Equipment maintenance decision and task scheduling joint optimization method and system in fixed-point maintenance mode

Abstract

The invention belongs to the technical field of equipment maintenance, and discloses an equipment maintenance decision and task scheduling joint optimization method in a fixed-point maintenance mode, which comprises the following steps of: firstly, synthesizing the problems of numerous maintenance modes, complex maintenance constraints, cumbersome maintenance decision conditions and the like of equipment maintenance in the fixed-point maintenance mode; factors such as multiple maintenance modes, time conflicts, the number of spare parts, maintenance time and priorities are considered; secondly, a part changing decision is put forward for the first time and combined with a selective maintenance decision, the combination of the secondary use time and priority of maintenance equipment serves as maintenance income, the maintenance income under limited time and maintenance equipment number maximization serve as targets, and a more perfect mathematical model is established; finally, corresponding encoding and decoding modes are adopted according to the particularity of a solution in fixed-point maintenance; an adaptive hybrid particle swarm genetic algorithm (NSPSO-AD) is designed for a multi-objective optimization problem, and the algorithm can effectively improve the solving speed and quality.

Description

Technical Field

[0001] The present invention belongs to the technical field of equipment maintenance, and in particular relates to a method and system for joint optimization of equipment maintenance decision and task scheduling in a fixed-point maintenance mode. Background Art

[0002] Modern equipment is gradually developing towards unmanned and human-machine collaboration, and the traditional equipment maintenance support based on simple superposition of experience, technology, information, etc. has become relatively outdated. Therefore, how to quickly and effectively implement highly intelligent, highly agile, and highly accurate maintenance decision-making and task scheduling optimization is crucial.

[0003] At present, some scholars have begun to focus on the new idea of ​​maintenance decision-making and maintenance task scheduling optimization in the study of maintenance support issues. They fully consider the relationship between maintenance resources and maintenance tasks, and are committed to achieving reasonable resource allocation and task scheduling optimization under the constraint of limited maintenance resources.

[0004] Among the maintenance support modes, the fixed-point centralized maintenance mode belongs to base-level maintenance. Compared with the accompanying maintenance, it accounts for a larger proportion, bears less maintenance risk, and has stronger maintenance capabilities. When a drone or unmanned vehicle fails during a mission, it needs to return to the maintenance base for centralized maintenance as soon as possible because it carries critical information. Therefore, as a mode of maintenance support, fixed-point maintenance plays an important role in the equipment maintenance process.

[0005] Equipment maintenance decision-making is the basis for the implementation of maintenance support activities. In terms of maintenance decision-making, researchers often divide maintenance into different levels or stages such as major maintenance, medium maintenance, and minor maintenance based on the maintenance scale through fault analysis, and then specify maintenance plans to complete the entire maintenance decision-making process. However, this model combines data with subjective evaluation, separates decision-making from scheduling, and directly makes maintenance decisions, lacking consideration of the correlation between decision-making and scheduling.

[0006] As an important part of maintenance support, task scheduling optimization needs to ensure scientific resource allocation and rapid scheduling optimization in a complex battlefield environment with limited resources to meet the scientific and feasible implementation of equipment maintenance modernization. Considering the attributes and constraints such as maintenance task classification, priority, maintenance personnel and capacity load, a dynamic scheduling model is constructed and solved using the improved taboo algorithm. Considering factors such as time window, non-traversal, repair capacity, and repair status, the total number of repairs, the sum of importance, and the secondary combat time are taken as the goals, and the improved non-dominated sorting genetic algorithm is used to solve it. An equipment maintenance task scheduling model with equipment importance, maintenance time, and maintenance cost as indicators is established, and an improved particle swarm algorithm is proposed to solve the equipment maintenance task scheduling model. In view of the complex characteristics of maintenance personnel types and maintenance personnel levels, a maintenance process scheduling problem with the shortest maintenance time and the minimum total human resource load as the objective function is established, and the improved adaptive hybrid particle swarm genetic algorithm is used to solve it.

[0007] However, in terms of mathematical modeling, there is currently a lack of research on the optimization of maintenance decisions and task scheduling considering different failure modes in the fixed-point maintenance mode. Some studies abstract the optimization of maintenance task scheduling into a target optimization problem with simple constraints. The optimization results obtained are contrary to engineering requirements and ignore the fact that maintenance types correspond to different maintenance decisions when repairing different equipment. They ignore emergencies that occur during the maintenance process, such as "part replacement", "different maintenance types", "dynamic random failures" and other situations in the maintenance guarantee process. They do not consider the large number of equipment and the complex failure modes. They only abstract the faulty equipment into simple faulty equipment and then perform unified scheduling optimization, which is inconsistent with the actual equipment maintenance needs.

[0008] In view of the current research deficiencies, the present invention considers four different maintenance modes of equipment: periodic maintenance, scheduled maintenance, selective maintenance, and random fault maintenance, determines the priorities of different maintenance tasks and the deadlines of maintenance tasks, selects appropriate faulty equipment to be added to the maintenance sequence, and forms an initial maintenance decision. Taking into account factors such as time windows, number of spare parts, maintenance time, multiple fault modes, and maintenance task priority, a parts replacement strategy is proposed for the first time. Considering the situation of insufficient spare parts during the maintenance process, reasonable implementation of strategies such as parts replacement and selective maintenance can achieve equipment maintenance scheduling optimization. At the same time, the maintenance task priority, secondary execution task duration, and number of maintenance equipment are combined to establish a multi-objective optimization model with maintenance efficiency and maintenance quantity maximization as the optimization goals.

[0009] In terms of algorithm design and solution, most current studies use traditional algorithms to deal with high-complexity, discrete optimization problems. The algorithms have problems such as slow convergence, premature maturity, and easy to fall into local optimal solutions, which need to be solved.

[0010] In the traditional fixed-point maintenance mode, due to the single maintenance mode and the complexity of decision-making, it often leads to unreasonable maintenance task allocation, frequent time conflicts, and waste of resources, which cannot meet the needs of efficient equipment maintenance in complex scenarios. At the same time, the existing technology has a single measurement indicator for maintenance benefits, and fails to comprehensively consider the secondary use time and task priority of the equipment, resulting in limited optimization effect of maintenance decisions. In addition, in multi-objective optimization, traditional algorithms have difficulty balancing efficiency and accuracy, and cannot efficiently handle the dynamic scheduling and resource allocation of a large number of maintenance tasks. Summary of the invention

[0011] In view of the problems existing in the prior art, the present invention provides a joint optimization method for equipment maintenance decision and task scheduling in a fixed-point maintenance mode.

[0012] The present invention is implemented as follows: a joint optimization method for equipment maintenance decision and task scheduling in a fixed-point maintenance mode includes:

[0013] Step 1: Equipment maintenance under the integrated fixed-point maintenance mode has many maintenance modes, complex maintenance constraints, and cumbersome maintenance decision-making situations. Multiple maintenance modes, time conflicts, spare parts quantity, maintenance time, and priority factors are considered;

[0014] Step 2: For the first time, the decision of serial replacement is proposed, combined with the selective maintenance decision, and the secondary use time of the maintenance equipment is combined with the priority as the maintenance benefit. The maintenance benefit and the maximum number of maintenance equipment under limited time are taken as the goal to establish a more complete mathematical model;

[0015] Step 3: Adopt corresponding encoding and decoding methods according to the particularity of the solution in fixed-point maintenance; design an adaptive hybrid particle swarm genetic algorithm for multi-objective optimization problems; verify the effectiveness of the model and algorithm by designing simulation experiments and indicator evaluation, and realize the optimization of equipment maintenance decision-making and task scheduling under the fixed-point maintenance mode.

[0016] Furthermore, the mathematical model:

[0017] The model is established as follows:

[0018]

[0019] Among them, formula (1) is the objective function 1, which represents the maintenance benefit generated by maintenance according to the maintenance decision. The benefit is composed of the secondary use time of the equipment and the equipment importance coefficient;

[0020] Formula (2) is the objective function 2, which represents the maximum number of maintenance equipment;

[0021] Formula (3) shows that when the maintenance task execution timing is within the time window, the maintenance benefit is not affected; otherwise, the penalty factor A is introduced to appropriately reduce the maintenance benefit;

[0022] Formula (4) is the spare parts constraint, which is the necessary resource condition for maintenance equipment. Maintenance is carried out on the basis of ensuring sufficient spare parts. The replacement strategy can be appropriately added to change the number of spare parts. The decision is passed up to the maintenance timing, which affects the final maintenance decision.

[0023] Formula (5) and (6) are singular constraints, which means that maintenance is not performed or is performed only once;

[0024] Formula (7) is the constraint of serial replacement, which ensures the non-repeatability of the serial replacement strategy;

[0025] Formula (8) is the maintenance deadline constraint;

[0026] R 1 ~R 4 They represent the maintenance task sets of periodic maintenance, scheduled maintenance, preventive maintenance, and random failure maintenance equipment respectively; the maintenance task set R 2 ,R 3 ,R 4 It can be abstracted into a set of periodic maintenance tasks R 1 , the transformation process is:

[0027] 1)

[0028] Scheduled maintenance tasksx 3 ∈R 2 , after the calculation of the remaining life, it is converted into the maintenance time e 3 , it can be abstracted as a time window (-∞,e 3 );

[0029] 2)

[0030] Preventive maintenance tasksx 4 ∈R 3 , after detecting the fault curve, the time when the fault threshold σ is reached is e 4 , it can be abstracted as a time window (-∞,e 4 );

[0031] 3)

[0032] Random fault repair x 5 ∈R 4 , the time of random equipment failure is unknown, and can be abstracted as a time window (-∞,+∞).

[0033] Another object of the present invention is to provide a joint optimization system for equipment maintenance decision and task scheduling in a fixed-point maintenance mode, comprising:

[0034] Comprehensive module, used for equipment maintenance under the comprehensive fixed-point maintenance mode, which has many maintenance modes, complex maintenance constraints, and cumbersome maintenance decision-making situations, taking into account multiple maintenance modes, time conflicts, spare parts quantity, maintenance time, and priority factors;

[0035] The mathematical model building module is used to propose the replacement decision for the first time, combine it with the selective maintenance decision, combine the secondary use time of the maintenance equipment with the priority as the maintenance benefit, and establish a more complete mathematical model with the maintenance benefit and the maximum number of maintenance equipment under limited time as the goal;

[0036] The scheduling optimization module is used to adopt corresponding encoding and decoding methods according to the particularity of the solution in fixed-point maintenance; an adaptive hybrid particle swarm genetic algorithm is designed for multi-objective optimization problems; the effectiveness of the model and algorithm is verified by designing simulation experiments and indicator evaluation, and the equipment maintenance decision-making and task scheduling optimization under the fixed-point maintenance mode are realized.

[0037] Another object of the present invention is to provide a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method for jointly optimizing equipment maintenance decisions and task scheduling in the fixed-point maintenance mode.

[0038] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to execute the steps of the method for jointly optimizing equipment maintenance decisions and task scheduling in the fixed-point maintenance mode.

[0039] Another object of the present invention is to provide an information data processing terminal, which is used to implement a joint optimization system for equipment maintenance decision-making and task scheduling in the fixed-point maintenance mode.

[0040] In combination with the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solutions to be protected by the present invention are as follows:

[0041] First, the optimization problem of maintenance decision and task scheduling under the fixed-point maintenance mode is studied, and multiple failure modes, time windows, number of spare parts, maintenance priority, selective maintenance, serial replacement and other factors are comprehensively considered. The relevant mathematical model is established with the dual optimization goals of maximizing the number of maintenance and maintenance benefits.

[0042] The whole algorithm process was analyzed, and the corresponding encoding and decoding methods were designed according to the problem. The advantages of the commonly used intelligent optimization algorithms were combined to design related operators and propose an improved NSPSO-AD algorithm. The relevant parameters were set, and a comparative experiment was designed based on the above content. By comparing the solution results and convergence speed of the algorithm for the model, the scientific nature of the model and algorithm proposed in this paper was verified, and the maintenance decision-making and task scheduling optimization problems under the fixed-point maintenance mode were well solved. The research results show that the above model and algorithm can assist the maintenance decision-making and task scheduling optimization under the fixed-point maintenance mode, and can be effectively applied to practical engineering.

[0043] (1) The expected benefits and commercial value of the technical solution of the present invention after transformation are:

[0044] The present invention can be applied to the actual equipment maintenance and support process work, assisting maintenance decision-making in task scheduling optimization. The provided algorithm can effectively improve the convergence speed and accuracy of the solution set, and reduce the time, manpower and material resources spent on the entire maintenance decision.

[0045] (2) The technical solution of the present invention fills the technical gap in the industry at home and abroad:

[0046] The replacement strategy is proposed for the first time, providing new research content for maintenance decision-making and task scheduling optimization.

[0047] Second, the present invention proposes a replacement parts decision model for the first time, combines it with the selective maintenance decision, integrates the secondary use time and priority factors of the equipment, and constructs a mathematical model with the goal of maximizing maintenance benefits within a limited time. This model can flexibly deal with problems such as multi-maintenance mode switching, time conflicts, and resource constraints in complex maintenance scenarios, making maintenance task decisions more accurate and reasonable, and significantly improving maintenance efficiency.

[0048] The present invention designs an adaptive hybrid particle swarm genetic algorithm, which combines the global search capability of the particle swarm algorithm and the local optimization capability of the genetic algorithm, dynamically adjusts the search strategy for multi-objective optimization problems, and significantly improves the efficiency and accuracy of the algorithm. Compared with traditional methods, the optimization algorithm of the present invention can converge to the optimal solution faster, and can flexibly handle complex constraints such as task priority, spare parts allocation and time conflicts, ensuring the dynamic optimization scheduling of maintenance tasks.

[0049] Through the collaboration of software and hardware, the present invention realizes the dynamic optimization of real-time task scheduling and resource allocation. Combined with the high-performance hardware acceleration module, the system of the present invention can quickly adjust the allocation strategy of maintenance tasks and dynamically optimize resource allocation, effectively reducing the task conflict rate and resource waste. At the same time, through simulation experiments and multi-dimensional indicator evaluation, the present invention proves its superior performance in complex maintenance scenarios, greatly improves maintenance efficiency and task completion rate, and provides an efficient and reliable solution for the field of fixed-point maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a flow chart of a joint optimization method for equipment maintenance decision and task scheduling in a fixed-point maintenance mode provided by an embodiment of the present invention.

[0051] Figure 2 It is a structural block diagram of a joint optimization system for equipment maintenance decision and task scheduling in a fixed-point maintenance mode provided by an embodiment of the present invention.

[0052] Figure 3 It is a general flow chart of maintenance and guarantee activities provided by an embodiment of the present invention.

[0053] Figure 4 This is a spare parts replacement diagram provided by an embodiment of the present invention.

[0054] Figure 5 It is a target solving process diagram provided by an embodiment of the present invention.

[0055] Figure 6 It is a basic flow chart of the algorithm provided by the embodiment of the present invention.

[0056] Figure 7 It is a diagram of the encoding situation provided by an embodiment of the present invention.

[0057] Figure 8 It is a crossover operator graph provided by an embodiment of the present invention.

[0058] Fig. 9 It is a mutation operator graph provided by an embodiment of the present invention.

[0059] Fig.10 It is a box plot provided by an embodiment of the present invention.

[0060] Fig.11 It is a comparison diagram of the convergence speed of the algorithms provided by the embodiments of the present invention. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0062] The invention provides a method for joint optimization of equipment maintenance decision and task scheduling in a fixed-point maintenance mode, comprising the following steps:

[0063] (a) Comprehensively consider equipment maintenance under the fixed-point maintenance mode, solve the problems of diverse maintenance modes, complex maintenance constraints, and cumbersome maintenance decision-making, and consider factors such as multiple maintenance modes, time conflicts, number of spare parts, maintenance time and priority;

[0064] (b) Propose a replacement decision and combine it with a selective maintenance decision, using the secondary use time and priority of the maintenance equipment as the maintenance benefit, maximize the maintenance benefit within a limited time, optimize the number of maintenance equipment, and establish a mathematical model;

[0065] (c) Design appropriate encoding and decoding methods based on the particularity of the solutions in fixed-point maintenance;

[0066] (d) Design an adaptive hybrid particle swarm genetic algorithm for multi-objective optimization problems, and verify the effectiveness of the model and algorithm through simulation experiments and indicator evaluation;

[0067] (e) Optimize equipment maintenance decisions and task scheduling under the fixed-point maintenance mode.

[0068] The mathematical model includes:

[0069] (a) Objective function 1, which is used to represent the maintenance benefit generated by performing maintenance according to the maintenance decision. The benefit is composed of the secondary use time of the equipment and the equipment importance coefficient;

[0070] (b) Objective function 2, used to represent the maximum number of maintenance equipment;

[0071] (c) When the maintenance task is executed within the time window, the maintenance benefit will not be affected. Otherwise, a penalty factor is introduced to appropriately reduce the maintenance benefit.

[0072] (d) Spare parts constraints: ensuring the resource conditions required for equipment maintenance and introducing appropriate replacement strategies to change the number of spare parts;

[0073] (e) singularity constraint, which means that the maintenance task is performed only once or not at all;

[0074] (f) Serial switching constraints to ensure the non-repeatability of serial switching strategies;

[0075] (g) Maintenance deadline constraints are used to control the time limit of maintenance tasks.

[0076] The maintenance task set in the model includes the periodic maintenance task set, and the conversion process is as follows:

[0077] (a) Scheduled maintenance tasks are converted into maintenance waiting times after the remaining life is calculated, which are abstracted as time windows;

[0078] (b) Preventive maintenance tasks, after detecting the fault curve, are converted into time windows when the fault threshold is reached;

[0079] (c) Random failure maintenance task, the time of random equipment failure is unknown and converted into a time window.

[0080] The adaptive hybrid particle swarm genetic algorithm is used to optimize the maintenance task scheduling. The algorithm includes the following steps:

[0081] (a) Initial solution generation, maintenance task scheduling based on the coding scheme;

[0082] (b) Using the particle swarm optimization (PSO) part to search for the global optimal solution;

[0083] (c) Using genetic algorithm, crossover and mutation to optimize candidate solutions and ensure algorithm diversity;

[0084] (d) Adaptively adjust the search strategy to avoid falling into the local optimal solution;

[0085] (e) Optimize maintenance decisions under complex constraints and improve the efficiency of the entire maintenance process.

[0086] The optimization method verifies the effectiveness of the model and algorithm through simulation experiments and indicator evaluation. The simulation experiments include:

[0087] (a) Set up experimental scenarios with different maintenance modes, maintenance task constraints, and the number of maintenance equipment;

[0088] (b) Use multiple indicators to evaluate the optimization effect, including maintenance benefits, equipment maintenance quantity, task completion time, etc.;

[0089] (c) By comparing with traditional optimization methods, the advantages of the proposed optimization method in decision-making and task scheduling are verified to ensure its effectiveness.

[0090] like Figure 1 As shown, a method for joint optimization of equipment maintenance decision and task scheduling in a fixed-point maintenance mode provided by an embodiment of the present invention includes the following steps:

[0091] S101, equipment maintenance under the comprehensive fixed-point maintenance mode has many maintenance modes, complex maintenance constraints, and cumbersome maintenance decision-making situations, considering multiple maintenance modes, time conflicts, spare parts quantity, maintenance time, and priority factors;

[0092] S102, for the first time, proposes a replacement decision, combines it with a selective maintenance decision, combines the secondary use time of the maintenance equipment with the priority as the maintenance benefit, takes the maintenance benefit and the maximum number of maintenance equipment under a limited time as the goal, and establishes a more complete mathematical model;

[0093] S103, adopt corresponding encoding and decoding methods according to the particularity of the solution in fixed-point maintenance; design an adaptive hybrid particle swarm genetic algorithm for multi-objective optimization problems; verify the effectiveness of the model and algorithm by designing simulation experiments and indicator evaluation, and realize the optimization of equipment maintenance decision-making and task scheduling under the fixed-point maintenance mode.

[0094] The hardware and software equipment includes sensors, data acquisition modules and embedded control systems. Sensors are installed on maintenance equipment to monitor the equipment's operating status, fault type, spare parts usage and other data in real time, and transmit them to the central control system through the data acquisition module. The initialization module receives the equipment's real-time data and historical maintenance records, and integrates the multi-dimensional constraints under the fixed-point maintenance mode, including maintenance mode, time conflict, spare parts quantity, etc., to provide basic information for the establishment of subsequent decision-making models.

[0095] The system converts the equipment status parameters and maintenance requirements collected from the data into mathematical model inputs through the collaborative work of software and hardware. The model introduces the decision of replacing parts for the first time, and calculates the maintenance benefits by combining the secondary use time and priority of the equipment. At the same time, the model sets multi-dimensional constraints such as maintenance time, number of spare parts, and task priority to maximize the maintenance benefits and the number of maintenance equipment within a limited time. The model design takes into account the complex time conflict problem and the need to switch between multiple maintenance modes.

[0096] The hardware and software collaboration uses an adaptive hybrid particle swarm genetic algorithm as the core optimization algorithm. This algorithm combines the global search capability of the particle swarm and the local optimization capability of the genetic algorithm to dynamically adjust the search strategy for multi-objective optimization problems. At the hardware level, high-performance computing chips (such as FPGA or GPU) are used to accelerate the iteration process of the optimization algorithm and improve the solution efficiency. At the software level, the optimization algorithm decomposes the maintenance tasks into unit tasks through specific encoding and decoding methods, and dynamically schedules the maintenance tasks.

[0097] The hardware equipment includes an embedded scheduling controller and a data distribution system. The scheduling module dynamically allocates maintenance tasks to different maintenance points based on the results of the optimization algorithm output, and allocates corresponding spare parts and time resources. The scheduling controller updates the task execution progress in real time and dynamically adjusts the task allocation during the task execution process to ensure the maximum resource utilization and the stability of the maintenance plan. The task and resource allocation results are delivered to the on-site operators through the data distribution system.

[0098] Through the hardware simulation platform, the fixed-point maintenance scenario is simulated to verify the effectiveness of the model and algorithm. The simulation platform can simulate a variety of maintenance scenarios, including complex scenarios such as time conflicts and multi-task parallel execution. At the software level, the evaluation module analyzes the actual effect of the optimization algorithm through indicators such as maintenance benefits, equipment quantity, and task completion rate, and outputs a detailed evaluation report to provide a basis for subsequent improvements.

[0099] After the repair is completed, the data acquisition module records various data during the task execution, including repair time, spare part usage, task completion rate, etc. Through the data feedback mechanism, the system inputs the actual data into the optimization algorithm module for parameter correction and algorithm iteration. Through multiple optimizations, the efficiency and accuracy of the repair task scheduling are gradually improved, forming a dynamic adaptive joint optimization system.

[0100] Through the cooperation of software and hardware, integrating multi-dimensional data analysis, complex mathematical modeling, efficient algorithm optimization, and dynamic resource scheduling, this method realizes the joint optimization of equipment repair decision-making and task scheduling in the fixed-point repair mode, significantly improving the repair efficiency and task completion rate.

[0101] The mathematical model provided by the embodiments of the present invention:

[0102] The established model is as follows:

[0103]

[0104] Among them, Equation (1) is the objective function 1, representing the repair benefit generated by performing repairs according to the repair decision-making. The benefit consists of the secondary use time of the equipment and the equipment importance coefficient;

[0105] Equation (2) is the objective function 2, representing the maximum number of repaired equipment;

[0106] When the execution time of the repair task is within the time window, it does not affect the repair benefit; otherwise, introduce the penalty factor A to appropriately reduce the repair benefit;

[0107] Equation (4) is the spare part constraint, which is the necessary resource condition for repairing equipment, ensuring that repairs are carried out on the basis of sufficient spare parts; The strategy of substituting parts can be appropriately added to change the number of spare parts, and the decision is passed up to the repair time, affecting the final repair decision;

[0108] Equations (5)(6) are the uniqueness constraints, indicating that the repair is not executed or only executed once;

[0109] Equation (7) is the substituting parts constraint, ensuring the non-repeatability of the substituting parts strategy;

[0110] Equation (8) is the repair deadline constraint;

[0111] R 1 ~R 4 respectively represent the repair task sets of periodically repaired, regularly repaired, preventively repaired, and randomly failed repaired equipment; The repair task set R 2 ,R 3 ,R 4 can be abstracted as the periodically repaired task set R 1 , and the conversion process is:

[0112] 1)

[0113] Scheduled maintenance tasksx 3 ∈R 2 , after the calculation of the remaining life, it is converted into the maintenance time e 3 , it can be abstracted as a time window (-∞,e 3 );

[0114] 2)

[0115] Preventive maintenance tasksx 4 ∈R 3 , after detecting the fault curve, the time when the fault threshold σ is reached is e 4 , it can be abstracted as a time window (-∞,e 4 );

[0116] 3)

[0117] Random fault repair x 5 ∈R 4 , the time of random equipment failure is unknown, and can be abstracted as a time window (-∞,+∞).

[0118] like Figure 2 As shown, an embodiment of the present invention provides a joint optimization system for equipment maintenance decision and task scheduling in a fixed-point maintenance mode, including:

[0119] Comprehensive module, used for equipment maintenance under the comprehensive fixed-point maintenance mode, which has many maintenance modes, complex maintenance constraints, and cumbersome maintenance decision-making situations, taking into account multiple maintenance modes, time conflicts, spare parts quantity, maintenance time, and priority factors;

[0120] The mathematical model building module is used to propose the replacement decision for the first time, combine it with the selective maintenance decision, combine the secondary use time of the maintenance equipment with the priority as the maintenance benefit, and establish a more complete mathematical model with the maintenance benefit and the maximum number of maintenance equipment under limited time as the goal;

[0121] The scheduling optimization module is used to adopt corresponding encoding and decoding methods according to the particularity of the solution in fixed-point maintenance; an adaptive hybrid particle swarm genetic algorithm is designed for multi-objective optimization problems; the effectiveness of the model and algorithm is verified by designing simulation experiments and indicator evaluation, and the equipment maintenance decision-making and task scheduling optimization under the fixed-point maintenance mode are realized.

[0122] The present invention is specifically implemented:

[0123] 1 Basic Description

[0124] 1.1 Problem Description

[0125] When equipment breaks down suddenly during a mission, the remaining life of parts is insufficient, or the fixed service life is reached, it needs to be sent to a maintenance center for fixed-point maintenance. Due to differences in equipment attributes, failure modes, and failure levels, combined with the characteristics of the equipment, maintenance modes are divided into periodic maintenance, fixed-point maintenance, preventive maintenance, and random failure maintenance. Before performing maintenance support activities, it is necessary to consider maintenance decisions for different equipment, determine maintenance modes, and formulate maintenance plans, thereby forming a queue of equipment to be repaired. For the maintenance queue, it is crucial to fully consider the relevant constraints in the maintenance process, reasonably allocate maintenance resources, and scientifically schedule tasks in order to quickly and effectively repair equipment and restore the equipment's ability to perform tasks. This is an urgent need for fixed-point maintenance task scheduling.

[0126] The present invention mainly studies the following problems: The equipment maintenance decision and task scheduling optimization problem under the fixed-point maintenance mode consists of the decision and optimization of maintenance sequences of different maintenance categories. First, after the equipment in the equipment library to be used completes the corresponding task, a maintenance task sequence to be processed is formed. The maintenance mode of the equipment to be maintained in the maintenance sequence consists of four complex maintenance modes: periodic maintenance, scheduled maintenance, preventive maintenance, and random fault maintenance. Among them, periodic maintenance takes into account the periodic nature of the equipment and ensures that the equipment is maintained within the pre-set time window. Scheduled maintenance takes into account the deadline for equipment maintenance and performs maintenance before the deadline. Preventive maintenance monitors the remaining life of the equipment according to the equipment health management system (Prognostics and Health Management, PHM)

[10] and executes the maintenance plan after reaching a fixed maintenance threshold. Random faults occur in the entire decision-making and scheduling process and are randomly added to the maintenance sequence to ensure that the system has the ability to handle random faults. The key to solving the problem is to take advantage of the different maintenance mode characteristics of different equipment and adopt different maintenance decisions and task scheduling.

[0127] For equipment in a certain maintenance mode, to ensure that the number of spare parts and time are limited and conflicting, maintenance strategies such as maintenance timing, maintenance decision, and replacement parts are considered to construct model constraints and decision variables. Multiple fault characteristics in each maintenance task are converted into maintenance urgency and maintenance time to further determine the maintenance priority. The objective function is constructed with the dual optimization goals of maximizing the number of equipment maintenance and maximizing the maintenance benefits in the maintenance sequence.

[0128] Combining the above model constraints and decision variables, a joint optimization model of maintenance decision and task scheduling is constructed. The model is solved and optimized by the designed intelligent optimization algorithm NSPSO-AD. After the optimization result is obtained, a maintenance plan is formulated, and the equipment sequence after maintenance is returned to the equipment library to be used, realizing the integration of equipment use, decision-making, optimization and maintenance. The overall process of maintenance support activities is as follows: Figure 3.

[0129] 1.2 Problem Assumptions

[0130] 1) It is assumed that before carrying out equipment maintenance support activities, the maintenance time, spare parts, maintenance mode, etc. required for the maintenance equipment have been given and will not change during the decision-making and scheduling process.

[0131] 2) Assume that all the equipment to be repaired in a maintenance queue belongs to the same equipment category and the faulty parts can be replaced with each other.

[0132] 3) Except for some spare parts that need to be replaced, other maintenance resources are sufficient.

[0133] 4) It does not consider the decision-making and optimization differences caused by external environmental factors such as weather conditions, maintenance personnel, and maintenance technology during the maintenance process.

[0134] 2 Model Construction

[0135] 2.1 Parameter Description

[0136] In order to effectively solve the equipment maintenance decision-making and task scheduling optimization problem under the fixed-point maintenance mode, this section uses mathematical language to abstract and quantify the various influencing factors, providing a basis for subsequent model establishment and solution. The symbols are shown in Table 1.

[0137] Table 1 Symbols

[0138]

[0139] 2.2 Establishing maintenance decision-making and scheduling optimization model

[0140] Based on the above analysis, the model is established as follows:

[0141]

[0142] Among them, formula (1) is objective function 1, which represents the maintenance benefit generated by maintenance according to the maintenance decision. The benefit is composed of the secondary use time of the equipment and the equipment importance coefficient. Formula (2) is objective function 2, which represents the maximum number of maintenance equipment. Formula (3) When the maintenance task execution time is within the time window, it does not affect the maintenance benefit. On the contrary, the introduction of penalty factor A appropriately reduces the maintenance benefit. Formula (4) is the spare parts constraint, which is the necessary resource condition for maintenance equipment, ensuring that maintenance is carried out on the basis of sufficient spare parts. The replacement strategy can be appropriately added to change the number of spare parts, and the decision is passed up to the maintenance timing, affecting the final maintenance decision. Formula (5) (6) are singleness constraints, indicating that the maintenance is not performed or only performed once. Formula (7) is the replacement constraint, which ensures the non-repeatability of the replacement strategy. Formula (8) is the maintenance deadline constraint.

[0143] In addition, different maintenance task sets can be converted to simplify model building and optimization. 1 ~R 4 They represent the maintenance task sets of periodic maintenance, scheduled maintenance, preventive maintenance, and random failure maintenance equipment. Maintenance task set R 2 ,R 3 ,R 4 It can be abstracted into a set of periodic maintenance tasks R 1 , the transformation process is:

[0144] 1)

[0145] Scheduled maintenance tasksx 3 ∈R 2 , after the calculation of the remaining life, it is converted into the maintenance time e 3 , it can be abstracted as a time window (-∞,e 3 ).

[0146] 2)

[0147] Preventive maintenance tasksx 4 ∈R 3 , after detecting the fault curve, the time when the fault threshold σ is reached is e 4 , it can be abstracted as a time window (-∞,e 4 ).

[0148] 3)

[0149] Random fault repair x 5 ∈R 4 , the time of random equipment failure is unknown, and can be abstracted as a time window (-∞,+∞).

[0150] Based on the above analysis, the problem is transformed into a fixed-point maintenance problem with time windows.

[0151] 3 Design of improved adaptive hybrid particle swarm genetic algorithm

[0152] 3.1 Algorithm Analysis

[0153] After completing the analysis and modeling of the above problems, it is necessary to refine the process of solving the target value of the maintenance sequence. When there are sufficient spare parts in the maintenance queue, sufficient spare parts after the replacement strategy, and random failures occur with a certain probability during the entire maintenance process, the maintenance benefit solution process can be broken down into the following sub-situations:

[0154] Case 1: Sufficient spare parts and no need to replace parts

[0155] For a maintenance queue {…,r,i,…}, the maintenance equipment r→i, then the time update is: Spare parts update: Objective function value update:

[0156] Case 2: Spare parts are insufficient, but sufficient spare parts are available after replacement

[0157] For the maintenance queue {…,r,i,…,j,…}, the maintenance equipment r→i and the parts are replaced, then the time is updated: Spare parts update: θ j =θ j -ρ r (τ ri ); objective function value update:

[0158] The replacement diagram of spare parts is as follows Figure 4 :

[0159] Case 3: Equipment does not perform preventive maintenance (does not exceed the fixed threshold σ)

[0160] For the maintenance queue {…,r,i}, if the maintenance equipment r→i, then the time, spare parts and target values ​​are not updated.

[0161] Case 4: Equipment performs preventive maintenance (exceeds a fixed threshold σ)

[0162] For a maintenance queue {…,r,i}, the maintenance equipment r→i, then the time update is: Spare parts update: Objective function value update:

[0163] Case 5: Repair within the time window

[0164] For a maintenance queue {…,r,i,…}, the maintenance equipment r→i, then the time update is: Spare parts update: Objective function value update:

[0165] Case 6: Repair outside the time window before the deadline

[0166] For a maintenance queue {…,r,i,…}, the maintenance equipment r→i, then the time update is: Spare parts update: Objective function value update:

[0167]

[0168] Case 7: Insert equipment to be repaired after random failure

[0169] For the maintenance queue {…,r,i,…}, if equipment k is inserted, the maintenance sequence is updated to {…,r,i,…,k}.

[0170] By analyzing and decomposing the entire maintenance benefit solution process in detail, we provide directions for model solution, algorithm design and optimization.

[0171] Based on the above, we first generate an initial population according to the rules to obtain a maintenance sequence that satisfies the constraints and the corresponding target value, and then optimize the objective function to obtain the optimal maintenance sequence. We then perform decision making and optimization after solving the target value through the initial solution set of the maintenance queue. Figure 5 .

[0172] The specific steps of the entire maintenance decision algorithm are as follows:

[0173] Step 1: According to the maintenance decision of the equipment, an initial maintenance queue with different failure modes and maintenance task priorities is obtained.

[0174] Step 2 determines whether all maintenance equipment in the maintenance sequence meets the maintenance constraints. If not, jump to step 1 to continuously obtain the initialization queue that meets the maintenance constraints.

[0175] Step 3: Make special maintenance strategies, such as: part replacement strategy, maintenance selection strategy.

[0176] Step 4: Calculate the maintenance benefit generated by the entire maintenance according to the maintenance decision and maintenance sequence order, and take the maintenance benefit as the optimization target value.

[0177] The entire maintenance sequence initialization process and the process of obtaining the maintenance benefit target value by considering various decisions through the maintenance sequence can be summarized in detail as Algorithm 1.

[0178]

[0179] After obtaining the initialized maintenance sequence and initializing the solution to the target through maintenance decision, it is necessary to use an intelligent optimization algorithm to optimize the maintenance sequence.

[0180] The task decision and scheduling optimization problem under the multi-maintenance mode proposed in this invention belongs to the Np-hard problem. Compared with the multi-objective optimization problems in other studies, the situation is complex and the solution set is discrete, which is difficult to handle using conventional multi-objective optimization methods.

[0181] Similar to other maintenance and support related problems, after modeling the scheduling optimization problem, the model is solved by combining intelligent optimization algorithms such as NSGA-Ⅱ[11-12] and MOPSO[13-14]. Genetic algorithms are used to select the individuals with the highest fitness in the population, and crossover and mutation are continuously performed to select excellent offspring and combine with part of the parent generation to form a new solution. The advantage is that it does not destroy the optimal solution of the parent population and improves global convergence.

[0182] The entire maintenance scheduling optimization algorithm process is:

[0183] Step 1: Use Algorithm 1 to solve the target values ​​corresponding to the initial maintenance queue set and all maintenance sequences in the maintenance queue set.

[0184] Step 2 sorts the maintenance sequence according to the non-dominated algorithm and selects individuals with better survivability as parents.

[0185] Step 3 uses operators such as crossover, mutation, and selection on the sorted maintenance sequence to generate offspring.

[0186] Step 4 determines whether the number of iterations is met, if not, jumps to step 2.

[0187] Step 5 outputs the optimized maintenance sequence and determines the maintenance timing and strategy of all equipment in the maintenance sequence.

[0188] The whole process is summarized as Algorithm 2.

[0189]

[0190]

[0191] In order to further improve the algorithm's search speed, scope and quality for solutions, the present invention improves the algorithm in the above process and designs the NSPSO-AD algorithm. The NSPSO-AD algorithm draws on the advantages of the genetic operators of the NSGA-Ⅱ algorithm, designs and improves operators that meet the model solving requirements, performs a global search on the solution set, and improves the global search capability of the algorithm. Combined with the search characteristics of the MOPSO algorithm for the solution set, as the number of iterations increases, when it is close enough to the optimal solution set, the search range is reduced to improve the local convergence characteristics of the algorithm. According to the relationship between the model and the solution set, new encoding and decoding methods and new operators are designed to adapt to the improved algorithm. Finally, the offspring are adaptively selected according to the number of iterations to improve the overall optimization performance of the algorithm.

[0192] The analysis of the algorithm provides a direction for solving and optimizing the maintenance decision and scheduling problem model. The maintenance decision can get the target value through the initial solution. Taking the target value as the optimization goal, the optimization is continuously scheduled and finally iterated to the optimal solution. The whole process of joint optimization of maintenance decision and task scheduling is shown in Figure 6。

[0193] 3.2 Encoding and Decoding

[0194] To improve the convergence speed and quality of the algorithm, new encoding and decoding methods need to be designed. For encoding, decisions such as the maintenance timing, maintenance decision, swapping strategy, and random failure maintenance timing of the equipment sequence play a decisive role in the change of maintenance benefits. Therefore, the present invention adopts a four - segment encoding, which is respectively used to represent the maintenance timing, maintenance decision, swapping strategy, and random failure maintenance strategy. The entire encoding situation is shown in Figure 7 。Since the solution set is discrete and the process is complex during the solution process, using real - number encoding is more in line with the actual requirements. Therefore, the first - segment encoding adopts sequential encoding (1 to N, where N represents the total number of equipment to be maintained in the maintenance sequence), representing the maintenance order; the second - segment encoding adopts integer encoding (0 or 1), representing the decision whether to perform maintenance; the third - segment encoding adopts integer encoding (0 or 1 to N), representing whether to execute the swapping part strategy and the timing of swapping parts. A positive integer indicates that this equipment is swapped with the equipment recording this positive integer; 0 indicates not to execute the swapping strategy. The fourth - segment encoding is integer encoding, representing the execution timing and order of the equipment with randomly occurring failures.

[0195] Due to the discrete solution set, no additional operations are required after using real - number encoding for operator operations, which is convenient for optimizing the solution process of the entire objective. However, during the process from the initial solution set to the solution of the target value, using integers for calculation is more in line with the requirements. Therefore, for decoding, it is necessary to convert real numbers into integers for processing according to the corresponding relationship.

[0196] Take Figure 7Taking the maintenance sequence in as an example, the specific decoding scheme is as follows: the first layer represents the maintenance order, and the maintenance sequence obtained by decoding is: {7,3,8,5,2,12,1,4,9,6,11,10}. The maintenance is carried out in order according to the maintenance sequence, and the maintenance order is 7-3-8-5-2-12-1-4-9-6-11-10. Combined with the maintenance decision of the second layer and the decision of the maintenance threshold, equipment 6 and 11 are not selected for maintenance, and the final maintenance order is: 7-3-8-5-2-12-1-4-9-10. The third layer represents the replacement strategy. After decoding, it can be seen that the equipment set of replacement parts is {(1,3), (5,6)}, which means that two replacement strategies appear in the whole process to meet the spare parts quantity constraint required for maintenance. The spare parts of equipment 1 need to be replaced with those of equipment 3, that is, the initial spare parts quantity of equipment 3 is reduced to fill the spare parts required for repairing equipment 1. After the maintenance is completed, the remaining spare parts are returned to the spare parts center. The whole process of replacing equipment 5 and 6 is the same as above. The fourth layer indicates that after the random failure occurs, it is added to the maintenance sequence. The maintenance occurs after equipment 5 and equipment 9, and the final maintenance sequence is: 7-3-8-5-x1-2-12-1-4-9-x2-10 (x1 and x2 are newly added equipment).

[0197] 3.3 Algorithm Improvement Plan

[0198] Chromosome is the basic unit of algorithm solution and optimization. Designing different operators to process the update iteration of chromosome is of great significance to algorithm solution and optimization. 4 layers of encoding and decoding correspond to 4 different segments of chromosome. Since different segments of chromosome represent different meanings, the same crossover, mutation and selection operators cannot be used. Therefore, the present invention adopts multi-mode and segmented genetic operator design. Among them, multi-mode is to combine genetic operators with optimization operators in particle swarms to optimize the algorithm in global and local optimization capabilities. Segmented is to use different operators for 4 segments of chromosome according to different characteristics.

[0199] According to the coding rules, the individuals with N equipment to be repaired in the maintenance sequence are initialized. The first-layer chromosome encoding of individual i is The second-level chromosome is encoded as The third level chromosome is encoded as The 4th level chromosome is encoded as The design details of the operator of the present invention are as follows:

[0200] like Figure 8 As shown, 1) crossover operator

[0201] Since the first layer of chromosomes needs to ensure unique coding rules, that is, integers 1 to N must exist and be unique, this ensures that each equipment has been repaired and can only be repaired once. Since the first layers of two chromosomes are cross-linked at a single point or multiple points, it is difficult for the chromosomes after cross-linking to meet the coding rules. Therefore, the present invention designs a random reversal crossover operator for single chromosomes. Taking individual i in the population as an example, the first layer chromosome is encoded as Generate a random number of rand(N) = k, and reverse and cross the chromosome part before the random number with the chromosome part after the random number to generate a new solution p that meets the coding requirements. 1 ' i .

[0202]

[0203] The second layer performs the crossover operator. The chromosomes after the crossover still meet the coding rules, and a single-point crossover can be used. For the crossover operator, the single-point random crossover operator of the genetic algorithm is combined with the intermediate recombination operator. Taking individuals i and j in the population as an example, and Perform a single-point random crossover to generate a random number rand(N) = k, perform a single-point crossover on the chromosome, and generate a new solution p that meets the coding requirements 2 ' i With p 2 ' j .

[0204]

[0205] remember Chromosome encoding for individuals waiting for crossover, adaptively select another parent chromosome encoding V from the Pareto dominated set with adaptive probability P = 0.8 (1-g / G) + 0.2 as the number of iterations g∈{1,2,3,…,G} increases father2 According to formula (9), the intermediate reorganization is performed to improve the global search capability in the early stage of the algorithm and the local optimization capability in the later stage of the algorithm. Where α∈[-1,1] is uniformly distributed.

[0206] V son =V father1 +α(V father2 -V father1 )(9)

[0207] Since the third and fourth layer coding rules cannot maintain their special coding characteristics after cross-processing, no cross-operation is performed.

[0208] 2) Mutation Operator

[0209] The first layer of chromosomes does not adopt mutation operations according to the coding rules. For the second layer of coding, a certain position is randomly selected to modify the state of the maintenance decision. For the third layer of coding, the cross-talk strategy adopts single-point mutation and double-point exchange operations.

[0210] Three-layer coding and It indicates that the spare parts required for repairing equipment r are insufficient, and a replacement strategy needs to be implemented to find equipment with serial numbers between r and N to perform spare parts replacement. Single-point mutation refers to generating a random number rand(r~N)=k in the range of r~N, determining the serial number k of the replaced equipment, and providing spare parts support for equipment r. Double-point exchange refers to exchanging the serial numbers r and k of the equipment that need to be replaced in the encoding, and finally completing the mutation operator of the three-layer encoding.

[0211] the following Fig. 9 For example, the 5th position of the three-layer code is a non-negative number, indicating that the spare parts required for maintenance equipment 5 are insufficient and a serial exchange operation is performed. Subsequently, a non-negative number 9 between 5 and 12 is generated as the variation of position 5, completing a single-point variation. Finally, the number in position 5 in the code is replaced with 9, and the number in position 9 is replaced with 5, completing a double-point exchange.

[0212] 3) Select Operator

[0213] The purpose of the selection operator is to determine whether the individuals in the population are good, select individuals with strong survival ability from the current population to participate in the evolution, and form the next generation of the population with some parent individuals to approach the optimal solution. The present invention refers to NSGA-Ⅱ to use non-dominated rank and crowding strategy to quickly sort the rules, obtain the Pareto solution set by sorting the encoding set and the corresponding target value, calculate the fitness by the dominance rank and crowding of the solution set through formula (10), use formula (11) to calculate the selection probability, and select the offspring by roulette.

[0214]

[0215]

[0216] 4) Boundary processing

[0217] Each time the chromosome performs an operator operation, it generates a solution that does not meet the coding requirements and needs to be improved. For solutions that exceed the coding range, the coding is reinitialized to improve the robustness of the algorithm for different operator operations.

[0218] According to the above designed operators, the present invention improves Algorithm 2 based on NSGA-Ⅱ and MOPSO, and proposes an improved adaptive hybrid particle swarm genetic algorithm. The specific steps are as follows:

[0219] Step 1: Initialize the solution set. Randomly initialize the maintenance sequence, maintenance decision, replacement strategy, and random failure strategy to meet the coding rules designed by the present invention to form an initialization set res_set;

[0220] Step 2: Crossover and mutation. The selected individuals are crossovered and mutated according to the improved operator scheme designed by the present invention to form a descendant set son, and the solution set res_set is updated to res_set = [res_set; son];

[0221] Step 3: Calculate the fitness. Calculate the target value according to Algorithm 1, and calculate the fitness according to the fitness calculation formula (10);

[0222] Step 4: Non-dominated sorting. Calculate the dominance level and crowding degree according to the fitness, and calculate the individuals selected for operator calculation;

[0223] Step 5: Selection. According to the non-dominated sorting and elite retention strategy, the inferior solutions are eliminated, the solution set size is maintained, and a new set res_set = Rank is formed. Non_dominated [1:N];

[0224] Step 6 determines whether the iteration termination condition is met; if the termination condition is not met, jump to step 2; if the termination condition is met, output the optimization result.

[0225] 4 Experimental simulation

[0226] The research of this invention is intended to be applied to the problem of fixed-point fault UAV equipment maintenance strategy and scheduling in a certain UAV maintenance unit. The equipment maintenance problem of UAV considers four fault maintenance modes. The maintenance time, time window length, and maintenance deadline of each equipment are taken from reference [2]. The method of reference

[15] sets priorities for different equipment, and dynamically adjusts the priority according to the difference between the latest maintenance time of the faulty equipment and the current time to meet the real engineering needs.

[0227] The preventive maintenance threshold is σ = 0.2. Referring to the health status of equipment in the PHM system in real engineering scenarios, the curve F = e -0.007t Used to simulate the remaining life of preventive maintenance equipment. Set p = 0.05 as the probability of random failure to simulate and verify the algorithm's ability to handle dynamic maintenance tasks. For the parameter setting of the switching strategy, if the parameter setting is too large, frequent switching will affect the maintenance efficiency of the equipment. Therefore, set the probability of switching occurrence p change =0.3.

[0228] Due to the lack of research on the initial number of spare parts, spare parts required for maintenance, the number of spare parts warehouses and other related information in the relevant literature, the present invention generates relevant data through investigation and consultation with relevant departments, and ensures the rationality of the data. The initial spare parts of the equipment are set to 10, the number of spare parts required for maintenance of each piece of equipment is set to 20, and the number of spare parts warehouses is [20, 40, 60, 80, 100], which are the numbers of spare parts in five average time periods respectively.

[0229] In order to ensure the universality of the model and algorithm for data, the method of reference

[16] was used to randomly generate 10 sets of data case1-case10

[17] through uniform distribution. Since the magnitude of the optimization objectives is different, normalization is performed to facilitate intuitive comparison of the solution and optimization results.

[0230] In addition, with respect to the relevant parameter settings of the algorithms, all algorithms involved in the present invention set the initial population size to 100, the number of iterations to 500, and the mutation probability to 0.1.

[0231] The experiment was conducted with case 1 data to verify the scientific nature of the model and algorithm proposed in the present invention. The results are shown in Tables 1-5. In order to verify the influence of the switching strategy proposed in the present invention on problem solving, a comparative experiment of switching and non-switching without random failure was set up. Tables 1 and 2 show the results of switching without random failure, and Table 3 shows the results of non-switching under the same conditions.

[0232] At the same time, in order to verify the model's ability to dynamically handle faulty equipment, a comparative experiment was set up to determine whether random failure modes occurred under the same conditions. The results are shown in Tables 4 and 5.

[0233] Table 2 Maintenance results (no random failure and replacement)

[0234]

[0235] Table 3 Conversion results

[0236]

[0237] Note: Target value is: 800,659.

[0238] Table 4 Repair results without random failure and non-sequential replacement

[0239]

[0240]

[0241] Note: Target value is: 600,489.

[0242] Table 5 Maintenance results (random failure and replacement)

[0243]

[0244] Table 6 Conversion results

[0245]

[0246] Note: Target values ​​are: 600, 402.

[0247] In order to verify the improvement effect of the algorithm proposed in the present invention, the present invention sets up a comparative experiment, and uses uniform distribution to generate case1-case10 data. The rest of the data and parameter settings are consistent with the above experiment. 100 experiments are conducted on NSGA-Ⅱ, MOPSO, NSPSO, and NSPSO-AD algorithms respectively, and the optimal value is taken to eliminate the influence of the initial distribution of the solution on the algorithm. In order to facilitate the setting of comparative experiments, the process does not involve the failure mode of random failures. The performance of the algorithm is verified by solving the target value size. The results are shown in Table 5. Fig.10 The box plot in the figure shows the algorithm solution and optimization results intuitively. Finally, taking case 1 as a reference, the convergence speed comparison of the algorithm is obtained. The results are as follows Fig.11 .

[0248] Table 7 Average optimal solution (×10 2 )

[0249]

[0250] Table 8 Convergence times to the optimal solution

[0251]

[0252] Tables 2 and 4 give the maintenance decisions and optimization results obtained in the maintenance sequence without considering the random fault mode. The corresponding target values ​​are given in the table notes. From the results, the encoding and decoding methods proposed in the algorithm can ensure that the model constraints are met while continuously searching, and finally solve the solution set. The substitution strategy significantly improves the solution quality of the model. The data in Table 4 show that before the execution of the substitution decision, due to the lack of spare parts for some equipment during maintenance, the maintenance of this equipment can only be abandoned. Table 3 gives the substitution results. Before the maintenance of equipment 1 and equipment 9, due to the lack of spare parts, compared with Table 3, some equipment does not perform maintenance, resulting in a window period for spare parts or ignoring the remaining spare parts after the maintenance. In contrast, the NSPSO-AD algorithm gives a substitution strategy, which supplements the relevant spare parts for maintenance, meets the maintenance needs, and improves the maintenance capacity of the fixed-point maintenance center. As can be seen from Tables 5 and 6, the algorithm can dynamically handle random faults in the maintenance sequence and has the ability to dynamically handle maintenance equipment.

[0253] It can be seen from Table 7 that for the comparative experiments conducted on case 1-case 10, the solution sets obtained by the NSPSO-AD algorithm proposed in the present invention are better than those of other algorithms under the same conditions. Fig.10 The box plot of the target values ​​is given, and it can be seen intuitively that the objective function values ​​of target 1 and target 2 obtained by other algorithms are inferior to those of the NSPSO-AD algorithm, and the quality and stability of the solution set searched by the NSPSO-AD algorithm are significantly improved.

[0254] Table 8 gives the average convergence speed comparison of different algorithms to the Pareto optimal solution. Fig.11 , the curve quickly reaches stability and the smaller the area of ​​the solution set, the faster the dual-objective optimization converges and the better the quality of the solution set. Fig.11 It can be seen that compared with other algorithms, the NSPSO-AD algorithm is not easy to fall into the local optimal solution. While ensuring the convergence speed, the global optimization ability in the early stage and the local optimization ability in the later stage of iteration are strong.

[0255] It can be seen from the simulation results that, compared with the commonly used intelligent optimization algorithms, the NSPSO-AD algorithm designed in the present invention is applied to the maintenance decision-making and task scheduling problems of fixed-point maintenance, and the quality of the solution set and the convergence speed are significantly improved. Among them, through the analysis of the problem and the model, the encoding and decoding methods are designed to ensure the mapping relationship between the solution set and the actual needs. Combining the advantages and disadvantages of various algorithms, crossover, mutation, and selection operators are designed to ensure the search capability of the solution set. In summary, the present invention, through the setting of comparison and verification experiments, demonstrates the improvement of the NSPSO-AD algorithm in the early stage of iteration fast global search and local optimal search capabilities in encoding and decoding design, operator design, and adaptive design, and verifies the scientific nature of the model and algorithm proposed in the present invention.

[0256] The present invention studies the optimization problem of maintenance decision and task scheduling under the fixed-point maintenance mode, comprehensively considers factors such as multiple fault modes, time windows, number of spare parts, maintenance priority, selective maintenance, and serial replacement parts, takes the maintenance quantity and maintenance benefit maximization as dual optimization goals, and establishes relevant mathematical models; analyzes the entire algorithm flow, designs corresponding encoding and decoding methods according to the problem, combines the advantages of commonly used intelligent optimization algorithms, designs relevant operators, and proposes an improved NSPSO-AD algorithm; sets relevant parameters, designs comparative experiments based on the above content, and verifies the scientific nature of the model and algorithm proposed by the present invention by comparing the solution results and convergence speed of the algorithm for the model, and the optimization problem of maintenance decision and task scheduling under the fixed-point maintenance mode is well solved. The research results show that the above model and algorithm can assist the optimization of maintenance decision and task scheduling under the fixed-point maintenance mode, and can be effectively applied to practical engineering.

[0257] In addition, combining the analysis of the present invention and related research, the next key work focuses on the issues of improving the efficient and available maintenance support of high-tech equipment such as unmanned aerial vehicles and unmanned vehicles by integrating artificial intelligence during the mission execution process in complex battlefield environments.

[0258] Embodiment 1: Equipment Maintenance and Scheduling Optimization of Power Transmission and Distribution Systems

[0259] In the power transmission and distribution industry, the regular maintenance of equipment and the handling of sudden failures are crucial for the stability of power supply. Power companies need to perform preventive maintenance before equipment failures and ensure timely response when equipment fails. The joint optimization method of equipment maintenance decision-making and task scheduling in the fixed-point maintenance mode can effectively optimize the allocation of maintenance resources and ensure the efficient operation of power equipment.

[0260] 1. Maintenance Decision-making: Combining multiple maintenance modes (such as preventive maintenance, time-based maintenance, breakdown maintenance, etc.), generate maintenance decisions based on factors such as the operating status, service life, and priority of the equipment to ensure the timely maintenance of important equipment.

[0261] 2. Task Scheduling: Optimize the scheduling of maintenance tasks through an adaptive hybrid particle swarm genetic algorithm, balance the maintenance time, spare part usage, and arrangement of maintenance personnel, avoid excessive extension of equipment downtime, and reduce maintenance time and costs.

[0262] 3. Simulation and Evaluation: Conduct simulation tests to evaluate the effectiveness of the optimization scheme under different working conditions, and verify the effectiveness of the optimization results through multi-objective evaluation (maintenance cost, downtime, maintenance revenue, etc.).

[0263] This method can help power companies achieve the optimal allocation of resources in equipment maintenance task scheduling, reduce the downtime caused by equipment failures, improve equipment utilization efficiency, and at the same time reduce maintenance costs.

[0264] Embodiment 2: Regular Maintenance and Task Scheduling Optimization of Airline Fleets

[0265] The maintenance management of airline fleets is a complex process that needs to handle various types of maintenance tasks, such as regular inspections, fault repairs, and emergency repairs. The joint optimization of equipment maintenance decision-making and task scheduling can significantly improve the maintenance efficiency of the fleet and ensure the on-time departure and safe operation of flights.

[0266] 1. Maintenance Decision-making: Combine factors such as the aircraft model, operating duration, and maintenance history in the fleet to make maintenance decisions and determine the specific maintenance tasks for each aircraft. Through the decision-making strategy of component substitution, optimize the use of spare parts and avoid maintenance delays caused by insufficient spare parts.

[0267] 2. Task Scheduling: Use the adaptive hybrid particle swarm genetic algorithm to optimize the scheduling of maintenance tasks to ensure the rational allocation of maintenance personnel and equipment resources. Consider factors such as aircraft priority and maintenance task timing to optimize maintenance schedules, reduce downtime, and ensure fleet operation efficiency.

[0268] 3. Simulation and evaluation: Conduct multiple simulation experiments to simulate the aircraft's maintenance cycle and failure scenarios, and verify the effectiveness of the optimization plan through evaluation indicators (such as maintenance time, resource utilization, cost, etc.).

[0269] By adopting the maintenance decision-making and task scheduling optimization method under the fixed-point maintenance model, airlines have improved the maintenance efficiency of their fleets, shortened aircraft maintenance time, reduced maintenance costs, and ensured flight punctuality and flight safety.

[0270] These two embodiments demonstrate how to apply the joint optimization method of equipment maintenance decision and task scheduling under the fixed-point maintenance mode to different industrial fields, so as to optimize equipment management and task scheduling, improve efficiency and reduce resource waste.

[0271] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. It can be understood by a person of ordinary skill in the art that the above-mentioned devices and methods can be implemented using computer executable instructions and / or contained in a processor control code, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. Such code is provided on the carrier medium. The device and its modules of the present invention can be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, and can also be implemented by a combination of the above-mentioned hardware circuits and software, such as firmware.

[0272] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principle of the present invention should be covered by the protection scope of the present invention.

Claims

1. A joint optimization system for equipment maintenance decision and task scheduling in a fixed-point maintenance mode, characterized in that: Contains the following modules: (a) Maintenance decision module, which is used to analyze equipment maintenance decisions by comprehensively considering multiple maintenance modes, time conflicts, spare parts quantity, maintenance time, priority, etc.; (b) A parts replacement decision module, which is used to propose a parts replacement decision and combine it with the selective maintenance decision to optimize the maintenance benefit and maximize the maintenance benefit and the number of maintenance equipment within a limited time; (c) Task scheduling module, used to optimize maintenance task scheduling through adaptive hybrid particle swarm genetic algorithm and verify its effect; (d) Simulation and evaluation module, which is used to verify the effectiveness of system models and algorithms through simulation experiments and indicator evaluation, and ensure the efficiency of maintenance decisions and task scheduling.

2. The equipment maintenance decision and task scheduling joint optimization system in the fixed-point maintenance mode according to claim 1 is characterized in that: The maintenance decision module includes: (a) Maintenance task analysis unit, used to analyze various maintenance modes and constraints and build a maintenance decision model; (b) An objective function optimization unit is used to optimize the benefits of maintenance tasks and maximize the number of maintenance equipment and maintenance revenue according to the secondary use time and priority of maintenance equipment.

3. The equipment maintenance decision and task scheduling joint optimization system in the fixed-point maintenance mode according to claim 1 is characterized in that: The serial replacement decision module includes: (a) a replacement parts selection unit, which is used to decide whether to adopt a replacement parts strategy and adjust the number of spare parts according to the actual situation of the maintenance task; (b) A spare parts scheduling unit, which is used to adjust the timing of maintenance tasks according to the replacement decision and ensure the adequacy of spare parts.

4. The equipment maintenance decision and task scheduling joint optimization system in the fixed-point maintenance mode according to claim 1 is characterized in that: The task scheduling module includes: (a) An adaptive optimization unit, which is used to optimize the maintenance task scheduling scheme by combining the particle swarm optimization algorithm and the genetic algorithm; (b) Multi-objective optimization unit, which is used to deal with multi-objective optimization problems in fixed-point maintenance, including constraints such as maintenance time, task priority, and maintenance resources.

5. The equipment maintenance decision and task scheduling joint optimization system in the fixed-point maintenance mode according to claim 1 is characterized in that: The simulation and evaluation module includes: (a) Experimental simulation unit, used to simulate the maintenance task scenario in the fixed-point maintenance mode and perform simulation tests on the maintenance tasks; (b) Index evaluation unit, which is used to evaluate the system optimization effect based on indicators such as maintenance benefits, number of maintenance equipment, and task completion time to ensure its effectiveness in practical applications.

6. A joint optimization method for equipment maintenance decision and task scheduling in a fixed-point maintenance mode, characterized in that: The following steps are involved: Step 1: Equipment maintenance under the integrated fixed-point maintenance mode has many maintenance modes, complex maintenance constraints, and cumbersome maintenance decision-making situations. Multiple maintenance modes, time conflicts, spare parts quantity, maintenance time, and priority factors are considered; Step 2: For the first time, the decision of serial replacement is proposed, combined with the selective maintenance decision, and the secondary use time of the maintenance equipment is combined with the priority as the maintenance benefit. The maintenance benefit and the maximum number of maintenance equipment under limited time are taken as the goal to establish a more complete mathematical model; Step 3: Adopt corresponding encoding and decoding methods according to the particularity of the solution in fixed-point maintenance; design an adaptive hybrid particle swarm genetic algorithm for multi-objective optimization problems; verify the effectiveness of the model and algorithm by designing simulation experiments and indicator evaluation, and realize the optimization of equipment maintenance decision-making and task scheduling under the fixed-point maintenance mode.

7. The joint optimization method for equipment maintenance decision and task scheduling in the fixed-point maintenance mode according to claim 6 is characterized in that: The mathematical model: The model is established as follows: Among them, formula (1) is the objective function 1, which represents the maintenance benefit generated by maintenance according to the maintenance decision. The benefit is composed of the secondary use time of the equipment and the equipment importance coefficient; Formula (2) is the objective function 2, which represents the maximum number of maintenance equipment; Formula (3) shows that when the maintenance task execution timing is within the time window, the maintenance benefit is not affected; otherwise, the penalty factor A is introduced to appropriately reduce the maintenance benefit; Formula (4) is the spare parts constraint, which is the necessary resource condition for maintenance equipment. Maintenance is carried out on the basis of ensuring sufficient spare parts. The replacement strategy can be appropriately added to change the number of spare parts. The decision is passed up to the maintenance timing, which affects the final maintenance decision. Formula (5) and (6) are singularity constraints, which means that maintenance is not performed or is performed only once; Formula (7) is the constraint of serial replacement, which ensures the non-repeatability of the serial replacement strategy; Formula (8) is the maintenance deadline constraint; R1~R4 represent the maintenance task sets of periodic maintenance, scheduled maintenance, preventive maintenance, and random failure maintenance equipment respectively; the maintenance task sets R2, R3, and R4 can be abstracted into the periodic maintenance task set R1, and the conversion process is: 1) The scheduled maintenance task x3∈R2, after the calculation of the remaining life, is converted into the maintenance time e3, which can be abstracted as the time window (-∞, e3); 2) The preventive maintenance task x4∈R3, after detecting the fault curve, reaches the fault threshold σ at the time e4, which can be abstracted as a time window (-∞, e4); 3) Random failure maintenance x5∈R4, the time of random equipment failure is unknown, so it can be abstracted as a time window (-∞,+∞).

Images