Improved particle swarm algorithm-based method and device for dynamic rescheduling of maintenance support resources

Abstract

The application provides a maintenance support resource dynamic rescheduling method and device based on improved particle swarm optimization, and relates to the technical field of resource scheduling. The method comprises the following steps: determining an attribute decision set of multiple maintenance support tasks according to multiple influence factors of the multiple maintenance support tasks and priorities of the influence factors; determining a priority order of the multiple maintenance support tasks according to weights of the multiple influence factors and the attribute decision set; correcting an initial scheduling scheme population to obtain a corrected scheduling scheme population; updating each corrected position code according to each corrected position code of the corrected scheduling scheme population, a fitness function value corresponding to each corrected position code, a coding value corresponding to each corrected position code and a crossover operation, obtaining a new scheduling scheme population, and realizing initial planning optimization of resource scheduling of the maintenance support tasks; and when a new maintenance support task demand occurs, realizing dynamic rescheduling of the maintenance support resources. In this way, the time for dynamic rescheduling of the maintenance support resources is relatively short.

Description

Technical Field

[0001] This invention relates to the field of resource scheduling technology, and in particular to an improved particle swarm optimization algorithm for dynamic rescheduling of maintenance and support resources. Background Technology

[0002] Equipment maintenance and support encompasses all activities undertaken to maintain, restore, and improve the prescribed technical condition of equipment. Task allocation and scheduling are crucial components of equipment maintenance and support, playing a vital role in the proper functioning of equipment maintenance and support command. Currently, equipment maintenance and support task allocation and scheduling technology mainly includes two parts: task priority determination and task scheduling. Reasonable task priority classification is a prerequisite, ensuring the full utilization of maintenance and support resources. Task scheduling is key, enabling rapid response to real-time maintenance and support needs. Equipment maintenance and support often occurs in a collaborative manner, requiring close cooperation, resource sharing, and coordinated maintenance. How to quantify, rationally allocate, and dynamically schedule maintenance and support resources based on dynamic task requirements, shorten the waiting time for each maintenance and support task, and fully utilize all maintenance and support resources is a pressing issue that needs to be addressed in equipment maintenance and support.

[0003] Currently, the priority determination of maintenance support tasks mainly employs methods such as fuzzy comprehensive evaluation and analytic hierarchy process (AHP). Decision-makers base their decisions on relevant equipment information, combined with their own knowledge and experience. However, this process is influenced by subjective factors, making it difficult to ensure the reliability and accuracy of priority decisions. For dynamic resource scheduling planning for maintenance support tasks, conventional scheduling planning methods are used. However, given multiple maintenance support tasks, multiple maintenance support resource conditions, and the possibility of multiple resources supporting a single task, achieving optimal resource scheduling is a complex planning problem. Conventional scheduling planning methods rarely yield satisfactory optimization results in a short time, resulting in prolonged dynamic rescheduling of maintenance support resources. Summary of the Invention

[0004] The purpose of this invention is to provide an improved method and apparatus for dynamic rescheduling of maintenance and support resources using particle swarm optimization, thereby solving the problem of long rescheduling times for maintenance and support resources.

[0005] To address the aforementioned technical problems, the embodiments of the present invention provide the following technical solutions:

[0006] The first aspect of this invention provides an improved method for dynamic rescheduling of maintenance and support resources based on the particle swarm optimization algorithm. The method includes:

[0007] Based on multiple influencing factors corresponding to multiple maintenance and support tasks and the priority of each influencing factor, an attribute decision set for multiple maintenance and support tasks is determined. The attribute decision set is used to indicate the quantitative values ​​of multiple influencing factors.

[0008] Based on the weights and attribute decision sets of multiple influencing factors, the priority order of multiple maintenance and support tasks is determined;

[0009] Based on priority order and preset length, multiple maintenance and support tasks are coded to obtain multiple location codes, and each location code is used as a resource scheduling scheme to generate an initial scheduling scheme population.

[0010] Based on the preset requirements, preset carrying capacity, target carrying capacity, current location of maintenance support resources, and location of maintenance support tasks, the initial scheduling scheme population is modified to obtain the modified scheduling scheme population.

[0011] Based on the corrected position codes of the population of the corrected scheduling scheme, the fitness function value corresponding to each corrected position code, the code value corresponding to each corrected position code, and the crossover operation, update each corrected position code to obtain a new population of the scheduling scheme.

[0012] The initial scheduling scheme population and the new scheduling scheme population are merged until the iteration conditions meet the preset conditions, and the optimal scheduling scheme population is obtained.

[0013] According to the optimal scheduling scheme population, multiple maintenance and support tasks are scheduled. When a new maintenance and support task is obtained, the latest maintenance and support task is determined based on the current status of the maintenance and support task and the new maintenance and support task.

[0014] Determine the new maintenance support resources and their remaining carrying capacity corresponding to the latest maintenance support task, update the resource location of the new maintenance support resources, and determine the current location of maintenance support resources in the scheduling state. The resource location is the previous location of the maintenance support task.

[0015] The latest maintenance and support task is treated as multiple maintenance and support tasks, the new maintenance and support resource is treated as maintenance and support resource, the remaining carry capacity is treated as the target carry capacity, the current position is treated as the current position, the next position of the resource position is treated as the maintenance and support task position, and the steps of determining the attribute decision set of multiple maintenance and support tasks based on multiple influencing factors corresponding to multiple maintenance and support tasks and the priority of each influencing factor are returned to generate the final scheduling scheme population.

[0016] The second aspect of this application provides an improved particle swarm optimization algorithm for dynamic rescheduling of maintenance and support resources. The dynamic rescheduling device for maintenance and support resources includes:

[0017] The first determining module is used to determine the attribute decision set of multiple maintenance and support tasks based on multiple influencing factors corresponding to multiple maintenance and support tasks and the priority of each influencing factor. The attribute decision set is used to indicate the quantitative values ​​of multiple influencing factors.

[0018] The second determination module is used to determine the priority order of multiple maintenance and support tasks based on the weights and attribute decision sets of multiple influencing factors.

[0019] The first generation module is used to encode multiple maintenance and support tasks according to priority order and preset length, obtain multiple location codes, and use each location code as a resource scheduling scheme to generate an initial scheduling scheme population.

[0020] The correction module is used to correct the initial scheduling scheme population based on the preset requirements, preset carrying capacity, target carrying capacity of maintenance support resources, current location of maintenance support resources and location of maintenance support tasks corresponding to multiple maintenance support tasks, so as to obtain a corrected scheduling scheme population.

[0021] The update module is used to update each correction position code based on the correction position code of the population of the corrected scheduling scheme, the fitness function value corresponding to each correction position code, the code value corresponding to each correction position code, and the crossover operation, so as to obtain a new population of the scheduling scheme.

[0022] The population merging module is used to merge the initial scheduling scheme population and the new scheduling scheme population until the iteration conditions meet the preset conditions to obtain the optimal scheduling scheme population.

[0023] The third determination module is used to schedule multiple maintenance and support tasks according to the optimal scheduling scheme population. When a new maintenance and support task is obtained, the latest maintenance and support task is determined based on the current status of the maintenance and support task and the new maintenance and support task.

[0024] The fourth determination module is used to determine the new maintenance support resources and the corresponding remaining carrying capacity corresponding to the latest maintenance support task, update the resource location of the new maintenance support resources, and determine the current location of the maintenance support resources in the scheduling state. The resource location is the previous location of the maintenance support task.

[0025] The second generation module is used to take the latest maintenance support task as multiple maintenance support tasks, the new maintenance support resource as maintenance support resource, the remaining carry capacity as target carry capacity, the current position as the current position, the next position of the resource position as the maintenance support task position, and return the steps of determining the attribute decision set of multiple maintenance support tasks based on multiple influencing factors corresponding to multiple maintenance support tasks and the priority of each influencing factor, so as to generate the final scheduling scheme population.

[0026] Compared to existing technologies, the improved particle swarm optimization algorithm-based dynamic rescheduling method and apparatus for maintenance and support resources provided by this invention can first determine a new scheduling scheme population based on the fitness function value and crossover operation in the genetic operator to complete the initial planning. When a new maintenance and support task is obtained, the latest maintenance and support task, the new maintenance and support resources and the corresponding remaining carrying capacity are determined, and the resource position of the new maintenance and support resources is updated. The current position of the maintenance and support resources in the scheduling state is determined, and the final scheduling scheme population is dynamically generated to realize the dynamic rescheduling task planning of maintenance and support resources. The dynamic rescheduling task planning of maintenance and support resources can make the dynamic rescheduling time of maintenance and support resources shorter. Attached Figure Description

[0027] The above and other objects, features, and advantages of exemplary embodiments of the present invention will become readily apparent upon reading the following detailed description with reference to the accompanying drawings. In the drawings, several embodiments of the invention are illustrated by way of example and not limitation, with the same or corresponding reference numerals denoteing the same or corresponding parts, wherein:

[0028] Figure 1 A flowchart illustrating the dynamic rescheduling method for maintenance support resources based on the improved particle swarm optimization algorithm is shown.

[0029] Figure 2 A schematic diagram illustrating the classification of maintenance and support tasks is provided.

[0030] Figure 3 A schematic diagram illustrating various factors influencing the priority of maintenance and support tasks is provided.

[0031] Figure 4 A schematic diagram illustrating the initial maintenance support tasks and the location distribution of maintenance resources is provided.

[0032] Figure 5 A schematic diagram of the encoding scheme is shown.

[0033] Figure 6 A schematic diagram of the crossover operator is shown.

[0034] Figure 7(a) schematically illustrates the task completion rate;

[0035] Figure 7(b) schematically illustrates the total execution time of the task;

[0036] Figure 8(a) schematically illustrates the Gantt chart of the initial maintenance support task execution;

[0037] Figure 8(b) schematically illustrates the Gantt chart of the initial maintenance support task execution;

[0038] Figure 8(c) schematically illustrates the Gantt chart of the initial maintenance support task execution;

[0039] Figure 9 The diagram illustrates the location distribution of maintenance support tasks and maintenance resources through dynamic rescheduling.

[0040] Figure 10 The diagram illustrates the algorithm optimization iteration curves during the optimization process of the dynamic rescheduling scheme for maintenance resources.

[0041] Figure 11 The diagram illustrates the execution of dynamic rescheduling of maintenance resources using a Gantt chart.

[0042] Figure 12 The schematic diagram illustrates the structure of a dynamic rescheduling device for maintenance support resources based on an improved particle swarm optimization algorithm. Detailed Implementation

[0043] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the invention and to fully convey the scope of the invention to those skilled in the art.

[0044] It should be noted that, unless otherwise stated, the technical or scientific terms used in this invention should have the ordinary meaning as understood by one of ordinary skill in the art.

[0045] The methods described in the embodiments of the present invention will be explained in detail below.

[0046] Figure 1 A flowchart illustrating a dynamic rescheduling method for maintenance support resources based on an improved particle swarm optimization algorithm, as shown in this embodiment of the invention, is provided. Figure 1 As shown, the method may include:

[0047] S101. Based on the multiple influencing factors corresponding to multiple maintenance and support tasks and the priority of each influencing factor, determine the attribute decision set of multiple maintenance and support tasks.

[0048] Among them, the attribute decision set is used to indicate the quantitative values ​​of multiple influencing factors.

[0049] Each maintenance and support task corresponds to an attribute decision set. Multiple maintenance and support tasks are affected by multiple factors, including the importance of the maintenance and support task, the environmental requirements of the maintenance and support task, maintenance time, and maintenance and support timeliness.

[0050] Generally speaking, command equipment is of the highest importance, followed by combat equipment, while support equipment is of relatively lower importance. Therefore, based on the type of maintenance and support mission and its contribution to battlefield operations, maintenance and support missions are divided into three categories: Category 1: Maintenance and support missions for command equipment and emergency combat execution equipment (combat equipment that will immediately participate in subsequent military activities) are classified as emergency maintenance and support missions; Category 2: Maintenance and support missions for combat execution equipment (execution equipment that will not immediately participate in combat) are classified as general maintenance and support missions; Category 3: Other maintenance and support missions (such as logistics equipment maintenance and support missions) are classified as non-emergency maintenance and support missions.

[0051] Figure 2 The diagram illustrates the classification of maintenance and support tasks. For different types of maintenance and support tasks, the execution order is always as follows: emergency maintenance and support tasks are executed first, general maintenance and support tasks are executed second, and non-emergency maintenance and support tasks are executed last.

[0052] The higher the quantified value of the maintenance support task's importance, the higher its urgency and priority. Similarly, a higher quantified value for the environmental requirements of a maintenance support task indicates a harsher battlefield environment, greater difficulty in getting maintenance resources to the task's location, and a lower priority. A higher quantified value for maintenance time also indicates a lower priority. The maintenance support task's support timeliness is the latest time when the available resources for the maintenance support task can reach the point where maintenance can begin. Therefore, a lower quantified value for the support task's support timeliness means the maintenance support task needs to be executed earlier, and the higher its priority.

[0053] The fuzzy evaluations given for each maintenance and support task regarding its importance, environmental requirements, maintenance time, and timeliness are converted into quantitative values, forming attribute decision sets for multiple maintenance and support tasks. Figure 3 This diagram schematically illustrates various factors influencing the priority of maintenance support tasks. Multiple factors corresponding to various maintenance support tasks include task importance, environmental requirements, maintenance time, and timeliness. The quantitative value for task importance is given by the task classification: emergency maintenance support task, general maintenance support task, and non-emergency maintenance support task, with corresponding quantitative values ​​of 3, 2, and 1, respectively. The quantitative value for environmental requirements is given based on factors such as the degree of threat from enemy artillery fire to the task location and the difficulty of road access for maintenance resources, categorized as severe, moderate, and easy, with corresponding quantitative values ​​of 3, 2, and 1, respectively.

[0054] S102. Determine the priority order of multiple maintenance and support tasks based on the weights and attribute decision sets of multiple influencing factors.

[0055] Specifically, based on the weights of multiple influencing factors and the attribute decision set, the priority order of multiple maintenance and support tasks is determined, including:

[0056] Step A1: Obtain the weights of multiple influencing factors based on the correspondence table of multiple influencing factors and their weights.

[0057] The correspondence table is obtained by multiple experts subjectively scoring the impact of various influencing factors on the priority determination of maintenance and support tasks. Then, the scores of multiple experts are combined to give the weighted average of the influencing factor scores for each maintenance and support task. Finally, the table is normalized to give the weight of the multiple influencing factors for each maintenance and support task.

[0058] Step A2: Construct an initial evaluation matrix based on the quantified values ​​of multiple influencing factors in the attribute decision set.

[0059] The priority order of multiple maintenance and support tasks can be determined based on the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS).

[0060] The attribute decision set obtained by quantifying the values ​​of M maintenance and support tasks constitutes the initial evaluation matrix X:

[0061]

[0062] Where X is the initial evaluation matrix, A M For the Mth maintenance and support task, a Mm It is the m-th influencing factor of the M-th maintenance and support task, where M is the number of maintenance and support tasks and m is the number of influencing factors.

[0063] Step A3: Normalize all influencing factors in the initial evaluation matrix to obtain the evaluation standard matrix.

[0064] The initial evaluation matrix is ​​normalized by vector normalization to eliminate the influence of dimensions between indicators, thus obtaining the evaluation normalization matrix.

[0065] For the two influencing factors, namely the importance of maintenance support tasks and the environmental requirements of maintenance support tasks, the following formula is used for normalization:

[0066]

[0067] Among them, c1bh To evaluate the h-th influencing factor (importance of maintenance support task and environmental requirements of maintenance support task) of the b-th maintenance support task in the specification matrix, a1 bh Let h be the h-th influencing factor (importance of maintenance support task and environmental requirements of maintenance support task) of the b-th maintenance support task in the initial evaluation matrix, M be the number of maintenance support tasks, m be the number of influencing factors, b∈(1,M), h∈(1,m).

[0068] For the two influencing factors of maintenance support task support timeliness and maintenance time, the following formula is used for normalization:

[0069]

[0070] Among them, c2 bh To evaluate the h-th influencing factor (maintenance timeliness and maintenance time) of the b-th maintenance support task in the specification matrix, a2 bh Let h be the h-th influencing factor (timeliness and maintenance time) of the b-th maintenance support task in the initial evaluation matrix, where M is the number of maintenance support tasks, m is the number of influencing factors, b∈(1,M), and h∈(1,m).

[0071] Step A4: Multiply each column of influencing factors in the evaluation norm matrix by its corresponding weight to obtain the weighted evaluation norm matrix.

[0072] Multiply each column of the evaluation norming matrix by its corresponding weight to obtain the weighted evaluation norming matrix:

[0073] E = (e bh ) M×m =(ω h ×c bh ) M×m ;

[0074] Where E is the evaluation weighted norm matrix, e bh To evaluate the h-th influencing factor of the b-th maintenance and support task in the weighted normative matrix, ω h c represents the weight of the influencing factors. bh To evaluate the h-th influencing factor of the b-th maintenance support task in the specification matrix, where M is the number of maintenance support tasks, m is the number of influencing factors, b∈(1,M), h∈(1,m).

[0075] Step A5: Determine the priority order of multiple maintenance and support tasks based on the evaluation weighted specification matrix and multiple maintenance and support tasks.

[0076] Specifically, based on the evaluation weighted specification matrix and multiple maintenance support tasks, the priority order of multiple maintenance support tasks is determined, including:

[0077] Step A51: Determine the ideal solution and negative ideal solution corresponding to the evaluation weighted specification matrix.

[0078] In evaluating the weighted normalization matrix, the maximum value is selected from each column of influencing factors. The row of factors corresponding to each column of influencing factors with the maximum value is then determined as the ideal solution S. + In evaluating the weighted normalization matrix, the minimum value is selected from each column of influencing factors. The row of factors corresponding to each column of influencing factors with the minimum value is then determined as the negative ideal solution S. - .

[0079] Step A52: Based on the ideal solution, the negative ideal solution, and multiple maintenance support tasks, determine the proximity of multiple maintenance support tasks to obtain the proximity of all maintenance support tasks.

[0080] Based on the ideal solution, the negative ideal solution, and multiple maintenance support tasks, specifically, the Euclidean distance between each maintenance support task and the ideal solution and the negative ideal solution is calculated, and the closeness between the maintenance support task and the ideal solution is defined as:

[0081]

[0082] Among them, Q b For the sake of closeness, A b For each maintenance and support mission, S + For the ideal solution, S - For a negative ideal solution, dis() is used to calculate the Euclidean distance, where b∈(1,M).

[0083] Step A53: Sort the proximity of multiple maintenance and support tasks from largest to smallest to obtain the sorted proximity scores.

[0084] Step A54: Determine the priority order of multiple maintenance and support tasks according to their proximity in the sorting.

[0085] The higher the proximity value of the maintenance support task, the closer the evaluation index of the maintenance support task is to the ideal solution, and the higher the task priority. The priority of maintenance support tasks is determined by sorting them from largest to smallest based on the proximity value. Figure 4 The diagram illustrates the initial maintenance support tasks and the location distribution of maintenance resources, where the higher the priority of the maintenance support task, the larger the circle representing the maintenance support task.

[0086] S103. Based on priority order and preset length, encode multiple maintenance and support tasks to obtain multiple location codes, and use each location code as a resource scheduling scheme to generate an initial scheduling scheme population.

[0087] Particle Swarm Optimization (PSO) is a swarm intelligence-based optimization algorithm that finds optimal solutions by simulating the movement of individuals within a swarm in the search space. A particle's current position represents a candidate solution to the corresponding optimization problem, and its movement is its search process. The particle's velocity can be dynamically adjusted based on its historical best position and the population's historical best position. Particles have two attributes: velocity and position. Velocity represents the speed of movement, and position represents the direction of movement. The optimal solution found by each particle individually is called its individual extreme value, and the best individual extreme value in the swarm is considered the current global optimal solution. The algorithm iterates continuously, updating velocity and position, until a final optimal solution satisfying the termination condition is obtained. While PSO converges quickly, it is prone to getting trapped in local optima. An improved PSO algorithm, incorporating mutation and crossover operators, maintains population diversity and ensures global optimality.

[0088] The preset length is M×N, where M is the number of maintenance and support tasks and N is the number of maintenance and support resource bits.

[0089] A position code X in the designed discrete particle swarm optimization algorithm i ={x i1 ,x i2 ,…,x i(M×N) This paper presents a solution to the maintenance support resource scheduling problem for maintenance support tasks. The specific location coding rule is as follows: Assuming there are M maintenance support tasks, and each task corresponds to N maintenance support resource bits, a location code of length M×N is constructed. Integer coding is used, and the coding is performed according to the priority order of the maintenance support tasks. Every N codes represent a resource scheduling scheme for a maintenance support task. Each location code corresponds to multiple first code values ​​and multiple second code values. The first code value is 1, indicating that the maintenance resource scheduling executes the maintenance support task; the second code value is 0, indicating that the maintenance resource scheduling does not execute the maintenance support task. Where X... i To encode the i-th position, x i2 It is the second bit of the encoded value for the i-th position.

[0090] Figure 5 The diagram illustrates the coding scheme. After random initialization, multiple location codes, i.e., multiple maintenance support tasks, are constructed to form a maintenance support resource scheduling scheme population X = {X1, ..., X...}. S}, X1 is the first position code. The length of a position code is M×N, there are M maintenance support tasks, and each maintenance support task corresponds to N maintenance support resource bits, for a total of S position codes.

[0091] S104. Based on the preset requirements, preset carrying capacity, target carrying capacity, current location of maintenance support resources, and location of maintenance support tasks, the initial scheduling scheme population is modified to obtain the modified scheduling scheme population.

[0092] Among them, the values ​​corresponding to each position code in the initial scheduling scheme population include multiple first code values ​​and multiple second code values. The multiple first code values ​​are used to indicate that maintenance support resources schedule multiple maintenance support tasks, and the multiple second code values ​​are used to indicate that maintenance support resources do not schedule multiple maintenance support tasks. The target carrying capacity of maintenance support resources includes the first target carrying capacity of maintenance support resources corresponding to the multiple first code values.

[0093] Specifically, based on the preset requirements, preset carrying capacity, target carrying capacity, current location of maintenance resources, and location of maintenance tasks, the initial scheduling scheme population is modified to obtain a modified scheduling scheme population, including:

[0094] Step B1: Determine whether the initial scheduling scheme population meets the preset requirements.

[0095] Preset requirements are the maintenance resource requirements for maintenance support tasks.

[0096] Determine whether the initial scheduling scheme population meets the preset requirements. If not, execute step B2a or step B2b below; if yes, execute step S105.

[0097] Step B2a: When the first target carrying capacity exceeds the preset carrying capacity, determine the first time sequence of the maintenance support resources arriving at the maintenance support task location based on the current location of the maintenance support resources and the maintenance support task location corresponding to the multiple first coding values, and accumulate the first carrying capacity of the maintenance support resources according to the first time sequence. If the first carrying capacity is equal to the preset carrying capacity, determine the first maintenance support resource corresponding to the first carrying capacity, and correct the multiple first coding values ​​corresponding to the second maintenance support resource to multiple second coding values.

[0098] The second maintenance support resource refers to maintenance support resources other than the first maintenance support resource.

[0099] If the initial scheduling scheme population does not meet the preset requirements, for the resource scheduling scheme corresponding to the location code, the first target carrying amount of maintenance support resources corresponding to multiple first coding values, that is, the carrying amount of coding value 1, exceeds the resource requirements of the maintenance support task. According to the order of arrival time of the resources at the maintenance support task location, their resource carrying amounts are accumulated one by one. When this total carrying amount of resources just meets the preset carrying amount, the remaining unused maintenance support resources with coding value 1 are deleted, that is, the coding value in the corresponding location code is corrected from 1 to 0.

[0100] Step B2b: When the first target carrying capacity does not exceed the preset carrying capacity, determine the second time sequence of the maintenance support resources arriving at the maintenance support task location based on the current location of the maintenance support resources and the maintenance support task location corresponding to the multiple second coding values, and accumulate the second carrying capacity of the maintenance support resources according to the second time sequence. If the second carrying capacity is equal to the preset carrying capacity, determine the third maintenance support resource corresponding to the second carrying capacity, and correct the multiple second coding values ​​corresponding to the fourth maintenance support resource to multiple first coding values.

[0101] The fourth maintenance support resource is the maintenance support resource other than the third maintenance support resource.

[0102] If the initial scheduling scheme population does not meet the preset requirements, for the resource scheduling scheme corresponding to the location code, the first target carrying capacity of maintenance support resources corresponding to multiple first coding values, i.e., the carrying capacity of coding value 1, is lower than the resource requirements of the maintenance support task. The time when the remaining resources, i.e. the resources corresponding to coding value 0, arrive at the location of the maintenance support task is calculated. According to the order of arrival time, new maintenance support resources are added sequentially to the location of the maintenance task, i.e., the coding value is corrected from 0 to 1. The total carrying capacity of the corresponding maintenance support resources is calculated. When the total carrying capacity of resources meets the preset carrying capacity, the addition of new maintenance support resources is stopped.

[0103] S105. Based on the corrected position codes of the population of the corrected scheduling scheme, the fitness function value corresponding to each corrected position code, the code value corresponding to each corrected position code, and the crossover operation, update each corrected position code to obtain a new population of the scheduling scheme.

[0104] Specifically, based on the corrected position codes of the revised scheduling scheme population, the fitness function values ​​corresponding to each corrected position code, the corresponding code values, and the crossover operations, the corrected position codes are updated to obtain a new scheduling scheme population, including:

[0105] Step C1: Determine the velocity code based on the corrected position codes of the population in the corrected scheduling scheme, the fitness function value corresponding to each corrected position code, and the scaling factor.

[0106] Specifically, based on the corrected position codes of the population according to the corrected scheduling scheme, the fitness function values ​​corresponding to each corrected position code, and the scaling factor, the velocity code is determined, including:

[0107] Step C11: Decode each corrected position code of the corrected scheduling scheme population and calculate the first fitness function value corresponding to each corrected position code.

[0108] The Particle Swarm Optimization (PSO) algorithm simulates the movement of a resource scheduling scheme in the search space to find the optimal solution by updating the population and the position codes of each particle. First, the modified position codes of the modified scheduling scheme population are decoded, and the first fitness function value corresponding to each modified position code is calculated. Each position code corresponds to a first fitness function.

[0109] The fitness function is the evaluation criterion for iteration in the entire particle swarm optimization algorithm. Particles with better fitness have position codes closer to the optimal solution. Simultaneously, during algorithm iteration, other particle codes gradually move closer to the fitness function, thus converging the code to an optimal solution. For the task scheduling model in this invention, the optimization objective is to minimize the total execution time of maintenance and support tasks while maximizing the overall completion rate. The formula for calculating fitness is as follows:

[0110] fitness={maxRate,minTime};

[0111] Here, Rate represents the completion rate of maintenance and support tasks, and Time represents the total execution time of maintenance and support tasks.

[0112] Step C12: Determine the optimal location code and the optimal resource scheduling scheme code based on the first fitness function value.

[0113] Based on the conditions of maximizing the completion rate of maintenance support tasks and minimizing the total execution time of maintenance support tasks, the optimal first fitness function is selected from all the first fitness functions. Based on the conditions of maximizing the completion rate of maintenance support tasks and minimizing the total execution time of maintenance support tasks, the optimal first fitness function is selected through multiple iterations. Based on the optimal first fitness function, the optimal location code and the optimal resource scheduling scheme code can be determined.

[0114] Step C13: Determine the speed code based on the optimal location code, the optimal resource scheduling scheme code, and the scaling factor.

[0115] The particle's velocity code, i.e., the particle's optimal movement direction, is constructed using the following mutation operator. Specifically, the velocity code is determined based on the optimal position code, the optimal resource scheduling scheme code, and the scaling factor, using the following formula:

[0116] V i,G=X i,G +ξ·(X gbest -X i,G )+ξ·(X pbest -X r1,G );

[0117] Among them, V i,G The velocity encoding is the velocity encoding corresponding to the i-th position in the G-th iteration, X. i,G To encode the i-th position in the G-th iteration, ξ is a scaling factor with a value range of [0, 1], X gbest The optimal position encoding, i.e., the position encoding of the historically optimal resource scheduling scheme of the entire population during the algorithm iteration process, X pbest Encoding the optimal resource scheduling scheme, i.e., the historical optimal resource scheduling scheme encoding for an individual particle, X r1,G To encode the r1-th position in the G-th iteration, V i,G ={v i1 ,…,v ij ,…,v i(M×N)}, X i,G ={x i1 ,…,x ij ,…,x i(M×N)}, v ij For multiple bits of the velocity code, that is, the j-th bit of the velocity code corresponding to the i-th position code in the G-th iteration, x ij This is the encoding value corresponding to each corrected position encoding, that is, the j-th bit encoding value of the i-th position encoding.

[0118] Step C2: Update each corrected position code according to the speed code and the corresponding code value of each corrected position code to obtain the updated scheduling scheme population.

[0119] Among them, the speed code includes multiple bits of the speed code.

[0120] Specifically, based on the speed code and the corresponding code values ​​of each corrected position code, the corrected position codes are updated to obtain the updated scheduling scheme population, including:

[0121] Based on the multiple bits of the velocity code and the corresponding code values ​​of each corrected position code, update each corrected position code to obtain the updated scheduling scheme population:

[0122]

[0123] Wherein, s(v ij ) represents the mapping value of the velocity encoding, v ij Multiple bits for speed encoding, x is an intermediate value in the velocity encoding mapping process. ijFor each correction position, the corresponding encoded value is ~x. ij To invert the code value corresponding to each corrected position code, i is the i-th position code, and j is the j-th bit in the multi-bit velocity code.

[0124] The modulus value of each bit in speed coding |v ij The larger the value, the greater the deviation between the position code and the optimal scheduling scheme. The first formula above maps this value to a value closer to 1. In the second formula above... i,j The closer the value is to 1, the greater the probability that the position code of that bit will be inverted, and the closer the updated position code will be to the optimal scheduling scheme.

[0125] Step C3: Perform a crossover operation on the updated position codes of the updated scheduling scheme population to obtain a new scheduling scheme population.

[0126] Based on the updated scheduling scheme population after the location coding scheme is updated, a crossover operation of location coding is performed to further increase the diversity of resource scheduling schemes. The crossover operator adopts an exponential crossover method. Figure 6 A schematic diagram of the crossover operator is shown, which encodes the position X. i,G Arbitrarily select an integer K in [1, M×N] as the starting position of the crossover, and then randomly select another integer k in [1, M×N] as the number of codes to be crossovered. Finally, perform exponential crossover to generate a new scheduling scheme population corresponding to the new maintenance and support task.

[0127] S106. Merge the initial scheduling scheme population and the new scheduling scheme population until the iteration conditions meet the preset conditions to obtain the optimal scheduling scheme population.

[0128] By comparing the fitness values ​​corresponding to each position code, the initial and new scheduling scheme populations are merged to achieve a second optimization of the scheduling scheme population, thus preserving the elite of the better resource scheduling schemes.

[0129] By comparing the fitness values ​​corresponding to each position code, the old and new populations are merged. Specifically, this involves sorting all fitness values ​​of the determined initial scheduling scheme population and all fitness values ​​of the new scheduling scheme population in descending order, and proposing scheduling schemes from the second half of the fitness values ​​in the scheduling scheme population to achieve population merging.

[0130] Once the iteration conditions of the improved particle swarm optimization algorithm meet the preset conditions, the iteration stops, and the position code with the optimal fitness value is obtained. The position code is then decoded, and the optimal scheduling scheme population obtained by the algorithm iteration optimization is output.

[0131] S107. According to the optimal scheduling scheme population, schedule multiple maintenance and support tasks. When a new maintenance and support task is obtained, determine the latest maintenance and support task based on the current status of the maintenance and support task and the new maintenance and support task.

[0132] The determination status includes "in execution" and "not executed". The determination status also includes "executed", and the "not executed" status includes "unexecuted". Unexecuted maintenance support tasks cannot guarantee timeliness constraints. Maintenance support resources are divided into three states: scheduling, waiting, and execution.

[0133] During the execution of the initial maintenance and support task resource scheduling scheme, new maintenance and support task requirements may arise. At this time, based on the initial maintenance and support task resource scheduling scheme, the maintenance status of the current maintenance and support task will be determined and the maintenance and support resource information will be updated. Then, through the designed improved particle swarm algorithm, the maintenance and support task will be replanned and resources will be rescheduled, and new maintenance and support resources will be dynamically allocated.

[0134] Specifically, multiple maintenance and support tasks are scheduled according to the optimal scheduling scheme population. When a new maintenance and support task is obtained, the latest maintenance and support task is determined based on the current status of the maintenance and support task and the new maintenance and support task, including:

[0135] According to the optimal scheduling scheme population, multiple maintenance and support tasks are scheduled. When a new maintenance and support task is obtained, the current maintenance and support task that is in the execution state, the current maintenance and support task that is not in the execution state, and the new maintenance and support task are combined to form the latest maintenance and support task.

[0136] Re-evaluate priorities, delete maintenance and support tasks that do not meet priority constraints and release the occupied maintenance and support resources.

[0137] S108. Determine the new maintenance support resources and their remaining carrying capacity corresponding to the latest maintenance support task, update the resource location of the maintenance support resources, and determine the current location of the maintenance support resources in the scheduling state.

[0138] The resource location is the location preceding the maintenance and support task location.

[0139] The formula for calculating the current location of maintenance and support resources in the scheduling state is as follows:

[0140]

[0141] in, This refers to the current location of maintenance and support resources in a scheduling state, i.e., the location of the resource set when a new maintenance and support task demand arrives. The resource location for maintenance and support, i.e., the location before resource aggregation and scheduling, s v To ensure the speed of maintenance and resource support, time insert For the new maintenance and support tasks, time s The time when maintenance and support resources begin to be allocated.

[0142] S109. The latest maintenance support task is treated as multiple maintenance support tasks, the new maintenance support resource is treated as maintenance support resource, the remaining carry capacity is treated as the target carry capacity, the current position is treated as the current position, the next position of the resource position is treated as the maintenance support task position, and the steps of determining the attribute decision set of multiple maintenance support tasks based on multiple influencing factors corresponding to multiple maintenance support tasks and the priority of each influencing factor are returned to generate the final scheduling scheme population.

[0143] There are currently two main strategies for dynamic task rescheduling: one is to dynamically repair existing plans, which causes less disruption to the original plan but only focuses on local impacts and is difficult to maximize overall task efficiency; the other is to completely reschedule unexecuted tasks and dynamically inserted new tasks, generating a new scheduling plan. This approach can guarantee the maximization of the overall efficiency of the scheduling plan, but it consumes relatively more computational resources. This invention addresses the dynamic planning of priority maintenance and support tasks. Dynamic repair often struggles to simultaneously guarantee task priority constraints and maximize overall efficiency, so a complete rescheduling strategy is adopted to achieve better maintenance and support task efficiency.

[0144] The latest maintenance and support task is treated as multiple maintenance and support tasks, the new maintenance and support resource is treated as maintenance and support resource, the remaining carry capacity is treated as the target carry capacity, the current position is treated as the current position, and the next position of the resource position is treated as the maintenance and support task position. Then, the process returns to step S101, which is the step of determining the attribute decision set of multiple maintenance and support tasks based on multiple influencing factors corresponding to multiple maintenance and support tasks and the priority of each influencing factor, in order to generate a final scheduling scheme population. The final scheduling scheme population includes the maintenance and support resource dynamic rescheduling execution scheme of the maintenance and support task.

[0145] To verify the proposed improved particle swarm optimization (PSO) algorithm for dynamic rescheduling of maintenance support resources, the algorithm parameters shown in Tables 1, 2, and 3, and the expert scoring table shown in Table 4, are used to verify the reliability of the invention. The performance of existing PSO algorithms, improved PSO algorithms, and the invention is compared. Simultaneously, the simulation examples of new maintenance support tasks shown in Table 6 are used to verify the reliability of the invention in maintenance support task replanning and resource rescheduling.

[0146] Table 1 Initial Maintenance and Support Task Requirements

[0147]

[0148]

[0149] Table 2 Initial Resource Set Configuration Table

[0150]

[0151] Table 3 Algorithm Parameter Table

[0152]

[0153] Table 4 Expert Scoring Sheet

[0154]

[0155] Analysis of experimental results:

[0156] Based on the data obtained from Tables 1 to 4 above, the priority ranking of maintenance and support tasks is 5→7→1→6→2→3→8→9→4→10. Refer to Figure 7 for the algorithm performance comparison and Figure 8 for the task execution Gantt charts obtained from different algorithms. Improved Particle Swarm Optimization Algorithm 1 is a maintenance and support equipment resource scheduling method based on a new position encoding update method. Improved Particle Swarm Optimization Algorithm 2 is a maintenance and support resource scheduling method based on Improved Particle Swarm Optimization Algorithm 1, incorporating mutation and crossover operators.

[0157] Figure 7(a) schematically illustrates the task completion rate. Figure 7(a) shows that the improved particle swarm optimization algorithm for dynamic rescheduling of maintenance and support resources proposed in this invention, along with the discrete particle swarm optimization algorithm and the improved particle swarm optimization algorithm, can achieve the highest initial task completion rate within a finite number of iterations. Figure 7(b) schematically illustrates the total task execution time. Figure 7(b) shows that compared with the discrete particle swarm optimization algorithm, both improved methods have improved optimization effects. However, the improved particle swarm optimization algorithm 1 can only converge to a better result in the later stages of the algorithm, while the improved particle swarm optimization algorithm 2 with added mutation and crossover operators can obtain the best resource scheduling scheme optimization in the early stages. This proves that the proposed new position encoding update method combined with mutation and crossover operators is conducive to obtaining a better maintenance and support task resource scheduling scheme.

[0158] Figure 8(a) schematically shows the Gantt chart of the initial maintenance and support task execution (general particle swarm), Figure 8(b) schematically shows the Gantt chart of the initial maintenance and support task execution (improved particle swarm 1), and Figure 8(c) schematically shows the Gantt chart of the initial maintenance and support task execution (improved particle swarm 2). In each chart, a color represents the maintenance execution of a maintenance and support task, the start of each colored rectangle is the time when the maintenance and support resource arrives at the location of the maintenance and support task, and the end of the rectangle is the end time when the maintenance and support task completes maintenance. Each row is a sequence of maintenance and support tasks arranged by maintenance and support resources.

[0159] Comparison and analysis of optimization results of different algorithms:

[0160] Table 5 Comparison of Algorithm Optimization Results

[0161]

[0162] The optimization results and analysis of dynamic scheduling of maintenance resources when new maintenance support tasks arise:

[0163] Table 6 Requirements for New Maintenance and Support Tasks

[0164]

[0165]

[0166] Based on the initial resource scheduling scheme for maintenance and support tasks presented in the improved Particle Swarm Optimization Algorithm 2, the task status is first determined, and maintenance and support tasks 2 that do not meet the priority constraints are deleted. The priority decision results for the remaining maintenance and support tasks are 15→7→11→1→6→12→3→13→8→9→4→14→10. Figure 9 The diagram illustrates the location distribution of maintenance support tasks and resources under dynamic rescheduling. It shows the location distribution of new maintenance support tasks and resources after dynamic programming task preprocessing following the arrival of new maintenance support task requirements. The higher the priority of the maintenance support task, the larger the circle representing the maintenance support task.

[0167] Figure 10 The diagram illustrates the algorithm optimization iteration curves during the optimization process of the dynamic rescheduling scheme for maintenance resources. Figure 10 (a) is the dynamic adjustment of the task completion rate iteration curve, that is, the task completion rate iteration curve of new maintenance and support tasks during the algorithm iteration process. It can be seen that during the algorithm iteration optimization process, the given maintenance resource scheduling scheme can gradually find a scheme with a task completion rate of 1, that is, it can enable all new maintenance and support tasks to be executed. Figure 10(b) To dynamically adjust the task execution time iteration curve, that is, the total execution time iteration curve of new maintenance and support tasks during the algorithm iteration process, under the condition that the task completion rate remains unchanged, the algorithm iteration optimization process can always optimize in the direction of making the total execution time of maintenance and support tasks shorter and shorter. When the task completion rate changes to 1, it can quickly find the resource scheduling scheme with the shortest execution time of all maintenance and support tasks and eventually converge. Figure 11 The diagram schematically illustrates a Gantt chart for the dynamic rescheduling of maintenance resources. The starting time is the moment when a new maintenance support task demand arrives and dynamic rescheduling begins. Each color represents the execution of a new maintenance support task. The beginning of each colored rectangle represents the moment when maintenance support resources arrive at the location of the new maintenance support task, and the end of the rectangle represents the end time when the new maintenance support task completes its maintenance. Each row represents a sequence of new maintenance support tasks dynamically rescheduled. The diagram demonstrates that the proposed improved particle swarm optimization algorithm-based dynamic rescheduling method for maintenance support resources can still quickly optimize and obtain the dynamic rescheduling scheme for maintenance support tasks with the highest task completion rate and shortest maintenance time after a new maintenance support task demand arrives.

[0168] Based on the above Figure 1As can be seen from the implementation method, the embodiments of the present invention update each modified position code according to the modified position code of each modified position code, the fitness function value corresponding to each modified position code, the code value corresponding to each modified position code, and the crossover operation to obtain a new scheduling scheme population; merge the initial scheduling scheme population and the new scheduling scheme population until the iteration condition meets the preset condition to obtain the optimal scheduling scheme population; schedule multiple maintenance support tasks according to the optimal scheduling scheme population; when a new maintenance support task is obtained, determine the latest maintenance support task according to the judgment status of the current maintenance support task and the new maintenance support task; determine the latest maintenance support task. The system calculates the new maintenance support resources and their remaining carrying capacity corresponding to the task, updates the resource location of the new maintenance support resources, determines the current location of the maintenance support resources in the scheduling state (the resource location is the previous location of the maintenance support task), treats the latest maintenance support task as multiple maintenance support tasks, treats the new maintenance support resources as maintenance support resources, treats the remaining carrying capacity as the target carrying capacity, treats the current location as the current location, treats the next location of the resource location as the maintenance support task location, and returns the steps of determining the attribute decision set of multiple maintenance support tasks based on multiple influencing factors corresponding to multiple maintenance support tasks and the priority of each influencing factor, in order to generate the final scheduling scheme population. In this way, a new scheduling scheme population can be determined first based on the fitness function value and crossover operation in the genetic operator to complete the initial planning. When a new maintenance and support task is obtained, the latest maintenance and support task, the new maintenance and support resources and the corresponding remaining carrying capacity are determined, and the resource position of the new maintenance and support resources is updated. The current position of the maintenance and support resources in the scheduling state is determined, and the final scheduling scheme population is dynamically generated to realize the dynamic rescheduling and re-scheduling task planning of maintenance and support resources. The dynamic rescheduling and re-scheduling task planning of maintenance and support resources can make the dynamic rescheduling time of maintenance and support resources shorter.

[0169] Based on the same inventive concept, as an implementation of the above-mentioned improved particle swarm optimization algorithm for dynamic rescheduling of maintenance and support resources, this embodiment of the invention also provides an improved particle swarm optimization algorithm for dynamic rescheduling of maintenance and support resources. Figure 12 This is a structural diagram of the device in an embodiment of the present invention. See also: Figure 12 As shown, the device may include:

[0170] The first determining module 1201 is used to determine the attribute decision set of multiple maintenance and support tasks based on multiple influencing factors corresponding to multiple maintenance and support tasks and the priority of each influencing factor. The attribute decision set is used to indicate the quantitative values ​​of multiple influencing factors.

[0171] The second determining module 1202 is used to determine the priority order of multiple maintenance and support tasks based on the weights and attribute decision sets of multiple influencing factors.

[0172] The first generation module 1203 is used to encode multiple maintenance and support tasks according to priority order and preset length, obtain multiple location codes, and use each location code as a resource scheduling scheme to generate an initial scheduling scheme population.

[0173] The correction module 1204 is used to correct the initial scheduling scheme population based on the preset requirements, preset carrying capacity, target carrying capacity of maintenance support resources, current location of maintenance support resources and location of maintenance support tasks corresponding to multiple maintenance support tasks, so as to obtain a corrected scheduling scheme population.

[0174] The update module 1205 is used to update each correction position code according to the correction position code of the population of the correction scheduling scheme, the fitness function value corresponding to each correction position code, the code value corresponding to each correction position code and the crossover operation, so as to obtain a new population of the scheduling scheme.

[0175] The population merging module 1206 is used to merge the initial scheduling scheme population and the new scheduling scheme population until the iteration conditions meet the preset conditions to obtain the optimal scheduling scheme population.

[0176] The third determination module 1207 is used to schedule multiple maintenance and support tasks according to the optimal scheduling scheme population. When a new maintenance and support task is obtained, the latest maintenance and support task is determined based on the current status of the maintenance and support task and the new maintenance and support task.

[0177] The fourth determination module 1208 is used to determine the new maintenance support resources and the corresponding remaining carrying capacity corresponding to the latest maintenance support task, update the resource location of the new maintenance support resources, and determine the current location of the maintenance support resources in the scheduling state. The resource location is the previous location of the maintenance support task.

[0178] The second generation module 1209 is used to take the latest maintenance support task as multiple maintenance support tasks, the new maintenance support resource as maintenance support resource, the remaining carrying capacity as target carrying capacity, the current position as the current position, the next position of the resource position as the maintenance support task position, and return the steps of determining the attribute decision set of multiple maintenance support tasks based on multiple influencing factors corresponding to multiple maintenance support tasks and the priority of each influencing factor, so as to generate the final scheduling scheme population.

[0179] The second determining module 1202 is specifically used to obtain the weights of multiple influencing factors based on a correspondence table of multiple influencing factors and their weights; construct an initial evaluation matrix based on the quantified values ​​of multiple influencing factors in the attribute decision set; normalize all influencing factors in the initial evaluation matrix to obtain an evaluation specification matrix; multiply each column of influencing factors in the evaluation specification matrix by its corresponding weight to obtain a weighted evaluation specification matrix; and determine the priority order of multiple maintenance and support tasks based on the weighted evaluation specification matrix and the multiple maintenance and support tasks.

[0180] The second determining module 1202 determines the priority order of multiple maintenance and support tasks based on the evaluation weighted specification matrix and multiple maintenance and support tasks, including: determining the ideal solution and negative ideal solution corresponding to the evaluation weighted specification matrix; determining the proximity of multiple maintenance and support tasks based on the ideal solution, negative ideal solution and multiple maintenance and support tasks to obtain the proximity of all maintenance and support tasks; sorting the proximity of multiple maintenance and support tasks from largest to smallest to obtain the sorted proximity; and determining the priority order of multiple maintenance and support tasks based on the sorted proximity.

[0181] The values ​​corresponding to the position codes in the initial scheduling scheme population in the first generation module 1203 include multiple first coding values ​​and multiple second coding values. The multiple first coding values ​​are used to indicate that maintenance support resources schedule multiple maintenance support tasks, and the multiple second coding values ​​are used to indicate that maintenance support resources do not schedule multiple maintenance support tasks. The target carrying capacity of maintenance support resources includes the first target carrying capacity of maintenance support resources corresponding to the multiple first coding values.

[0182] The correction module 1204 is specifically used to determine whether the initial scheduling scheme population meets the preset requirements. If not, when the first target carrying capacity exceeds the preset carrying capacity, the first time sequence of the maintenance support resources arriving at the maintenance support task location is determined based on the current location of the maintenance support resources corresponding to multiple first coding values ​​and the maintenance support task location. The first carrying capacity of the maintenance support resources is accumulated according to the first time sequence. If the first carrying capacity is equal to the preset carrying capacity, the first maintenance support resource corresponding to the first carrying capacity is determined, and the multiple first coding values ​​corresponding to the second maintenance support resource are corrected to multiple second coding values. The second maintenance support resource is a maintenance support resource other than the first maintenance support resource. When the first target carrying capacity does not exceed the preset carrying capacity, the second time sequence of the maintenance support resources arriving at the maintenance support task location is determined based on the current location of the maintenance support resources corresponding to multiple second coding values ​​and the maintenance support task location. The second carrying capacity of the maintenance support resources is accumulated according to the second time sequence. If the second carrying capacity is equal to the preset carrying capacity, the third maintenance support resource corresponding to the second carrying capacity is determined, and the multiple second coding values ​​corresponding to the fourth maintenance support resource are corrected to multiple first coding values. The fourth maintenance support resource is a maintenance support resource other than the third maintenance support resource.

[0183] The update module 1205 is used to determine the velocity code based on the corrected position codes of the corrected scheduling scheme population, the fitness function value corresponding to each corrected position code, and the scaling factor; update each corrected position code based on the velocity code and the code value corresponding to each corrected position code to obtain the updated scheduling scheme population; and perform a crossover operation on the updated position codes of the updated scheduling scheme population to obtain a new scheduling scheme population.

[0184] The update module 1205 determines the velocity code based on the corrected position codes of the corrected scheduling scheme population, the fitness function value corresponding to each corrected position code, and the scaling factor. This includes: decoding each corrected position code of the corrected scheduling scheme population and calculating the first fitness function value corresponding to each corrected position code; determining the optimal position code and the optimal resource scheduling scheme code based on the first fitness function value; and determining the velocity code based on the optimal position code, the optimal resource scheduling scheme code, and the scaling factor.

[0185] The update module 1205, which includes multiple bits of the speed code, updates each corrected position code based on the speed code and the corresponding code value, resulting in an updated scheduling scheme population. This includes: updating each corrected position code based on the multiple bits of the speed code and the corresponding code value, thus obtaining an updated scheduling scheme population.

[0186]

[0187] Wherein, s(v ij) represents the mapping value of the velocity encoding, v ij Multiple bits for speed encoding, x is an intermediate value in the velocity encoding mapping process. (ij) For each correction position, the corresponding encoded value is ~x. (ij) To invert the code value corresponding to each corrected position code, i is the i-th position code, and j is the j-th bit in the multi-bit velocity code.

[0188] The third determination module 1207 is specifically used to schedule multiple maintenance and support tasks according to the optimal scheduling scheme population. When a new maintenance and support task is obtained, the current maintenance and support task that is in the execution state, the current maintenance and support task that is not in the execution state, and the new maintenance and support task are combined to form the latest maintenance and support task. The determination state includes the execution state and the non-execution state.

[0189] It should be noted that the above description of the embodiment of the dynamic rescheduling device for maintenance and support resources based on the improved particle swarm optimization algorithm is similar to the description of the embodiment of the dynamic rescheduling method for maintenance and support resources based on the improved particle swarm optimization algorithm, and has similar beneficial effects. For technical details not disclosed in the embodiments of the dynamic rescheduling device for maintenance and support resources based on the improved particle swarm optimization algorithm of the present invention, please refer to the description of the embodiment of the dynamic rescheduling method for maintenance and support resources based on the improved particle swarm optimization algorithm of the present invention for understanding.

[0190] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for improving the dynamic rescheduling of maintenance support resources using a particle swarm optimization algorithm, characterized in that, The method comprises the following steps: According to the priority of each influence factor corresponding to the multiple maintenance support tasks and the multiple influence factors, an attribute decision set of the multiple maintenance support tasks is determined, wherein the attribute decision set is used to indicate the quantized values of the multiple influence factors; According to the weight of the multiple influence factors and the attribute decision set, a priority order of the multiple maintenance support tasks is determined; According to the priority order and a preset length, the multiple maintenance support tasks are encoded to obtain multiple position encodings, and each position encoding is taken as a resource scheduling scheme to generate an initial scheduling scheme population; According to the preset demand of the maintenance support resources corresponding to the multiple maintenance support tasks, the preset carrying capacity, the target carrying capacity of the maintenance support resources, the current position of the maintenance support resources and the position of the maintenance support tasks, the initial scheduling scheme population is modified to obtain a modified scheduling scheme population; According to each modified position encoding of the modified scheduling scheme population, the fitness function value corresponding to each modified position encoding, the encoding value corresponding to each modified position encoding and a crossover operation, each modified position encoding is updated to obtain a new scheduling scheme population; The initial scheduling scheme population and the new scheduling scheme population are merged until the iteration condition meets a preset condition to obtain an optimal scheduling scheme population; According to the optimal scheduling scheme population, the multiple maintenance support tasks are scheduled, when a new maintenance support task is obtained, the latest maintenance support task is determined according to the determination state of the current maintenance support task and the new maintenance support task; The new maintenance support resource corresponding to the latest maintenance support task and the corresponding residual carrying capacity are determined, and the resource position of the new maintenance support resource is updated, the current position of the maintenance support resource in the scheduling state is determined, and the resource position is the previous position of the position of the maintenance support task; The latest maintenance support task is taken as the multiple maintenance support tasks, the new maintenance support resource is taken as the maintenance support resource, the residual carrying capacity is taken as the target carrying capacity, the current position is taken as the current position, the next position of the resource position is taken as the position of the maintenance support task, and the step of determining the attribute decision set of the multiple maintenance support tasks according to the priority of each influence factor corresponding to the multiple maintenance support tasks and the multiple influence factors is returned to generate a final scheduling scheme population.

2. The method of claim 1, wherein, The step of determining the priority order of the multiple maintenance support tasks according to the weight of the multiple influence factors and the attribute decision set comprises the following steps: According to the correspondence table of the multiple influence factors and the weight of the multiple influence factors, the weight of the multiple influence factors is obtained; According to the quantized values of the multiple influence factors in the attribute decision set, an initial evaluation matrix is constructed; All influence factors in the initial evaluation matrix are normalized to obtain an evaluation standard matrix; Each column of influence factors in the evaluation standard matrix is multiplied by the weight of the corresponding influence factor to obtain an evaluation weighted standard matrix; and The method comprises the following steps: According to the priority of each influence factor corresponding to the multiple maintenance support tasks and the multiple influence factors, an attribute decision set of the multiple maintenance support tasks is determined, wherein the attribute decision set is used to indicate the quantized values of the multiple influence factors; According to the weight of the multiple influence factors and the attribute decision set, a priority order of the multiple maintenance support tasks is determined; According to the priority order and a preset length, the multiple maintenance support tasks are encoded to obtain multiple position encodings, and each position encoding is taken as a resource scheduling scheme to generate an initial scheduling scheme population; According to the preset demand of the maintenance support resources corresponding to the multiple maintenance support tasks, the preset carrying capacity, the target carrying capacity of the maintenance support resources, the current position of the maintenance support resources and the position of the maintenance support tasks, the initial scheduling scheme population is modified to obtain a modified scheduling scheme population; According to each modified position encoding of the modified scheduling scheme population, the fitness function value corresponding to each modified position encoding, the encoding value corresponding to each modified position encoding and a crossover operation, each modified position encoding is updated to obtain a new scheduling scheme population; The initial scheduling scheme population and the new scheduling scheme population are merged until the iteration condition meets a preset condition to obtain an optimal scheduling scheme population; According to the optimal scheduling scheme population, the multiple maintenance support tasks are scheduled, when a new maintenance support task is obtained, the latest maintenance support task is determined according to the determination state of the current maintenance support task and the new maintenance support task; The new maintenance support resource corresponding to the latest maintenance support task and the corresponding residual carrying capacity are determined, and the resource position of the new maintenance support resource is updated, the current position of the maintenance support resource in the scheduling state is determined, and the resource position is the previous position of the position of the maintenance support task; The latest maintenance support task is taken as the multiple maintenance support tasks, the new maintenance support resource is taken as the maintenance support resource, the residual carrying capacity is taken as the target carrying capacity, the current position is taken as the current position, the next position of the resource position is taken as the position of the maintenance support task, and the step of determining the attribute decision set of the multiple maintenance support tasks according to the priority of each influence factor corresponding to the multiple maintenance support tasks and the multiple influence factors is returned to generate a final scheduling scheme population. The step of determining the priority order of the multiple maintenance support tasks according to the weight of the multiple influence factors and the attribute decision set comprises the following steps: According to the correspondence table of the multiple influence factors and the weight of the multiple influence factors, the weight of the multiple influence factors is obtained; According to the quantized values of the multiple influence factors in the attribute decision set, an initial evaluation matrix is constructed; All influence factors in the initial evaluation matrix are normalized to obtain an evaluation standard matrix; Each column of influence factors in the evaluation standard matrix is multiplied by the weight of the corresponding influence factor to obtain an evaluation weighted standard matrix; and The method comprises the following steps: According to the priority of each influence factor corresponding to the multiple maintenance support tasks and the multiple influence factors, an attribute decision set of the multiple maintenance support tasks is determined, wherein the attribute decision set is used to indicate the quantized values of the multiple influence factors; According to the weight of the multiple influence factors and the attribute decision set, a priority order of the multiple maintenance support tasks is determined; According to the priority order and a preset length, the multiple maintenance support tasks are encoded to obtain multiple position encodings, and each position encoding is taken as a resource scheduling scheme to generate an initial scheduling scheme population; According to the preset demand of the maintenance support resources corresponding to the multiple maintenance support tasks, the preset carrying capacity, the target carrying capacity of the maintenance support resources, the current position of the maintenance support resources and the position of the maintenance support tasks, the initial scheduling scheme population is modified to obtain a modified scheduling scheme population; According to each modified position encoding of the modified scheduling scheme population, the fitness function value corresponding to each modified position encoding, the encoding value corresponding to each modified position encoding and a crossover operation, each modified position encoding is updated to obtain a new scheduling scheme population; The initial scheduling scheme population and the new scheduling scheme population are merged until the iteration condition meets a preset condition to obtain an optimal scheduling scheme population; According to the optimal scheduling scheme population, the multiple maintenance support tasks are scheduled, when a new maintenance support task is obtained, the latest maintenance support task is determined according to the determination state of the current maintenance support task and the new maintenance support task; The new maintenance support resource corresponding to the latest maintenance support task and the corresponding residual carrying capacity are determined, and the resource position of the new maintenance support resource is updated, the current position of the maintenance support resource in the scheduling state is determined, and the resource position is the previous position of the position of the maintenance support task; The latest maintenance support task is taken as the multiple maintenance support tasks, the new maintenance support resource is taken as the maintenance support resource, the residual carrying capacity is taken as the target carrying capacity, the current position is taken as the current position, the next position of the resource position is taken as the position of the maintenance support task, and the step of determining the attribute decision set of the multiple maintenance support tasks according to the priority of each influence factor corresponding to the multiple maintenance support tasks and the multiple influence factors is returned to generate a final scheduling scheme population. The step of determining the priority order of the multiple maintenance support tasks according to the weight of the multiple influence factors and the attribute decision set comprises the following steps: According to the correspondence table of the multiple influence factors and the weight of the multiple influence factors, the weight of the multiple influence factors is obtained; According to the quantized values of the multiple influence factors in the attribute decision set, an initial evaluation matrix is constructed; All influence factors in the initial evaluation matrix are normalized to obtain an evaluation standard matrix; Each column of influence factors in the evaluation standard matrix is multiplied by the weight of the corresponding influence factor to obtain an evaluation weighted standard matrix; and The method comprises the following steps: According to the priority of each influence factor corresponding to the multiple maintenance support tasks and the multiple influence factors, an attribute decision set of the multiple maintenance support tasks is determined, wherein the attribute decision set is used to indicate the quantized values of the multiple influence factors; According to the weight of the multiple influence factors and the attribute decision set, a priority order of the multiple maintenance support tasks is determined; According to the priority order and a preset length, the multiple maintenance support tasks are encoded to obtain multiple position encodings, and each position encoding is taken as a resource scheduling scheme to generate an initial scheduling scheme population; According to the preset demand of the maintenance support resources According to the evaluation-weighted norm matrix and the plurality of maintenance support tasks, a priority order of the plurality of maintenance support tasks is determined.

3. The method of claim 2, wherein, The method according to the evaluation-weighted norm matrix and the plurality of maintenance support tasks, a priority order of the plurality of maintenance support tasks is determined. An ideal solution and a negative ideal solution corresponding to the evaluation-weighted norm matrix are determined. According to the ideal solution, the negative ideal solution and the plurality of maintenance support tasks, a closeness degree of the plurality of maintenance support tasks is determined to obtain closeness degrees of all maintenance support tasks. The closeness degrees of the plurality of maintenance support tasks are sorted in descending order to obtain sorted closeness degrees. According to the sorted closeness degrees, a priority order of the plurality of maintenance support tasks is determined.

4. The method of claim 1, wherein, The values corresponding to the positions in the initial scheduling scheme population include a plurality of first encoding values and a plurality of second encoding values, the plurality of first encoding values are used to indicate that the maintenance support resources schedule the plurality of maintenance support tasks, the plurality of second encoding values are used to indicate that the maintenance support resources do not schedule the plurality of maintenance support tasks, and the target carrying amount of the maintenance support resources includes a first target carrying amount of the maintenance support resources corresponding to the plurality of first encoding values.

5. The method of claim 4, wherein, The initial scheduling scheme population is corrected according to the preset demand, the preset carrying amount, the target carrying amount of the maintenance support resources corresponding to the plurality of maintenance support tasks, the current positions of the maintenance support resources and the positions of the maintenance support tasks to obtain a corrected scheduling scheme population, including: It is judged whether the initial scheduling scheme population satisfies the preset demand; If not, when the first target carrying amount exceeds the preset carrying amount, a first time order in which the maintenance support resources arrive at the positions of the maintenance support tasks is determined according to the current positions of the maintenance support resources and the positions of the maintenance support tasks corresponding to the plurality of first encoding values, and a first carrying amount of the maintenance support resources is accumulated according to the first time order, if the first carrying amount is equal to the preset carrying amount, a first maintenance support resource corresponding to the first carrying amount is determined, the plurality of first encoding values corresponding to a second maintenance support resource are corrected to the plurality of second encoding values, and the second maintenance support resource is a maintenance support resource other than the first maintenance support resource; When the first target carrying amount does not exceed the preset carrying amount, a second time order in which the maintenance support resources arrive at the positions of the maintenance support tasks is determined according to the current positions of the maintenance support resources and the positions of the maintenance support tasks corresponding to the plurality of second encoding values, and a second carrying amount of the maintenance support resources is accumulated according to the second time order, if the second carrying amount is equal to the preset carrying amount, a third maintenance support resource corresponding to the second carrying amount is determined, the plurality of second encoding values corresponding to a fourth maintenance support resource are corrected to the plurality of first encoding values, and the fourth maintenance support resource is a maintenance support resource other than the third maintenance support resource.

6. The method of claim 1, wherein, The method comprises the following steps: According to the modified scheduling scheme population, the modified position encoding, the fitness function value corresponding to the modified position encoding, the encoding value corresponding to the modified position encoding, and the crossover operation, the modified position encoding is updated to obtain a new scheduling scheme population, which comprises the following steps: According to the modified scheduling scheme population, the modified position encoding, the fitness function value corresponding to the modified position encoding, and the scaling factor, a speed encoding is determined; According to the speed encoding and the encoding value corresponding to the modified position encoding, the modified position encoding is updated to obtain an updated scheduling scheme population; 7. The method of claim 6, wherein, The updated position encoding of the updated scheduling scheme population is subjected to a crossover operation to obtain the new scheduling scheme population. The method comprises the following steps: The modified position encoding of the modified scheduling scheme population is decoded, and a first fitness function value corresponding to the modified position encoding is calculated; According to the first fitness function value, an optimal position encoding and an optimal resource scheduling scheme encoding are determined; 8. The method of claim 6, wherein, According to the optimal position encoding, the optimal resource scheduling scheme encoding, and the scaling factor, a speed encoding is determined. The speed encoding comprises a plurality of bits of the speed encoding, and the method comprises the following steps: wherein s(v ij ) is a mapping value of the velocity code, v ij is a multi-bit of the velocity code, is an intermediate value of the velocity code mapping process, x ij is a corresponding code value of the each modified position code, ~x ij is a negation of the corresponding code value of the each modified position code, i is the i-th position code, and j is the j-th bit of the multi-bit of the velocity code.

9. The method of claim 8, wherein, According to the plurality of bits of the speed encoding and the encoding value corresponding to the modified position encoding, the modified position encoding is updated to obtain the updated scheduling scheme population. The state comprises an executing state and a non-executing state, and the method comprises the following steps:

10. An improved particle swarm algorithm maintenance support resource dynamic rescheduling device, characterized in that, According to the optimal scheduling scheme population, the plurality of maintenance support tasks are scheduled, and when a new maintenance support task is obtained, a latest maintenance support task is determined according to the state of the current maintenance support task and the new maintenance support task. The maintenance support resource dynamic rescheduling device comprises: A first determining module is configured to determine an attribute decision set of the plurality of maintenance support tasks according to a plurality of influence factors corresponding to the plurality of maintenance support tasks and a priority of each influence factor, wherein the attribute decision set is used to indicate quantized values of the plurality of influence factors; A second determining module is configured to determine a priority order of the plurality of maintenance support tasks according to weights of the plurality of influence factors and the attribute decision set; A first generating module is configured to encode the plurality of maintenance support tasks according to the priority order and a preset length to obtain a plurality of position encodings, and generate an initial scheduling scheme population by taking each position encoding as a resource scheduling scheme. The correction module is configured to correct the initial scheduling scheme population according to preset demands and preset carrying amounts of the maintenance support resources corresponding to the plurality of maintenance support tasks, target carrying amounts of the maintenance support resources, current positions of the maintenance support resources, and maintenance support task positions, to obtain a modified scheduling scheme population; The updating module is configured to update each modified position code according to the modified scheduling scheme population, corresponding fitness function values of the modified position codes, corresponding code values of the modified position codes, and a crossover operation, to obtain a new scheduling scheme population; The population merging module is configured to merge the initial scheduling scheme population and the new scheduling scheme population until a preset condition is met when an iteration condition is satisfied, to obtain an optimal scheduling scheme population; The third determining module is configured to schedule the plurality of maintenance support tasks according to the optimal scheduling scheme population, and determine a latest maintenance support task according to a current maintenance support task and the new maintenance support task when the new maintenance support task is obtained; The fourth determining module is configured to determine a new maintenance support resource corresponding to the latest maintenance support task and a corresponding remaining carrying amount, and update a resource position of the new maintenance support resource, and determine a current position of a maintenance support resource in a scheduling state, the resource position being a previous position of the maintenance support task position; The second generating module is configured to take the latest maintenance support task as the plurality of maintenance support tasks, take the new maintenance support resource as the maintenance support resources, take the remaining carrying amount as the target carrying amounts, take the current position as the current positions, take a next position of the resource position as the maintenance support task positions, and return to a step of determining an attribute decision set of the plurality of maintenance support tasks according to a plurality of influence factors corresponding to the plurality of maintenance support tasks and priorities of the influence factors, to generate a final scheduling scheme population.

Images