to probabilistically characterize the failure situations of all nodes within a certain period of time;

[0140] Within a certain period of time, all normally operating nodes generate a third random number within [0, 1], and the third random number is represented by random 3 to indicate;

[0141] Compare all the third random numbers with F(t; λ) respectively. If random 3 < F(t; λ), then the node is a failed node, and extract the relevant information of all association patterns of each failed node from the task network model respectively, to be used after the node recovers;

[0142] Select the node corresponding to the largest betweenness in the remaining nodes, and iteratively execute the above two steps until the generated third random number is greater than or equal to F(t; λ), and the node failure process at this moment ends;

[0143] Among them, betweenness represents the proportion of the number of paths passing through a node in all the shortest paths in the task network model to the total number of the shortest paths.

[0144] In this embodiment, betweenness refers to node betweenness, which represents the proportion of the number of paths passing through a certain specific node in all the shortest paths in the task network model to the total number of the shortest paths. The maximum betweenness failure strategy is similar to the maximum degree failure strategy.

[0145] Furthermore, the constructed node recovery strategy is a specific manifestation of the adaptation and recovery capabilities of the task network model of the unmanned cluster. In this embodiment, the node recovery strategy mainly includes the random recovery strategy, the maximum degree recovery strategy, and the maximum betweenness recovery strategy.

[0146] The random recovery strategy means that at any moment, construct a set of failed nodes according to the failure information of all nodes stored in the task network model. The set of failed nodes includes multiple failed nodes, and generate exponentially distributed random numbers for all the failed nodes correspondingly; if the random number corresponding to a failed node is less than the preset simulation parameter, it is considered that the failed node has completed recovery at this moment.

[0147] The maximum degree recovery strategy means that sort the degrees generated by all the failed nodes in the set of failed nodes in descending order, and select the failed node corresponding to the largest degree for recovery; if the failed node is successfully recovered, continue to recover other failed nodes according to this recovery step; if the failed node is not successfully recovered, the recovery process at this moment ends.

[0148] The maximum betweenness recovery strategy means: sort the nodes in descending order according to the betweenness generated by all failed nodes in the failed node set, and select the failed node corresponding to the maximum betweenness for recovery; if the failed node is successfully recovered, continue to recover other failed nodes according to this recovery step; if the failed node is not successfully recovered, the recovery process at this moment ends.

[0149] In step S103 of this embodiment, as Figure 4 shown, according to the node failure strategy and the node recovery strategy, formulate the performance index of the unmanned cluster, and formulate the resilience index of the unmanned cluster according to the dynamic change of the performance index of the unmanned cluster. Step S103 of this embodiment may include the following sub-steps:

[0150] Sub-step S1031: Select the appropriate node failure strategy and the node recovery strategy, and design relevant simulation experiments.

[0151] Sub-step S1032: In each simulation experiment cycle, calculate the initial global efficiency of the task network according to the initial states of all nodes in the task network model.

[0152] Sub-step S1033: Calculate the global efficiency of the task network according to the node failure strategy and the node recovery strategy at each moment in the simulation experiment cycle.

[0153] Furthermore, the expression of the global efficiency of the task network is:

[0154]

[0155] where NE on (t) represents the global efficiency of the task network, N represents the number of effective closed loops of all target nodes in the task network model; N r represents the number of target nodes in the task network model, t represents the t-th moment in the task cycle, t n represents the serial number parameter of the target node, represents the length of the shortest effective closed loop passing through the target node.

[0156] In the task network model of the unmanned cluster, a valid closed-loop path symbolizes the entire cycle from task startup to target completion. The global efficiency of the task network is a measure of the ability of the task network model of the unmanned cluster to achieve independent closed loops for each target node at the macroscopic level, and it maps the overall effectiveness of the task network model of the unmanned cluster and the efficiency of the overall collaborative task.

[0157] Sub-step S1034: Perform a summation operation on the global efficiency of the task network at all moments in the simulation experiment cycle, divide the obtained total efficiency by the initial global efficiency of the task network, and then multiply by time, that is, the resilience index of the unmanned cluster is obtained.

[0158] Taking the global efficiency of the task network as the performance index of the unmanned cluster, the resilience performance change of the task network model considering the node failure strategy and the node recovery strategy is studied. From Figure 4 As can be seen, the abscissa is time and the ordinate is performance. P(t) represents the global efficiency of the task network of the task network model at the t-th moment. Starting from time t 0 , the nodes in the task network model gradually start to fail according to the node failure strategy. At time t d , the performance of the task network model begins to decline. The unmanned cluster autonomously selects a recovery model and starts node recovery. At time t m , the performance of the task network model drops to the lowest point. After that, its performance gradually recovers and enters a stable state at time t f .

[0159] Furthermore, the expression of the resilience index of the unmanned cluster is:

[0160]

[0161] where R represents the resilience index of the unmanned cluster, t 0 represents the initial time, t f represents the time when the performance of the unmanned cluster recovers and stabilizes after improvement, P(t) represents the performance of the unmanned cluster at the t-th moment during the task cycle, and P(t 0 ) represents the initial performance of the unmanned cluster.

[0162] In step S104, the resilience of the unmanned cluster is evaluated using the resilience index of the unmanned cluster.

[0163] To verify the effectiveness of a method for evaluating the resilience of an unmanned cluster based on task network efficiency proposed in this application, the following simulation experiments were carried out.

[0164] In this simulation experiment, a task network of an unmanned cluster containing 20 target nodes, 45 sensing nodes, 10 decision-making nodes, and 30 execution nodes, a total of 105 nodes, was established for simulation verification.

[0165] As Figure 5 shown, Ar-Rr represents the random failure-random recovery strategy, Ar-Rd represents the random failure-maximum degree recovery strategy, and Ar-Rb represents the random failure-maximum betweenness recovery strategy.

[0166] From Figure 5 it can be seen that in the case of random failure, when the random failure-maximum degree recovery strategy is adopted, the global efficiency of the task network decreases the least, recovers relatively quickly, and the global efficiency of the task network when finally recovering to the new stable stage is also the highest.

[0167] When the random failure - maximum betweenness recovery strategy is adopted, the degree of reduction in the global efficiency of the task network is the same as that when the random failure - maximum degree recovery strategy is adopted. However, when the random failure - maximum betweenness recovery strategy is adopted, after a period of rapid recovery, the recovery speed decreases, and the global efficiency of the task network when it finally returns to the new stable stage is the lowest.

[0168] When the random failure - random recovery strategy is adopted, the global efficiency of the task network decreases the most, but the recovery rate is the highest, and the global efficiency of the task network when it finally returns to the new stable stage ranks second.

[0169] According to the expression of the resilience index of the unmanned cluster, these three strategies are calculated respectively, and we get: R Ar-Rr = 0.886, R Ar-Rd = 0.933 and R Ar-Rb = 0.912. This shows that under the random failure scenario, the performance degradation of the unmanned cluster is the lowest and the recovery is the best after adopting the random failure - maximum degree recovery strategy, and the resilience value of the resilience index of the unmanned cluster is the highest.

[0170] As Figure 6 shown, use Ad - Rr to represent the maximum degree failure - random recovery strategy, Ad - Rd to represent the maximum degree failure - maximum degree recovery strategy, and Ad - Rb to represent the maximum degree failure - maximum betweenness recovery strategy.

[0171] From Figure 6 it can be seen that in the case of maximum degree failure, when the maximum degree failure - random recovery strategy is adopted, the global efficiency of the task network decreases the least, and the global efficiency of the task network when it finally returns to the new stable stage ranks second.

[0172] When the maximum degree failure - maximum betweenness recovery strategy is adopted, the degree of reduction in the global efficiency of the task network is greater than that when the maximum degree failure - random recovery strategy is adopted, and the global efficiency of the task network when it finally returns to the new stable stage is the lowest.

[0173] When the maximum degree failure - maximum degree recovery strategy is adopted, the degree of reduction in the global efficiency of the task network is the same as that when the maximum degree failure - maximum betweenness recovery strategy is adopted, and the global efficiency of the task network when it finally returns to the new stable stage is the highest.

[0174] According to the expression of the resilience index of the unmanned cluster, these three modes are calculated respectively, and we get: R Ad-Rr = 0.934, R Ad-Rd = 0.933 and R Ad-Rb= 0.932. The results show that in the scenario of maximum-degree failure, the failed nodes are all nodes with relatively high degrees in the task network model. With the same parameters, the number of nodes adopting the maximum-degree failure-random recovery strategy is larger, and the degree of performance degradation is the least. Considering that the resilience indicators calculated by the three strategies differ very little, and considering that the performance recovery rate of the unmanned cluster is the highest when adopting the maximum-degree failure-maximum-degree recovery strategy, it is expected that in a wider time range, the maximum-degree failure-maximum-degree recovery strategy will be superior to the maximum-degree failure-random recovery strategy.

[0175] It can be seen from the simulation experiments that when adopting the random failure-maximum-degree recovery strategy and the maximum-degree failure-maximum-degree recovery strategy, not only can the performance of the unmanned cluster be effectively restored, but also the resilience of the task network model can be significantly improved, thus also reflecting the

[0176] In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the embodiments of the present application, the meaning of "a plurality" is two or more, unless otherwise specifically defined.

[0177] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example" or "some examples", etc. means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. In addition, those skilled in the art can combine and combine the different embodiments or examples described in this specification.

[0178] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of various equivalent modifications or substitutions within the technical scope disclosed by the present application, and these modifications or substitutions should all be covered within the protection scope of the present application.

[0179] Those skilled in the art will easily think of other implementation schemes of the present application after considering the specification and practicing the invention disclosed herein. The present application is intended to cover any variations, uses or adaptations of the present application, and these variations, uses or adaptations follow the general principles of the present application and include the common general knowledge or conventional technical means in the technical field not disclosed by the present application.

Claims

1. A method for evaluating the resilience of unmanned clusters based on task network efficiency, characterized in that: The following steps are involved: Constructing a task network model of an unmanned cluster, wherein the task network model includes a topological network structure consisting of a detection layer, a decision layer, an execution layer, and a task layer in sequence, wherein the topological network structure includes different types of nodes, all of which are communicatively connected, and all of which have multiple association modes; Design node failure strategy and node recovery strategy according to the topological network structure; Formulate unmanned cluster performance indicators according to the node failure strategy and the node recovery strategy, and formulate unmanned cluster resilience indicators according to dynamic changes of the unmanned cluster performance indicators; The unmanned cluster resilience index is used to evaluate the resilience of the unmanned cluster.

2. The unmanned cluster resilience assessment method based on task network efficiency according to claim 1 is characterized in that: The types of nodes include: the detection layer includes multiple perception nodes, the decision layer includes multiple decision nodes, the execution layer includes multiple execution nodes, and the task layer includes multiple target nodes; The set of all the sensing nodes is represented as: S = {S1, S2, ..., S i ,…,S j ,…,S I },S I Indicates the S I sensor nodes, I represents the number of sensor nodes; The set of all decision nodes is represented as: D = {D1, D2, ..., D i ,…,D j ,…,D J }, D J Indicates D J decision nodes, J represents the number of decision nodes; The set of all the execution nodes is represented as: W = {W1, W2, ..., W i ,…,W j ,…,W N },W N Indicates the W N execution nodes, N represents the number of execution nodes; The set of all the target nodes is represented as: T = {T1, T2, ..., T i ,…,T j ,…,T M }, T M Indicates the T M target nodes, and M represents the number of target nodes.

3. The unmanned cluster resilience assessment method based on task network efficiency according to claim 2 is characterized in that: The various association modes include: T i →S i , indicating the S i The sensing nodes detect the Tth i the target node, and from the Tth i Obtain relevant information from the target node; S i →S j , indicating the S i The sensing node obtains the relevant information and j The sensing nodes are exchanged; S i →D i , indicating the S i The sensing node transmits the acquired relevant information to the Dth i the decision nodes; D i →D j , indicating that the D i The decision nodes are from the D j The decision node receives the task instruction and implements the decision node D i and the decision node D j Decision-making coordination between D i →S i , indicating that the D i The decision node transmits the task instruction to the Sth i The sensing nodes make the Sth i The sensing nodes adjust the detection tasks; D i →W i , indicating that the D i The decision node transmits the task instruction to the Wth i The execution node W i The execution nodes execute specific tasks; W i →T i , indicating the Wth i The execution nodes affect the Tth i the target node.

4. The unmanned cluster resilience assessment method based on task network efficiency according to claim 3 is characterized in that: The detection layer is G S Indicates that the decision layer is represented by G D Indicated, the execution layer is represented by G W Indicates that the task layer is represented by G T express; All the association patterns between the detection layer and the decision layer are represented by E SD Indicates that all the association patterns between the decision layer and the execution layer are represented by E DW Indicates that all the association modes between the execution layer and the task layer are represented by E WT Indicates that all the association patterns between the task layer and the detection layer are represented by E TS express; The expression of the task network model is: Among them, G AN represents the task network model, G TS G represents the detection perception network from the task layer to the detection layer; SS represents the information sharing network between detection layers; G SD represents the information transmission network from the detection layer to the decision layer; G DS represents the information feedback network from the decision layer to the detection layer; G DD represents the information fusion network between decision layers; G DW G represents the command network from the decision-making level to the execution level; WT Represents the task allocation network from the execution layer to the task layer.

5. The unmanned cluster resilience assessment method based on task network efficiency according to claim 1 is characterized in that: The node failure strategy includes a random failure strategy, a maximum degree failure strategy and a maximum betweenness failure strategy; the maximum degree failure strategy and the maximum betweenness failure strategy are collectively referred to as intentional failure strategies; The random failure strategy is used to simulate the situation where the node of the task network model fails due to its own internal failure and external environmental interference; The intentional failure strategy is used to characterize the overall performance change of the task network model when it is affected by various intentional failure factors; The node recovery strategy includes a random recovery strategy, a maximum degree recovery strategy and a maximum betweenness recovery strategy; The random recovery strategy means that at any time, a failed node set is constructed according to the failure information of all the nodes stored in the task network model, the failed node set includes multiple failed nodes, and exponentially distributed random numbers are generated for all the failed nodes; if the random number corresponding to the failed node is less than the preset simulation parameter, it is considered that the failed node has completed recovery at this time; The maximum degree recovery strategy means sorting the degrees generated by all the failed nodes in the failed node set in descending order, and selecting the failed node corresponding to the maximum degree for recovery; If the failed node is successfully restored, continue to restore other failed nodes according to the restoration steps; If the failed node is not successfully restored, the recovery process at that moment ends; The maximum betweenness recovery strategy means sorting the betweennesses generated by all the failed nodes in the failed node set in descending order, and selecting the failed node corresponding to the maximum betweenness for recovery; If the failed node is successfully restored, continue to restore other failed nodes according to the restoration steps; If the failed node is not successfully restored, the recovery process at that moment ends.

6. The unmanned cluster resilience assessment method based on task network efficiency according to claim 5 is characterized in that: The steps of constructing the random failure strategy include: Using exponential distribution sampling method to perform probability characterization on the failure conditions of all the nodes within a certain period of time; Among them, the expression of the probability representation is: Among them, F(t; λ) represents the probability that a node operates normally from the initial moment to the failure moment, t represents the t-th moment within the task cycle, and λ represents the failure rate of nodes in the task network model; Within a certain period of time, all the nodes operating normally generate first random numbers within [0, 1], and the first random numbers are represented by random1; Compare all the first random numbers with F(t; λ) respectively. If random1 < F(t; λ), then the node is the failure node, and extract the relevant information of all the associated patterns of each failure node from the task network model respectively, to be used after the node recovers; The steps of constructing the maximum degree failure strategy include: Use the exponential distribution sampling method to probabilistically characterize the failure situations of all the nodes within a certain period of time; Within a certain period of time, all the nodes operating normally generate second random numbers within [0, 1], and the second random numbers are represented by random2; Compare all the second random numbers with F(t; λ) respectively. If random2 < F(t; λ), then the node is the failure node, and extract the relevant information of all the associated patterns of each failure node from the task network model respectively, to be used after the node recovers; Select the node corresponding to the maximum degree among the remaining nodes, and iteratively execute the above two steps until the generated second random number is greater than or equal to F(t; λ), at which time the node failure process ends; Among them, the degree represents the number of connections between a certain node in the task network model and all other nodes; The steps of constructing the maximum betweenness failure strategy include: Use the exponential distribution sampling method to probabilistically characterize the failure situations of all the nodes within a certain period of time; Within a certain period of time, all the nodes operating normally generate third random numbers within [0, 1], and the third random numbers are represented by random3; Compare all the third random numbers with F(t; λ) respectively. If random3 < F(t; λ), then the node is a failure node, and extract the relevant information of all the associated patterns of each failure node from the task network model respectively, to be used after the node recovers; Select the node corresponding to the maximum betweenness among the remaining nodes, and iteratively execute the above two steps until the generated third random number is greater than or equal to F(t; λ), at which time the node failure process at this moment ends; Among them, the betweenness represents the proportion of the number of shortest paths passing through the node in the task network model to the total number of shortest paths; 7. The unmanned cluster resilience assessment method based on task network efficiency according to claim 6 is characterized in that: The steps of designing the node failure strategy and the node recovery strategy according to the topological network structure include: Determine the initial states of all the nodes in the task network model, initialize the network parameters of the task network model, and set the seed of the random number generator according to the design strategy; At any time, for all nodes of the task network model, node failure sequences and node recovery sequences of all nodes are generated respectively according to the probability representation of the design strategy; All the failed nodes and all the restored nodes are selected, and the states of all the failed nodes and all the restored nodes are updated according to the probability representation, and relevant information of all the failed nodes and all the restored nodes is recorded.

8. The unmanned cluster resilience assessment method based on task network efficiency according to claim 7 is characterized in that: The step of formulating an unmanned cluster performance function according to the node failure strategy and the node recovery strategy, and formulating an unmanned cluster resilience index according to the dynamic change of the unmanned cluster performance function comprises: Select appropriate node failure strategies and node recovery strategies, and design relevant simulation experiments; In each simulation experiment cycle, the global efficiency of the initial task network is calculated according to the initial states of all the nodes in the task network model; Calculating the global efficiency of the task network according to the node failure strategy and the node recovery strategy at each moment in the simulation experiment cycle; The global efficiency of the task network at all times in the simulation experiment cycle is summed up, and the obtained total efficiency is divided by the initial task network global efficiency, and then multiplied by time to obtain the unmanned cluster resilience index.

9. The unmanned cluster resilience assessment method based on task network efficiency according to claim 8 is characterized in that: The expression of the global efficiency of the task network is: Among them, NE on (t) represents the global efficiency of the task network, N represents the number of effective closed loops of all target nodes in the task network model; N r represents the number of target nodes in the task network model, t represents the tth time in the task cycle, and t n The serial number parameter representing the target node. Indicates the length of the shortest valid closed loop passing through the target node.

10. The unmanned cluster resilience assessment method based on task network efficiency according to claim 9 is characterized in that: The expression of the unmanned cluster resilience index is: Among them, R represents the unmanned cluster resilience index, t0 represents the initial time, and t f It represents the moment when the unmanned cluster recovers to stability after performance improvement, P(t) represents the performance of the unmanned cluster at the tth moment in the mission cycle, and P(t0) represents the initial performance of the unmanned cluster.

Images