A SysML-based multi-agent collaborative system probability analysis method, electronic equipment, storage medium and program product

Abstract

The application discloses a multi-agent collaborative system probability analysis method, an electronic device, a storage medium and a program product, and comprises the following steps: modeling and describing the functions, behaviors and interaction interfaces between agents in the multi-agent collaborative system to obtain a model of the multi-agent collaborative system; converting structural elements and behavior elements in the model into symbolic state variables, symbolic states and state transition relationships based on a predefined mapping rule to obtain a symbolic state transition model for describing multi-agent collaborative behaviors; introducing a probability parameter into the state transition relationship based on the symbolic state transition model to characterize uncertain behaviors in a multi-agent collaborative process and constructing a probabilistic state transition model; and outputting corresponding probability analysis results based on the probabilistic state transition model.

Description

Technical Field

[0001] This invention belongs to the field of multi-agent system modeling and analysis technology, specifically a method based on... A probabilistic analysis method for multi-agent cooperative systems. Technical Background

[0002] With the continuous improvement of computer hardware performance and the ongoing expansion of system scale, the structure and function of software systems and complex engineering systems are becoming increasingly complex. Their reliability and security issues are gradually becoming significant factors restricting system applications. Multi-agent collaborative systems are widely used, especially in scenarios such as distributed control, collaborative decision-making, and complex task execution. These systems typically consist of multiple agents with autonomous decision-making capabilities. These agents work together to complete tasks through information exchange and collaborative mechanisms, exhibiting characteristics such as concurrency, dynamism, and uncertainty. Defects in system design or collaborative logic can lead to collaborative failures and even serious consequences.

[0003] In the engineering design of multi-agent cooperative systems, system modeling languages ​​are typically used to describe the system structure and behavior. For example, using... Modeling the functional modules, behavioral states, and interactions between agents allows for a more intuitive depiction of the system's hierarchical structure and collaborative relationships. Taking a typical collaborative task as an example, multiple agents are in various behavioral states such as "standby," "execution," and "completion," and these states switch when specific triggering conditions are met. Modeling can clearly describe the state sets of each agent and the transition relationships between states, and demonstrate the overall collaborative process of the system from an engineering perspective.

[0004] However, the aforementioned engineering modeling models primarily focus on describing the system's structure and behavior. Their states and transitions are typically expressed graphically or textually, lacking unified formal semantic support and making them difficult to directly analyze the system's behavioral evolution under different execution paths. For example, when multiple agents are simultaneously in different state combinations, the engineering model struggles to accurately depict the system's current overall state and is not convenient for analyzing the state change paths the system may experience under different triggering conditions.

[0005] To compensate for the shortcomings of engineering modeling in behavioral analysis, existing technologies also employ state transition models for formal modeling of system behavior. These methods typically abstract the system state into a set of state variables X, where each state variable x describes a specific behavioral attribute of the system or agent; a specific state S of the system is represented as a single combination of values ​​from the set of state variables X. Based on this, the behavioral changes or events of the agent are abstracted into a set of state transition relations T, used to describe the process by which the system transitions from one state to another when specific conditions are met.

[0006] For example, in a simple collaborative scenario, state variables can be defined. ,in This indicates whether a certain intelligent agent is in an executing state. This indicates whether communication is normal. At this point, a system state can be represented as ( =Execution, =Normal). When a certain condition is met. At that time, the system may be in a state Transition to state When another condition is met When that happens, the state transitions to the next state. In this way, the behavior of the system can be formally described, and the evolution of behavior under different combinations of states can be analyzed.

[0007] Although the formal modeling methods mentioned above can characterize system behavior to some extent, their modeling process is usually independent of the engineering model and difficult to integrate with... While these models directly correspond to engineering modeling languages, they often require manual conversion of engineering models into formal models, leading to a fragmented modeling process, a large workload, and a high risk of errors. Furthermore, these models typically assume deterministic state transitions, making it difficult to reflect the uncertainties prevalent in real-world systems.

[0008] In practical applications, multi-agent cooperative processes are often affected by factors such as communication delays, environmental changes, and uncertainties in agent behavior, resulting in non-deterministic system behavior. For example, in the same cooperative task, from state... to state The transition may occur successfully with a high probability, or it may fail to occur due to communication failure or resource constraints. Although some existing probabilistic analysis methods can assign probabilities to state transitions, they are usually modeled in independent probabilistic models, failing to form a unified process with engineering modeling models or formal state models. This makes it difficult to intuitively reflect the uncertainty and reliability of the system's cooperative behavior from an engineering design perspective.

[0009] Therefore, existing technologies generally suffer from the following shortcomings: there is a lack of a unified mapping mechanism between the engineering model and the formal state transition model; the formal model and the probabilistic analysis process are independent of each other, making it difficult to uniformly describe and analyze the states, transitions, and uncertainties of multi-agent cooperative behavior during the engineering modeling stage. This fragmented modeling and analysis approach is detrimental to the engineering design, behavioral analysis, and collaborative effect evaluation of multi-agent cooperative systems. Summary of the Invention

[0010] Purpose of the invention: To address the problems in existing technologies regarding the disconnect between engineering modeling and behavioral analysis in multi-agent cooperative systems, the difficulty in correlating formal modeling with engineering models, and the challenge of uniformly describing and evaluating the uncertainties of the cooperative process, this invention proposes a method based on... The probabilistic analysis method for multi-agent cooperative systems establishes a mapping relationship between engineering modeling models and formal state transition models, and introduces a probabilistic modeling mechanism on this basis to achieve unified modeling and probabilistic analysis of multi-agent cooperative behavior. This supports the analysis and evaluation of the evolution process and credibility of the cooperative behavior of multi-agent cooperative systems, and improves the engineering applicability of multi-agent cooperative system modeling and analysis.

[0011] Technical solution: In the first aspect, the present invention proposes a method based on Probabilistic analysis methods for multi-agent cooperative systems include:

[0012] Step 1: Assume the multi-agent cooperative system consists of multiple agents, and utilize... The functions, behaviors, and interaction interfaces of each agent in the multi-agent cooperative system are modeled and described to obtain the multi-agent cooperative system. The model; the behavior of an intelligent agent includes: behavioral state, events that trigger state changes, state triggering conditions, and cooperative constraints;

[0013] Step 2: Based on predefined mapping rules, the... The structural and behavioral elements in the model are converted into symbolic state variables, symbolic states, and state transition relationships to obtain a symbolic state transition model for describing the cooperative behavior of multiple agents. The symbolic state variables include the current state of each agent, and the symbolic state is represented by a combination of values ​​of the symbolic state variables. The state transition relationships include task initiation transition, single agent completion transition, cooperative success transition, and failure transition.

[0014] Step 3: Based on the symbolic state transition model, introduce probability parameters into the state transition relationship to characterize the uncertain behavior in the multi-agent cooperative process, and construct the probabilistic state transition model.

[0015] Step 4: Based on the probabilistic state transition model, output the corresponding probability analysis results, which include the success probability of the multi-agent collaborative task, the overall credibility index of the multi-agent collaborative system, or the probability of occurrence of key states.

[0016] Furthermore, in step 2, the predefined mapping rules include:

[0017] Will In the model, the agent's behavioral state is mapped to symbolic state variables, which are used to characterize the agent's current behavioral state;

[0018] Will The state change relationships between agent behaviors in the model are mapped to state transition relationships, and state triggering conditions and cooperative constraints are mapped to state transition triggering conditions or constraint conditions.

[0019] The interaction between multiple agents is represented by sharing state variables or synchronizing state transitions.

[0020] Furthermore, the probabilistic state transition model is expressed as follows: ,in, For a set of symbolic state variables, For the system state set, ; transition to any state After introducing probability parameters, it is expanded to , where p represents the transition probability, which represents the probability of transitioning from state s to state s′ under the condition of satisfying the cooperative constraint φ.

[0021] Furthermore, the probability parameters are configured to correspond one-to-one with specific state transition relationships.

[0022] Furthermore, the state transition probability associated with the action of a single agent is determined by the success probability of the agent's execution.

[0023] Furthermore, when the system state transition is triggered by the independent actions of multiple agents, the transition probability is determined by the action probabilities of each agent according to a predefined combination rule.

[0024] Furthermore, when any agent enters a failure state, the transition probability is determined by the corresponding agent's failure probability or its complement probability.

[0025] Secondly, the present invention provides an electronic device, the electronic device comprising:

[0026] At least one processor;

[0027] and a memory communicatively connected to the at least one processor;

[0028] The memory stores a computer program executable by the at least one processor, which is then executed by the at least one processor to enable the at least one processor to perform operations based on... A probabilistic analysis method for multi-agent cooperative systems.

[0029] Thirdly, the present invention provides a computer-readable storage medium storing computer instructions that are used to cause a processor to execute and implement based on A probabilistic analysis method for multi-agent cooperative systems.

[0030] Part Four: This invention proposes a computer program product, which includes a computer program that, when executed by a processor, implements a system based on... A probabilistic analysis method for multi-agent cooperative systems.

[0031] Beneficial effects: The method of this invention first utilizes The structure, behavior, and cooperative relationships of multi-agent systems are modeled; subsequently, predefined mapping rules are used to... The model is converted into a symbolic state transition model to formally express the cooperative behavior of multiple agents. Based on this, probability parameters are introduced to extend the state transition model, constructing a probabilistic state transition model to describe the uncertain behavior in the multi-agent cooperative process. Finally, based on the probabilistic state transition model, the success probability of the multi-agent cooperative task and the system reliability are evaluated and analyzed. This invention realizes an integrated modeling process of engineering modeling and probabilistic analysis, reducing the complexity of modeling and analyzing multi-agent cooperative systems and exhibiting good engineering applicability. It has the following advantages:

[0032] (1) Achieving a unified connection between engineering modeling and formal modeling: This invention achieves this by... The structural elements, behavioral states, and interaction relationships in the engineering model are mapped to state variables and state transition relationships in the symbolic transition system. This achieves a consistent expression between the engineering model and the formal behavioral model, avoids the problem of the separation between the engineering model and the formal model in the existing technology, and reduces the manual conversion cost of model construction and maintenance.

[0033] (2) Enhance the formal expression of multi-agent cooperative behavior: This invention describes the behavioral state and cooperative conditions of multi-agent systems by introducing symbolic state variables. The system state is represented by the values ​​of multiple state variables or the combination of constraints, which can effectively characterize the concurrent execution of multiple agents and the combination of multiple states, avoid explicit enumeration of complex state spaces, and improve the expressive power and scalability of complex cooperative behavior modeling.

[0034] (3) Improve the completeness and coverage of collaborative behavior analysis: This invention describes the state transition conditions in the form of constraint predicates in the symbolic transition system, so that the state evolution path of the system under different combinations of conditions can be systematically analyzed, ensuring that the collaborative behavior of multi-agents can be covered and analyzed under various execution paths, which helps to discover potential collaborative conflicts and design defects.

[0035] (4) Unified characterization of uncertainties in multi-agent collaboration: Based on the symbolic transition model, this invention introduces probability parameters to construct a probabilistic symbolic transition system, so that the probability of state transition can be uniformly described with engineering modeling and behavioral model, thereby reflecting the impact of uncertainties such as communication instability and resource constraints on collaborative behavior under the same model framework.

[0036] (5) Improve the accuracy and interpretability of the credibility analysis of collaborative tasks: By analyzing the probabilistic transition model, this invention can calculate the success probability and the probability of occurrence of key states in multi-agent collaborative tasks, providing quantitative indicators for the credibility assessment of collaborative systems, making the analysis results more intuitive and interpretable, and easier for engineers to understand and apply.

[0037] (6) Reduce the modeling complexity of complex multi-agent system analysis: by... Introducing directly mappable behavioral semantics during the engineering modeling phase reduces repetitive modeling of system behavior in subsequent formal modeling and probabilistic analysis, improves modeling efficiency, and reduces the overall complexity of complex system analysis.

[0038] (7) Improve the engineering applicability and promotion value of the method: The modeling and analysis process of the present invention has a clear structure, which is not only applicable to the engineering design of multi-agent collaborative systems, but also easy to use in combination with existing system modeling and analysis tools. It can support the behavior analysis and credibility assessment of complex collaborative systems in the design stage, and has good engineering applicability and promotion value. Attached Figure Description

[0039] Figure 1 Based on A flowchart illustrating the probabilistic analysis method for multi-agent cooperative systems;

[0040] Figure 2 This is a symbolic state transition diagram of the multi-agent cooperative system in Embodiment 1 of the present invention;

[0041] Figure 3 This is a schematic diagram of the probabilistic state transition of the multi-agent cooperative system in Embodiment 2 of the present invention. Detailed Implementation

[0042] This invention proposes a method based on A probabilistic analysis method for multi-agent cooperative systems includes the following steps:

[0043] Step 1: Assume that the multi-agent cooperative system consists of multiple agents, represented as: ,use The functional modules, behavioral states, state triggering conditions, interaction interfaces, and cooperative constraints of each agent in a multi-agent cooperative system are modeled and described to obtain the multi-agent cooperative system. Model. As one implementation method, based on Modeling a multi-agent cooperative system yields... Engineering Model ;in, This represents the set of functional modules corresponding to the intelligent agent; This represents the interaction interface relationship between intelligent agents; This represents the set of behavioral models for each intelligent agent.

[0044] For any intelligent agent Its behavioral model Represented as:

[0045]

[0046] in: For intelligent agents A set of behavioral states; A set of events that trigger state changes; This refers to the set of state triggering conditions or collaborative constraints.

[0047] Step 2: Based on predefined mapping rules, The structural and behavioral elements in the model are converted into symbolic state variables and state transition relationships, constructing a symbolic state transition model to describe the cooperative behavior of multiple agents. This symbolic state transition model is a Symbolic Transition System (STS), defined as follows: ,in: A set of symbolic state variables; This is a symbolic set of system states used to represent the system's behavioral states at different control positions. It is a set of state transition relations. This represents a constraint predicate defined on a set of symbolic state variables. In STS, the system state is represented by a combination of values ​​for one or more symbolic state variables, and state transitions are controlled by triggering conditions or cooperative constraints.

[0048] Predefined mapping rules can be specifically understood as... In the model, agent behavior states are mapped to symbolic state variables, state change relationships between agent behaviors are mapped to state transition relationships, and state triggering conditions and cooperative constraints are mapped to state transition triggering conditions or constraint conditions, specifically as follows:

[0049] For any agent's behavioral state Mapped to a symbolic state variable ;

[0050] Relationship to any state change ,in This is mapped to state transitions:

[0051]

[0052] in, Represents a symbolic constraint expression generated by triggering conditions or collaborative constraints;

[0053] The interaction between multiple agents is represented by sharing state variables or synchronizing state transitions.

[0054] Step 3: Based on the symbolic state transition model, probabilistic parameters are introduced into the state transition relationships to characterize the uncertain behavior in the multi-agent cooperative process, thus constructing a probabilistic state transition model. This probabilistic state transition model is a probabilistic symbolic transition system (SPTS), defined as follows: ,in, ;

[0055] For any state transition After introducing probability parameters, it is expanded to , where p represents the probability that the system will transition from state s to state s′ under the condition that constraint φ is satisfied.

[0056] The probability parameters are configured to correspond one-to-one with specific state transition relationships, and are used to characterize the probability of a state transition from the source state to the target state under the corresponding triggering conditions.

[0057] Step 4: Based on the probabilistic state transition model, analyze and evaluate the success probability and system reliability of the multi-agent collaborative task, and output the corresponding probability analysis results. The probability analysis results include the success probability of the multi-agent collaborative task, the overall system reliability index, or the probability of occurrence of key states. The probability analysis results are used to support the design optimization, operation evaluation, or decision support of the multi-agent collaborative system.

[0058] Collaborative Task and Probabilistic Analysis Objective Definition: Define the set of objective states for a multi-agent collaborative task as follows:

[0059]

[0060] Given an initial state Under certain conditions, the success probability of a multi-agent cooperative task is defined as the probability of the system starting from its initial state. via probabilistic state transition set Reaching the target state set The probability of:

[0061]

[0062] The success probability and the probability of occurrence of critical states are used together to evaluate the collaborative behavior of multi-agent cooperative systems.

[0063] Taking a simplified multi-agent cooperative task execution scenario as an example, assume there is a cooperative system consisting of two agents, agent A and agent B. The two agents cooperate to complete a joint task, and their basic cooperative behavior process includes:

[0064] The agent is in an idle state ( );

[0065] Upon receiving a task instruction, it enters the execution state. );

[0066] After execution, it enters the completion state. );

[0067] When any agent fails to execute its function, the system enters a failure state. ).

[0068] The requirement for this collaborative task is that the task is considered successfully completed only when both agents successfully complete their execution.

[0069] In this embodiment, firstly, the following is adopted: Engineering modeling of multi-agent collaborative systems.

[0070] Structural modeling: utilizing The Block Definition Diagram (BDD) defines the structural elements contained in the system, including: cooperative system Block; agent A Block; agent B Block; and communication interface between agents.

[0071] Behavioral modeling: utilizing A State Machine Diagram describes the behavior of each agent, and its set of states includes: Transitions between states are triggered by events or conditions, for example:

[0072] Event triggered → ;

[0073] Event triggered → ;

[0074] Event triggered → .

[0075] Through the above Modeling can intuitively depict the structural composition of a system and the behavioral state of an agent, but the model does not yet possess semantics that can be directly used for formal analysis.

[0076] In this embodiment, to achieve a unified representation of multi-agent cooperative systems from engineering modeling to formal analysis, a predefined... The mapping rules from engineering models to symbolic state transition systems (STS) are used to guide the construction of symbolic state variables, system states, and state transition relationships.

[0077] Mapping rules should include at least the following:

[0078] Rule 1: Mapping from structuring elements to state variables

[0079] The Block elements that describe the agent are mapped to symbolic state variables;

[0080] Each agent Block corresponds to a symbolic state variable, which is used to describe the current behavioral state of the agent.

[0081] Rule 2: Mapping from state machine states to variable ranges

[0082] The states defined in the state machine diagram (such as...) ), which is mapped to the range of values ​​of the corresponding symbolic state variables;

[0083] The set of values ​​for symbolic state variables It consists of all the states in its corresponding state machine diagram.

[0084] Rule 3: Mapping from state combinations to system states

[0085] The combination of values ​​of multiple symbolic state variables is mapped to a system state;

[0086] System state is used to represent the joint behavioral state of multiple agents at the same moment.

[0087] Rule 4: Mapping from state machine transitions to state transition relationships

[0088] Migration triggering conditions, events, or guard conditions are mapped to constraint predicates defined on symbolic state variables. ;

[0089] The state transitions in the state machine diagram are mapped to the state transition relationships in the STS. .

[0090] Rule 5: Unified mapping of abnormal or failure states

[0091] When any agent enters a failure state, the entire system enters a failure-related state.

[0092] In STS, failure states are modeled as failure absorption states, which are used to characterize unrecoverable failure scenarios in collaborative tasks.

[0093] Through the above mapping rules, it can be The structure, behavior, and interaction relationships in the engineering model are systematically transformed into a symbolic state transition system with formal semantics.

[0094] Based on predefined mapping rules, The model is converted into a symbolic state transition system (STS), such as Figure 2 As shown.

[0095] Define the symbolic state transition system as follows:

[0096]

[0097] in: A set of symbolic state variables; It is a symbolic set of states of the system, used to represent the behavioral state of the system at different control positions; It is a set of state transition relations; This represents a constraint predicate defined on a set of symbolic state variables.

[0098] Define the set of symbolic state variables as follows:

[0099]

[0100] in: This represents the current state of agent A; This represents the current state of agent B.

[0101] The range of values ​​for each state variable is:

[0102] ,

[0103] The symbolic state set of the system It is represented by the combination of values ​​of the symbolic state variables, that is:

[0104]

[0105] For example:

[0106] This indicates that both agents are in an idle state;

[0107] This indicates that both agents are in the execution state;

[0108] This indicates that agent A is currently executing, while agent B has completed.

[0109] This indicates that the collaborative task has been completed;

[0110] This indicates that the system is in a failure-related state.

[0111] State transition relationships are used to characterize the state evolution process of a system under certain conditions. In this embodiment, some typical state transitions are defined as follows:

[0112] ①Task Initiation Transfer:

[0113]

[0114] Indicates when a start event is received At that time, both intelligent agents simultaneously enter the execution state.

[0115] ② A single agent completes the transfer:

[0116]

[0117] This indicates that when agent A completes its execution, its state changes from... Become .

[0118] ③ Successful collaborative transfer:

[0119]

[0120] This indicates that the system has entered a stable completion state after the collaborative task has been successfully completed.

[0121] ④ Failure transfer:

[0122]

[0123] This indicates that if any agent fails to execute, the system enters a failure-related state.

[0124] Based on the STS model described above, formal analysis of system behavior can be performed, for example:

[0125] 1. Determine if there exists a state from the initial state. Departure Arrival The reachable path;

[0126] 2. Determine if there is a necessary entry. The execution path of the state;

[0127] 3. Analyze the conditions for collaborative completion under different execution orders.

[0128] Through the above methods, it was achieved from The complete technical process of engineering modeling → symbolic state transition system → formal analysis.

[0129] To further consider the uncertainties caused by factors such as communication instability and execution failure during multi-agent collaboration, the system model is extended probabilistically.

[0130] In practical multi-agent cooperative systems, the execution process of agent behavior is often not completely deterministic. For example:

[0131] 1. When an intelligent agent is performing a task, it may fail due to environmental interference;

[0132] 2. Communication between intelligent agents may experience packet loss or delays, leading to asynchrony in coordination;

[0133] 3. Under the same execution conditions, the success rate of task completion varies.

[0134] Therefore, it is necessary to characterize the probability of state transitions to reflect the stochastic characteristics in the multi-agent cooperative process.

[0135] For symbolic transfer systems Probabilistic extensions are performed to construct a probabilistic symbolic transfer system (SPTS), such as... Figure 3 As shown, it is defined as:

[0136]

[0137] in, For a set of symbolic state variables, For the system state set; It is a set of state transition relations with probability parameters.

[0138] Define the state transition probability rules from STS to SPTS, including:

[0139] Rule 1: Probabilistic Extension of Deterministic Transitions

[0140] For each state transition in STS In SPTS, this is expanded to: .in, This indicates that the constraint predicate is satisfied. Under the condition of state, Transition to state The probability of occurrence.

[0141] Rule 2: Modeling the probability of successful agent behavior

[0142] The state transitions associated with the actions of a single agent are probabilities determined by the agent's success probability; for example, the probability of a state transition representing the completion of agent A's action is denoted as... .

[0143] Rule 3: Probability Combination Rule for Cooperative Transfers

[0144] When a system state transition is triggered by the independent actions of multiple agents, the transition probability is determined by the action probabilities of each agent according to a predefined combination rule. In one implementation, the combination rule is in the form of a probability product, used to characterize the case where the actions of each agent are independent of each other.

[0145] Rule 4: Probabilistic Modeling of Failure Transfer

[0146] When any agent enters a failure state When this happens, the system transitions to a failure-related state. The probability of this transition is determined by the failure probability of the corresponding agent (e.g., ...). The probability of its complement is determined.

[0147] By using the above rules, probability parameters can be introduced to construct an SPTS model for analyzing the success probability of collaborative behavior and assessing system credibility without changing the STS structure.

[0148] In this embodiment of the invention, probability parameters are introduced into some state transition relationships to characterize the uncertainty of cooperative behavior.

[0149] For example, initiating a transfer for a task:

[0150]

[0151] After introducing probability parameters, it is expanded to:

[0152]

[0153] in This represents the probability that both agents will successfully enter the execution state after receiving the task start instruction.

[0154] Similarly, a probability parameter is introduced for the agent to complete the transition, for example:

[0155]

[0156] in This represents the probability that agent A successfully completes the task during execution.

[0157] If the agent fails to execute, a corresponding failure transition can be defined, for example:

[0158]

[0159] Defined collaborative task success status:

[0160]

[0161] The initial state of the system is defined as follows:

[0162]

[0163] In the probabilistic symbolic transition system SPTS, analysis is performed from the initial state... Starting from a point, the process involves probabilistic state transitions to reach the target state set. The path can be used to calculate the probability of successful completion of a multi-agent collaborative task.

[0164] In one example scenario, suppose:

[0165] 1. The probability of the task starting successfully is: ;

[0166] 2. The success probabilities of agent A and agent B are respectively... and ;

[0167] 3. The actions of each agent are independent of each other.

[0168] The success probability of a collaborative task can then be expressed as:

[0169]

[0170] By changing the values ​​of probability parameters, we can analyze the likelihood of the system completing the task under different cooperative conditions, thereby evaluating the behavioral reliability and trustworthiness of the multi-agent cooperative system.

[0171] By introducing probabilistic parameters into the symbolic transfer model, the logical conditions and probabilities of multi-agent cooperative behavior can be described simultaneously within a unified model framework, enabling quantitative analysis of the uncertainty of cooperative behavior. Compared with analytical methods that only use deterministic models, this approach more realistically reflects the behavioral characteristics of multi-agent cooperative systems in actual operation, providing more valuable analytical results for system design and operational decisions.

Claims

1. A method based on A probabilistic analysis method for multi-agent cooperative systems, characterized by: include: Step 1: Assume the multi-agent cooperative system consists of multiple agents, and utilize... The functions, behaviors, and interaction interfaces of each agent in the multi-agent cooperative system are modeled and described to obtain the multi-agent cooperative system. The model; the behavior of an intelligent agent includes: behavioral state, events that trigger state changes, state triggering conditions, and cooperative constraints; Step 2: Based on predefined mapping rules, the... The structural and behavioral elements in the model are converted into symbolic state variables, symbolic states, and state transition relationships to obtain a symbolic state transition model for describing the cooperative behavior of multiple agents. The symbolic state variables include the current state of each agent, and the symbolic state is represented by a combination of values ​​of the symbolic state variables. The state transition relationships include task initiation transition, single agent completion transition, cooperative success transition, and failure transition. Step 3: Based on the symbolic state transition model, introduce probability parameters into the state transition relationship to characterize the uncertain behavior in the multi-agent cooperative process, and construct the probabilistic state transition model. Step 4: Based on the probabilistic state transition model, output the corresponding probability analysis results, which include the success probability of the multi-agent collaborative task, the overall credibility index of the multi-agent collaborative system, or the probability of occurrence of key states.

2. A method based on claim 1 A probabilistic analysis method for multi-agent cooperative systems, characterized by: In step 2, the predefined mapping rules include: Will In the model, the agent's behavioral state is mapped to symbolic state variables, which are used to characterize the agent's current behavioral state; Will The state change relationships between agent behaviors in the model are mapped to state transition relationships, and state triggering conditions and cooperative constraints are mapped to state transition triggering conditions or constraint conditions. The interaction between multiple agents is represented by sharing state variables or synchronizing state transitions.

3. A method based on claim 1 A probabilistic analysis method for multi-agent cooperative systems, characterized by: The probabilistic state transition model is expressed as follows: ,in, For a set of symbolic state variables, For the system state set, ; transition to any state After introducing probability parameters, it is expanded to , where p represents the transition probability, which represents the probability of transitioning from state s to state s′ under the condition of satisfying the cooperative constraint φ.

4. A method based on claim 3 A probabilistic analysis method for multi-agent cooperative systems, characterized by: The probability parameters are configured to correspond one-to-one with specific state transition relationships.

5. A method based on claim 4 A probabilistic analysis method for multi-agent cooperative systems, characterized by: The state transitions associated with the actions of a single agent are determined by the probability of the agent's successful execution.

6. A method based on claim 4 A probabilistic analysis method for multi-agent cooperative systems, characterized by: When a system state transition is triggered by the independent actions of multiple agents, the transition probability is determined by the action probabilities of each agent according to a predefined combination rule.

7. A method based on claim 4 A probabilistic analysis method for multi-agent cooperative systems, characterized by: When any agent enters a failure state, the transition probability is determined by the corresponding agent's failure probability or its complement probability.

8. An electronic device, characterized in that, The electronic device includes: At least one processor; and a memory communicatively connected to the at least one processor; The memory stores a computer program executable by the at least one processor, which is executed by the at least one processor to enable the at least one processor to perform any one of claims 1-7 based on A probabilistic analysis method for multi-agent cooperative systems.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the implementation of any one of claims 1-7 based on A probabilistic analysis method for multi-agent cooperative systems.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the implementation based on any one of claims 1-7. A probabilistic analysis method for multi-agent cooperative systems.

Images