Distributed maintenance task distribution method for formation-level maintenance team

Abstract

The invention relates to a formation-level maintenance team distributed maintenance task allocation method, and belongs to the technical field of operation and maintenance guarantee and task allocation. Comprising the following steps: establishing a reliability model of a polymorphic system, analyzing the performance of the system in a task, and performing selective maintenance planning for a single repairable system; establishing a communication topology model for the formation, describing the income of each maintenance action, and establishing a maintenance task distribution model based on the initial positions of the maintenance stations and all members in the formation; and solving the formation maintenance task allocation model by taking maximization of the overall maintenance income of the formation as a target and uncertainty of maintenance time to obtain conflict-free allocation of the maintenance tasks, including tasks and path sequences of each maintenance team. According to the invention, conflict-free distributed maintenance task allocation can be realized on the basis of ensuring the performance of the formation task, and the availability of a maintenance allocation scheme when the maintenance time is uncertain is improved.

Description

Technical Field

[0001] The invention relates to a distributed maintenance task allocation method for a formation-level maintenance team, and belongs to the technical field of operation and maintenance guarantee and task allocation. Background Art

[0002] A repairable system formation can perform maintenance between two tasks when performing consecutive tasks with intervals; the formation's centralized maintenance decision center may not exist or be unavailable, and the maintenance team needs to allocate maintenance tasks to improve mission efficiency.

[0003] Currently, fleet-level maintenance decision-making methods do not consider the impact of maintenance provider scheduling on maintenance efficiency, which means that the results of maintenance decisions may not fully reflect the actual situation of the project. In addition, the uncertainty of the maintenance time required for each system in the fleet also leads to the situation that maintenance tasks cannot be completed as expected.

[0004] Therefore, in the interval between adjacent tasks, considering the characteristics of distributed formation maintenance task allocation, ensuring the conflict-free distribution of maintenance tasks, and considering the uncertainty of the maintenance time required by the system, so as to improve the maintenance efficiency of the formation, is the main purpose of formation maintenance decision-making and a problem that technical personnel in this field urgently need to solve. Summary of the invention

[0005] The present invention provides a distributed maintenance task allocation method for a formation-level maintenance team, which is mainly used to solve the problem of conflict-free maintenance task allocation considering the uncertainty constraint of system maintenance time when an equipment formation makes maintenance decisions.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] The present invention provides a distributed maintenance task allocation method for a formation-level maintenance team, which has the following special features:

[0008] Step 1: Analyze the mission process of the formation, describe the mission and maintenance time series of the system, model the repairable system as a multi-state system, and use the homogeneous Markov model to establish the relationship between the performance of the system in the mission and its health state when it starts to perform the mission;

[0009] Step 2: Analyze the maintainability of the system according to the structure of the repairable system, obtain the minimum performance constraint of the system according to the task requirements, optimize the maintenance strategy through heuristic methods, and obtain the selective maintenance plan for each system in the formation;

[0010] Step 3: Establish a communication topology model for the formation according to the actual communication strategy of the formation, describe the value of each maintenance action, and establish a distributed maintenance task allocation model among the maintenance teams based on the maintenance sites and the initial positions of all members in the formation;

[0011] Step 4: With the goal of maximizing the overall maintenance benefit of the formation and the uncertainty of maintenance time, the consensus-based task package algorithm is used to solve the formation maintenance task allocation model to obtain a conflict-free allocation of maintenance tasks, including the tasks and path sequences of each maintenance team.

[0012] Preferably, the expression of the distributed maintenance task allocation model in step 3 is:

[0013]

[0014] In the formula, R represents the total revenue of fleet maintenance, A represents the size of the maintenance team, and N f represents the size of the formation, c ij and τ ij They represent the benefits and time of maintenance actions respectively, and p i is a list of tasks assigned to the agent task package. The assignment relationship between maintenance agent i and repairable system j is defined as vector x i , N t 、N a denote the number of subsystems in the system and the number of components in the ith subsystem, ΔP denotes the performance increase after maintenance, and T max Indicates the time available for maintenance operations. The maintainer and system index sets are defined as If the maintenance task of system j is assigned to maintainer i, then the binary decision variable x ij is 1.

[0015] Preferably, in step 4, the consensus-based task package algorithm, in which each maintenance team is regarded as an intelligent agent, each intelligent agent i carries four types of vectors, namely, task list b i , path list p i , winning bid list y i and the list of winning agents z i , where task list b i Contains all maintenance tasks assigned to the agent, the path list p i is the ordered task sequence of agent i. For example, if agent i performs task j at the kth point in its path, then its kth element is j and the bidding list y i Represents the bidding information of all agents for the task that agent i has mastered during the bidding process, and the list of winning agents z iRepresents the winning bid situation of agent i, that is, which maintenance agent the maintenance task is assigned to.

[0016] Preferably, in step 4, the process of solving the problem using a consensus-based task package algorithm includes:

[0017] Task package construction: Each agent creates a task package and continuously adds tasks to it until it can no longer add tasks. Since different task execution times will result in different task benefits, for each newly added task, the algorithm will add it to the task package to the position that can produce the greatest benefit improvement;

[0018] In the conflict resolution phase, a consensus strategy will be used to resolve conflicts in which a task is assigned to multiple agents. Agents add tasks to the task package based on the set of tasks that have been assigned. Agents use the winning bid list y i 、List of winning agents z i It reaches a consensus on the timestamp list of this information obtained from other agents through communication. According to a series of defined rules, the agent can decide whether to update its several lists. If the task package is updated or reset, it returns to the task package construction phase.

[0019] Preferably, the consensus-based task bundle algorithm takes into account the uncertainty of system-level maintenance time. The algorithm allows the robot to consider the randomness of task duration when building and scoring bundles, reducing the situation where the robot cannot complete all tasks assigned in the bundle or the robot's capabilities are not fully utilized due to the randomness of task duration. The method includes statistically summing the random time of each task in the task bundle and comparing it with a given time window to determine whether the task bundle is full. In the proposed maintenance task scenario, the maintenance task duration is a random variable, and the time spent by the maintenance agent on the journey is generally fixed. Therefore, the total time spent by the maintenance agent to execute the task bundle can be divided into two parts: the deterministic time consumed in the journey and the uncertain time considered when investing in the maintenance task. Without considering the failure of some tasks to be completed, the time for maintenance agent i to complete the jth task in its sequence is:

[0020]

[0021] in, represents the fixed time taken by the maintenance agent to perform the kth task, that is, the time consumed during the journey, It represents the uncertainty time taken by the maintenance agent to perform the kth task, that is, the time required for maintenance.

[0022] Preferably, in the consensus-based task package algorithm, when constructing a task package, the agent predicts the total time to complete all tasks in the task package based on the probability distribution of the completion time of each task. If the probability that the task to be added to the package will cause the total time to exceed the maintenance interval duration is greater than the specified probability, the agent will not add the task to the package. This filtering strategy greatly reduces the risk that the agent will not successfully complete all tasks in the task package in the context of uncertain task duration, especially when the uncertainty is high.

[0023] It can be seen from the above technical solutions that, compared with the prior art, the present invention discloses a method for allocating distributed maintenance tasks for a formation-level maintenance team, which uses a selective maintenance planning method to obtain a selective maintenance plan for each system in the formation according to the health status of a single repairable system and its minimum performance constraint in the task. Then, a distributed maintenance task allocation model is established with the goal of maximizing the formation maintenance benefit, and the model is solved using a consensus-based task package algorithm. The consensus-based task package algorithm is optimized for task scenarios with uncertainty in maintenance time, thereby achieving conflict-free formation-level maintenance task allocation.

[0024] The present invention can realize conflict-free formation-level maintenance task allocation while taking into account the decentralized task allocation method and the uncertainty of system maintenance time, thereby improving maintenance efficiency and providing technical support for the research on formation-level maintenance decision-making methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0026] Figure 1 A flow chart of the fleet maintenance task allocation method provided by the present invention;

[0027] Figure 2 A simplified structural block diagram of a repairable system to which the method provided by the present invention is applicable;

[0028] Figure 3 A Gantt chart of maintenance tasks obtained by the consensus-based task package algorithm provided by the present invention;

[0029] Figure 4 This is a schematic diagram of a maintenance task allocation result provided by the present invention. DETAILED DESCRIPTION

[0030] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0031] Decentralized formation-level maintenance decisions need to consider the needs of the subsequent tasks that the formation needs to perform and the health status of each member in the formation, adopt selective maintenance plans for each member, and then assign the maintenance plans as tasks to all maintenance teams to obtain the optimal system maintenance plan and conflict-free maintenance task allocation plan.

[0032] In this regard, an embodiment of the present invention discloses a distributed maintenance task allocation method for a formation-level maintenance team. Through a reliability analysis method, the performance of the repairable system as a member in the task is analyzed when the formation task requirements and the health status of each member are known. A heuristic method is used to calculate a selective maintenance plan for each member, and then the maintenance task information of each member is transmitted to the maintenance teams located at different sites. Considering the uncertainty of maintenance time, the maintenance tasks of the formation are decentralized allocated through a consensus-based task package algorithm to obtain a conflict-free task allocation plan. This method can improve the maintenance efficiency of the formation, reduce the task risk caused by the uncertainty of system maintenance time, and provide an efficient and conflict-free solution and its theoretical basis for formation-level maintenance decisions that consider distributed communication and maintenance time uncertainty.

[0033] Specifically, the maintenance planning method of the present invention includes the following steps: Figure 1 As shown,

[0034] Step 1: Analyze the mission process of the formation, describe the system's mission and maintenance time series, model the repairable system as a multi-state system, and use the homogeneous Markov model to establish the relationship between the system's performance in the mission and its health state when it starts the mission.

[0035] Step 2: According to the structure of the repairable system, analyze the maintainability of the system, obtain the minimum performance constraints of the system according to the mission requirements, optimize the maintenance strategy through heuristic methods, and obtain the selective maintenance plan for each system in the formation.

[0036] Step 3: Establish a communication topology model for the formation according to the actual communication strategy of the formation, describe the value of each maintenance action, and establish a distributed maintenance task allocation model among the maintenance teams based on the maintenance sites and the initial positions of all members in the formation.

[0037] Step 4: With the goal of maximizing the overall maintenance benefit of the formation and the uncertainty of maintenance time, the consensus-based task package algorithm is used to solve the formation maintenance task allocation model to obtain a conflict-free allocation of maintenance tasks, including the tasks and path sequences of each maintenance team.

[0038] In order to make the technical approach, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0039] First, the subsequent mission requirements of the formation and the health status of each formation member are analyzed to obtain the performance of each system in the subsequent mission when the status after maintenance is known, including: analyzing the maintenance support mode of the formation, and combining the random process model to give a method for analyzing the health status of the formation in the mission.

[0040] When the state transition rate of the multi-state repairable system is known, the state of the system at the end of the task is obtained by solving the differential equation, so as to determine whether the system can maintain its mission capability to meet the requirements during the task.

[0041] Secondly, the health management system provides selective maintenance solutions for each repairable system as a member of the formation, including:

[0042] Under the condition of known mission conditions and system health status, the health management module carried by each system uses heuristic algorithms such as particle swarm algorithm to perform local optimization calculations in parallel to obtain the optimal selective maintenance plan for each system.

[0043] Secondly, the maintenance task characteristics and communication topology of the formation are analyzed, and a decentralized formation maintenance task allocation model is constructed, including:

[0044] The objectives and constraints of formation maintenance task allocation are analyzed, and a mathematical model describing the characteristics of formations and maintenance teams is established; the performance of the formation in the task is calculated in combination with the task requirements and the health status of the formation; a distributed formation-level maintenance task allocation model is established with the maximization of the total reward for formation maintenance as the optimization goal and the characteristics of decentralized maintenance task allocation and total maintenance time as constraints. By solving this model, a conflict-free allocation of formation maintenance tasks by maintenance teams located at different sites can be obtained.

[0045] In one embodiment, selectively allocating maintenance tasks to a repairable system includes:

[0046] The characteristics of the mission that the formation is about to perform are analyzed, the mission and maintenance time series of the system are described, the repairable system is modeled as a multi-state system, and the homogeneous Markov model is used to establish the relationship between the system's performance in the mission and its health state when it starts the mission.

[0047] According to the structure of the repairable system, the maintainability of the system is analyzed, and the minimum performance constraint of the system is obtained according to the mission requirements. The maintenance strategy is optimized through the particle swarm algorithm to obtain a selective maintenance plan for each system in the formation.

[0048] In the proposed maintenance task scenario, the duration of the maintenance task is a random variable, while the time spent by the maintenance agent on the journey is generally fixed. Therefore, the total time taken by the maintenance agent to execute the task package can be divided into two parts: the deterministic time consumed on the journey and the uncertain time considered when investing in the maintenance task. Without considering the failure of some tasks to be completed, the time it takes for the maintenance agent i to complete the jth task in its sequence is:

[0049]

[0050] in, represents the fixed time taken by the maintenance agent to perform the kth task, that is, the time consumed during the journey, It represents the uncertainty time taken by the maintenance agent to perform the kth task, that is, the time required for maintenance.

[0051] Based on the description and analysis of the problem, a selective maintenance model for a single repairable system can be established. Given the initial health state of each component before maintenance, the goal of selective maintenance is to select a reasonable maintenance plan for the system so that the mission capability of the system after maintenance can meet the upcoming mission requirements. This goal can be summarized as: Under the constraint that the system health state meets the next wave of mission requirements, plan appropriate maintenance levels for each component in the system to maximize the maintenance reward. If minimizing the system maintenance time is used as the maintenance reward, the optimization model of the system selective maintenance planning problem can be established as follows

[0052]

[0053] subject to:P - >P min

[0054]

[0055] In one embodiment, constructing a fleet-level maintenance task allocation model includes:

[0056] 1. Obtain the maintenance task information of each member in the formation, that is, the selective maintenance plan for each repairable system. The maintenance task information mainly includes the task location coordinates, task rewards and task duration, etc., which is the information that the maintenance team needs to master before task assignment.

[0057] 2. Establish a communication topology model for the formation according to the actual communication strategy of the formation, describe the value of each maintenance action, and establish a distributed maintenance task allocation model among the maintenance teams based on the maintenance sites and the initial positions of all members in the formation.

[0058] For a swarm, the goal of maintenance task allocation is to find a conflict-free match between the system and the maintenance agent, thereby maximizing some global reward. f The global reward function for repair task assignment is the sum of local reward values, and each local reward is determined by the repair agent and the repair task assigned to it. Suppose the reward c for repairing system j by agent i is ij The time to complete this maintenance is τ ij and increase the mission capability of the system Determine the time τ to complete this maintenance ij The path p of agent i i The only certainty is p i Depending on the list of tasks in the task bag assigned to the agent, the assignment relationship is defined as the vector x i Based on the above assumptions, the maintenance task allocation problem can be described by the following integer programming model:

[0059]

[0060] Thirdly, the present invention designs a consensus-based task package algorithm that takes into account the randomness of system maintenance time to solve the distributed formation-level maintenance task allocation model, including:

[0061] The formation-level maintenance task allocation problem is analyzed, and a distributed task allocation method is adopted based on the communication topology of the maintenance team. The consensus-based task package algorithm is improved while considering the uncertainty of maintenance time, and the distributed formation-level maintenance task allocation model is solved to obtain a conflict-free allocation of maintenance tasks that maximizes the formation maintenance reward.

[0062] In one embodiment, in a consensus-based task package algorithm considering task duration uncertainty, the agent cannot directly determine whether the task package is full during the construction process because the actual duration of the task is uncertain. In some cases, later tasks in the task package may not be completed correctly, while in other cases, all tasks in the task package may be completed prematurely, resulting in agent idle time. The consensus-based task package algorithm considering task duration uncertainty adopts a dynamic allocation mechanism during the iteration process. When building the package, the probability of maintenance task completion is taken into account, reducing the task risk caused by time uncertainty in the task allocation stage.

[0063] In this example, the proposed method is used to make maintenance decisions for a vehicle fleet, and a selective maintenance plan for the system and a conflict-free maintenance task allocation plan for the maintenance provider are obtained to illustrate the effectiveness of the proposed method. The repairable system studied in the vehicle is a grain collection system in a harvester cluster, which consists of Figure 2 The system consists of five subsystems, totaling 14 components. In the system, crops go through the cutting and collecting system, conveying system, threshing and separation system, conveying system and cleaning system in turn, and finally complete the processing and storage of crops. This system can be regarded as a flow transmission system, and this type of system structure has received much attention in previous maintenance planning research. According to the characteristics of the system, its mission capability can be expressed as:

[0064] G=min(G s1 ,G s2 ,...,G sn )

[0065] At any time during the mission, if the system's mission capability is less than the upcoming mission demand C min , the state of the system is considered to be failed. The proposed method is used to simulate the maintenance task decision of the case. Given the health status of all systems and the duration of the upcoming task, the PHM system of each repairable system uses a heuristic algorithm to complete the planning of the system's selective maintenance plan.

[0066] After obtaining the maintenance plan for each member of the formation through a heuristic method, the formation can pass the maintenance demand information to each maintenance provider, and then the maintenance providers can allocate tasks among them. This section uses a consensus-based task package algorithm that considers the duration of random tasks to solve this problem. In the maintenance task allocation scenario studied, the maintenance providers and vehicle formations are located in a two-dimensional plane space with an area of ​​W*W (W=2km). The upcoming task of the formation requires the maintenance task to be completed within the time window of [0-1200s]. The positions of each member of the formation and the initial positions of the maintenance sites where the maintenance providers are located are randomly generated in space. The movement speed of the maintenance provider is 5m / s. The uncertainty of the maintenance time of each member in the formation is characterized by a normal distribution with a mean of μ and a variance of σ. Its mean μ is calculated by the maintenance plan given by the heuristic algorithm, and the variance is unified as σ=30. It is assumed that there are three maintenance providers performing maintenance tasks on the formation, and a fully connected communication topology is established between them. The minimum performance requirement of the task for each system is set to 15.5 tons per hour. The algorithm is set to consider the case where the maximum maintenance time for each system reaches μ+σ.

[0067] Figure 4The task allocation of maintenance providers is shown, where the x-axis and y-axis represent the position coordinates of the maintenance provider or the repairable system formation in two-dimensional space, and the z-axis represents time. As can be seen from the figure, the maintenance provider converges to a conflict-free allocation by abandoning one goal, ensuring the completion of more system maintenance operations while meeting the maintenance time window constraint.

[0068] Finally, the distributed task allocation method obtains a set of conflict-free allocation schemes that maximize maintenance benefits for the fleet maintenance task allocation model. The running time of the consensus-based task package algorithm that considers the uncertainty of maintenance tasks is comparable to that of the baseline algorithm, because the uncertainty is processed in the stage of pre-processing the previous data. In effect, since redundancy is left for the fluctuation of the maintenance operation duration, the proposed method can avoid the situation where some maintenance operations cannot be completed as expected to a certain extent. In addition, if there is a maintenance operation that is highly likely to be incomplete, the proposed method tends to abandon the maintenance operation, thereby saving maintenance resources and notifying the target early so that it can be released from the next task and handle other matters.

[0069] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0070] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A distributed maintenance task allocation method for a formation-level maintenance team, characterized in that: The steps include: Step 1: Analyze the mission process of the formation, describe the mission and maintenance time series of the system, model the repairable system as a multi-state system, and use the homogeneous Markov model to establish the relationship between the performance of the system in the mission and its health state when it starts to perform the mission; Step 2: Analyze the maintainability of the system according to the structure of the repairable system, obtain the minimum performance constraint of the system according to the task requirements, optimize the maintenance strategy through heuristic methods, and obtain the selective maintenance plan for each system in the formation; Step 3: Establish a communication topology model for the formation according to the actual communication strategy of the formation, describe the value of each maintenance action, and establish a distributed maintenance task allocation model among the maintenance teams based on the maintenance sites and the initial positions of all members in the formation; Step 4: With the goal of maximizing the overall maintenance benefit of the formation and the uncertainty of maintenance time, the consensus-based task package algorithm is used to solve the formation maintenance task allocation model to obtain a conflict-free allocation of maintenance tasks, including the tasks and path sequences of each maintenance team.

2. A distributed maintenance task allocation method for a formation-level maintenance team according to claim 1, characterized in that: In step 3, the expression of the distributed maintenance task allocation model among the maintenance teams is: maximize: satisfy: In the formula, R represents the total revenue of fleet maintenance, A represents the size of the maintenance team, and N f represents the size of the formation, c ij and τ ij They represent the benefits and time of maintenance actions respectively, and p i is a list of tasks assigned to the agent task package. The assignment relationship between maintenance agent i and repairable system j is defined as vector x i , N t 、N a denote the number of subsystems in the system and the number of components in the ith subsystem, ΔP denotes the performance increase after maintenance, and T max Indicates the time available for maintenance operations. The maintainer and system index sets are defined as If the maintenance task of system j is assigned to maintainer i, then the binary decision variable x ij is 1.

3. The method for allocating distributed maintenance tasks for a formation-level maintenance team according to claim 1, characterized in that: In step 4, the consensus-based task package algorithm is used, in which each maintenance team is regarded as an agent, and each agent i carries four types of vectors, namely, task list b i , path list p i , winning list y i and the list of winning agents z i .

4. The method for allocating distributed maintenance tasks for a formation-level maintenance team according to claim 1, characterized in that: In step 4, the process of solving the problem using the consensus-based task package algorithm includes: Task package construction: Each agent creates a task package and continuously adds tasks to it until it can no longer add tasks. Since different task execution times will result in different task benefits, for each newly added task, the algorithm will add it to the task package to the position that can produce the greatest benefit improvement; In the conflict resolution phase, the consensus strategy will be used to resolve the conflict in which a task is assigned to multiple agents. The agents add tasks to the task package based on the set of tasks that have been assigned. The agents list the winning bids y i 、List of winning agents z i It reaches a consensus on the timestamp list of this information obtained from other agents through communication. According to a series of defined rules, the agent can decide whether to update its several lists. If the task package is updated or reset, it returns to the task package construction phase.

5. The consensus-based task package algorithm according to claim 4, characterized in that: The consensus-based task bundle algorithm takes into account the uncertainty of system-level maintenance time. The algorithm allows the robot to consider the randomness of task duration when building and scoring bundles, reducing the situation where the robot cannot complete all tasks assigned in the bundle or the robot's capabilities are not fully utilized due to the random task duration. The algorithm includes statistically summing the random time of each task in the task bundle and comparing it with the given time window to determine whether the task bundle is full. In the proposed maintenance task scenario, the maintenance task duration is a random variable, and the time spent by the maintenance agent on the journey is generally fixed. Therefore, the total time spent by the maintenance agent to execute the task bundle can be divided into two parts: the deterministic time consumed on the journey and the uncertain time considered when investing in the maintenance task. Without considering the failure of some tasks to be completed, the time for maintenance agent i to complete the jth task in its sequence is: in, represents the fixed time taken by the maintenance agent to perform the kth task, that is, the time consumed during the journey, It represents the uncertainty time taken by the maintenance agent to perform the kth task, that is, the time required for maintenance.

6. The consensus-based task package algorithm according to claim 4, characterized in that: When the task package is constructed, the agent predicts the total time to complete all tasks in the task package based on the probability distribution of the completion time of each task. If the probability that the task to be added to the package will cause the total time to exceed the maintenance interval duration is greater than the specified probability, the agent will not add the task to the package.

Images