A multi-objective equipment scheduling scheme generation method based on a genetic algorithm

Abstract

The application discloses a multi-target equipment scheduling scheme generation method based on a genetic algorithm, takes the visible time window of an observation target of equipment as a resource, combines and integrates multi-target scheduling task characteristics into the genetic algorithm under the condition that the resource is constrained, more fully and comprehensively utilizes the resource, can provide more efficient and more reasonable scheduling schemes for scheduling work, and improves a task completion rate and a resource utilization rate.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of resource scheduling, and particularly relates to a multi-target equipment scheduling scheme generation method based on a genetic algorithm. BACKGROUND

[0002] In an increasingly variable load observation, resource scheduling scene, a sudden event can occur at any time. The observation equipment not only needs to complete the observation of the target within a certain time, but also needs to complete the observation task of multiple targets at the same time, so it is particularly important to obtain the optimal equipment scheduling scheme within a period of time. However, there is no effective and sufficient scheduling scheme for the equipment scheduling problem with multi-target capability at present.

[0003] At the same time, considering that such scheduling problem is an NP problem, it is almost impossible to search for the optimal scheduling scheme within a limited time. Therefore, in this field, field search or heuristic algorithm is mostly used to obtain an approximate solution of the optimal scheme within a short time. However, most of them are scheduling scheme generation methods for task multi-target, observation capability unique, that is, one equipment can only observe one target at the same time. The application of the scheduling scheme generation method of the algorithm in the general scene where the equipment capability is not unique is not fully considered. SUMMARY

[0004] In view of the above problems in the prior art, the multi-target equipment scheduling scheme generation method based on the genetic algorithm provided by the application solves the problem of equipment scheduling scheme generation under the multi-target observation capability.

[0005] In order to achieve the above application purpose, the technical scheme adopted by the application is as follows: a multi-target equipment scheduling scheme generation method based on a genetic algorithm, comprising the following steps:

[0006] S1, constructing a scheduling data set in equipment scheduling;

[0007] S2, generating an initial scheduling scheme group according to the scheduling data set, and encoding the initial scheduling scheme group;

[0008] S3, calculating the fitness of the scheduling scheme in the current scheduling scheme group after encoding;

[0009] S4, determining the parent scheduling scheme in the current scheduling scheme group and supplementing the next generation scheduling scheme group according to the fitness of the scheduling scheme;

[0010] S5, performing scheduling scheme information exchange calculation and single scheduling scheme mutation calculation on the scheduling scheme in the current scheduling scheme group, and updating the current scheduling scheme group;

[0011] S6, repeating steps S3-S5 until a preset number of iterations is reached, and determining a final multi-objective equipment scheduling scheme according to the fitness of the current scheduling scheme group.

[0012] Further, the data in the scheduling data set in step S1 includes visible prediction data, equipment scheduling requirement data, and equipment capability data.

[0013] The visible prediction data is single equipment resource data in the form of a five-tuple, which constitutes an available resource set and includes a unique identifier, a start time, an end time, an observation target, and an observation equipment; the equipment scheduling requirement data includes an observation target and an observation frequency; and the equipment capability data is the maximum number of simultaneously observable targets of the equipment.

[0014] Further, in step S2, the method for generating the initial scheduling scheme group is specifically as follows:

[0015] S2-1, calculating the number of time period conflict resources for each available resource in the available resource set;

[0016] S2-2, sorting the available resources according to the number of time period conflict resources;

[0017] S2-3, searching for available resources that satisfy the equipment scheduling requirement and the equipment capability from the sorted available resources in a first-come-first-served manner for the tasks in the current task sequence until the task is completed or there is not enough available resource to satisfy the task;

[0018] S2-4, determining whether there is still an unassigned task in the task sequence;

[0019] If yes, returning to step S2-3;

[0020] If no, proceeding to step S2-5;

[0021] S2-5, determining whether the number of scheduling schemes reaches a preset number of scheduling schemes;

[0022] If no, shuffling the tasks in the current task sequence and returning to step S2-2;

[0023] If yes, generating an initial scheduling scheme group from all the scheduling schemes.

[0024] Further, in step S2, the method for encoding each scheduling scheme in the initial scheduling scheme group is specifically as follows:

[0025] Each scheduling scheme is encoded as an individual, each individual is encoded as an equipment code set, and the number of equipment codes in each individual is equal to the number of available equipment in the entire scheduling task;

[0026] The encoding mode of the individual code corresponding to a single scheduling scheme is:

[0027] Scheduling scheme: [equipment code 1 equipment code 2 … equipment code L1]

[0028] Wherein, L1 is the number of available equipment in the available resource set;

[0029] The encoding mode corresponding to each equipment code is:

[0030] Equipment code: [ability line 1 ability line 2 … ability line L2]

[0031] Wherein, L2 is the maximum number of simultaneous observation targets of the corresponding equipment in the available resource set;

[0032] The encoding mode corresponding to each ability line is:

[0033] Ability line: [resource id 1 resource id 2 … resource L3]

[0034] Wherein, L3 is the number of scheduled resources in a timeline in the scheduling scheme.

[0035] Further, in step S3, the fitness of the ith scheduling scheme i The calculation formula is:

[0036]

[0037] In the formula, cot i The number of tasks completed in the ith scheduling scheme, X is the total number of valid tasks in the equipment scheduling demand data, subscript i = 1, 2, 3, …, N, N is the total number of scheduling schemes in the current scheduling scheme group.

[0038] Further, the step S4 is specifically:

[0039] Taking the fitness of the scheduling scheme as the index, the top 20% of the scheduling schemes are selected as the parent scheduling schemes in the current scheduling scheme group, and the scheduling schemes are randomly selected to supplement the next generation of scheduling scheme group.

[0040] Further, in step S5, the method for scheduling scheme information exchange calculation is specifically:

[0041] S51-1, according to the preset exchange probability, determine the ith scheduling scheme and the jth scheduling scheme participating in information exchange calculation;

[0042] S51-2, randomly select the same number of equipment codes in the ith scheduling scheme and the jth scheduling scheme as the information exchange subject;

[0043] S51-3, randomly select a capability line in each of the selected two equipment encodings, and exchange with each other;

[0044] S51-4, perform a sorting operation on the scheduling scheme after the exchange of the capability lines, and obtain an extended solution space;

[0045] S51-5, search for available resources from the extended solution space in a task-oriented manner to complete the scheduling task, and realize scheduling scheme information exchange calculation.

[0046] The extended solution space is a set of available resources that can be scheduled after canceling the scheduling of part of the available resources.

[0047] Further, in the step S5, the scheduling scheme mutation calculation method is specifically:

[0048] S52-1, according to a preset mutation probability, determine the i th scheduling scheme and the equipment encoding participating in the mutation calculation;

[0049] S52-2, in the equipment encoding participating in the mutation calculation, search for a set of solution space critical points existing according to the capability line thereof;

[0050] S52-3, randomly select a solution space critical point from the set of solution space critical points, and randomly cancel the scheduling of a resource, and then obtain an extended solution space;

[0051] S52-4, search for available resources from the extended solution space in a task-oriented manner to complete the scheduling task, and realize scheduling scheme mutation calculation.

[0052] The solution space critical point is a set of resources that maximize the number of equipment observations, that is, a solution space critical point is a set of available resources with a maximum number of equipment observations.

[0053] Further, in the step S6, determine the scheduling scheme with the highest fitness in the current scheduling scheme group, and decode it, and take the decoded scheduling scheme as the final multi-objective equipment scheduling scheme.

[0054] The beneficial effects of the present application are:

[0055] (1) The present application aims to complete as many tasks as possible, and by taking advantage of the genetic algorithm in solving NP problems, starting from the main operations of encoding, decoding, mutation and exchange, a scheduling scheme generation method is designed, which integrates the characteristics of equipment scheduling problems, so that a higher task completion rate can be obtained in a limited time.

[0056] (2) Processing multiple targets, the present application especially considers the equipment scheduling problem in the case that the processing equipment has multiple target observation capabilities, through the method of capability value segmentation, the coding, decoding, mutation, and exchange operation processes are designed, so that the present application can process the equipment scheduling scheme in the case that the equipment has multiple target observation capabilities. BRIEF DESCRIPTION OF DRAWINGS

[0057] Figure 1 The multi-target equipment scheduling scheme generation method flow chart based on the genetic algorithm provided by the present application.

[0058] Figure 2 The scheduling scheme coding schematic provided by the present application.

[0059] Figure 3 The scheduling scheme information exchange calculation schematic provided by the present application.

[0060] Figure 4 The scheduling scheme mutation calculation schematic provided by the present application. DETAILED DESCRIPTION

[0061] The specific embodiments of the present application are described below to facilitate the understanding of the present application by those skilled in the art, but it should be clear that the present application is not limited to the scope of the specific embodiments, and for those skilled in the art, as long as various changes are within the spirit and scope of the present application defined and determined by the appended claims, these changes are obvious, and all the inventions utilizing the concept of the present application are within the scope of protection.

[0062] The embodiment of the present application provides a multi-target equipment scheduling scheme generation method based on a genetic algorithm, as shown in Figure 1 The method comprises the following steps:

[0063] S1, constructing a scheduling data set during equipment scheduling;

[0064] S2, generating an initial scheduling scheme group according to the scheduling data set, and coding the initial scheduling scheme group;

[0065] S3, calculating the fitness of the scheduling scheme in the coded current scheduling scheme group;

[0066] S4, determining the parent scheduling scheme in the current scheduling scheme group and supplementing the next generation scheduling scheme group according to the fitness of the scheduling scheme;

[0067] S5, performing scheduling scheme information exchange calculation and single scheduling scheme mutation calculation on the scheduling scheme in the current scheduling scheme group, and updating the current scheduling scheme group;

[0068] S6, repeating steps S3-S5 until a preset iteration number is reached, and determining a final multi-objective equipment scheduling scheme according to the fitness in the current scheduling scheme group.

[0069] The data in the scheduling data set in step S1 of the embodiment of the application comprises visible prediction data, equipment scheduling demand data, and equipment capability data.

[0070] The visible prediction data is single equipment resource data in the form of a five-tuple, which constitutes an available resource set and comprises a unique identifier, a start time, an end time, an observation target, and an observation equipment; the equipment scheduling demand data comprises an observation target and an observation frequency; and the equipment capability data is the maximum number of simultaneously observable targets of the equipment.

[0071] In the embodiment, the above three types of data are read and stored, and unusable resources are excluded, so as to obtain the available resource set.

[0072] In the process of generating the initial scheduling scheme group in step S2 of the embodiment of the application, an initial scheduling scheme group is generated by using a first-come-first-served heuristic greedy strategy. In the heuristic greedy strategy, resources are allocated as much as possible to ensure that conflicts are reduced. Therefore, the number of resource conflicts is used as a penalty cost, and the resources are sorted from low to high according to the penalty cost. In scheduling, resources are allocated in a task-oriented manner in a first-come-first-served manner. Task requirements and constraints are preferentially met, and the next task is completed. The process is repeated until all tasks are completed or no task can be completed. The task sequence is shuffled to achieve diversity and randomness of solutions in the initial scheduling scheme group, so as to achieve survival of the fittest in the subsequent mutation and exchange processes.

[0073] Based on this, in step S2 of the embodiment of the application, the method for generating the initial scheduling scheme group is specifically as follows:

[0074] S2-1, calculating the number of time period conflict resources corresponding to each available resource in the available resource set;

[0075] S2-2, sorting the available resources according to the number of time period conflict resources;

[0076] S2-3, searching for available resources that meet the equipment scheduling demand and the equipment capability from the sorted available resources in a first-come-first-served manner for a task in the current task sequence until the task is completed or there is not enough available resource to meet the task;

[0077] S2-4, determining whether there is an unallocated task in the task sequence;

[0078] If yes, returning to step S2-3;

[0079] If no, entering step S2-5;

[0080] S2-5, determining whether the number of scheduling schemes reaches the preset number of scheduling schemes;

[0081] If not, the tasks in the current task sequence are shuffled, and the step S2-2 is returned;

[0082] If yes, all the scheduling schemes are generated into an initial scheduling scheme group.

[0083] In step S2 of the embodiment, when encoding the scheduling schemes, according to the characteristics of the scheduling problem, it is necessary to focus on solving the scheduling characteristics that multiple targets can be observed by the equipment at the same time. Here, the individual encoding scheme is designed with the core of being able to intuitively reflect the scheduling scheme and facilitate the completion of the exchange process and the mutation process. An individual encoding is taken as a complete scheduling scheme. Each individual encoding is an equipment encoding set, and the number of equipment encodings in the individual is the number of available equipment in the entire scheduling. Each equipment encoding is composed of a two-dimensional vector, and the number of vectors in the two-dimensional vector is the maximum number of targets that can be observed by the equipment at the same time. Each vector can be regarded as a time axis, and the resources in the time axis do not conflict with each other. The unique identifier of the resource is saved in the vector.

[0084] Based on this, in step S2 of the embodiment, as shown in Figure 2 the method for encoding each scheduling scheme in the initial scheduling scheme group is specifically:

[0085] Each scheduling scheme is taken as an individual encoding, each individual encoding is an equipment encoding set, and the number of equipment encodings in each individual encoding is the number of available equipment in the entire scheduling task.

[0086] The encoding mode of the individual encoding corresponding to a single scheduling scheme is:

[0087] Scheduling scheme: equipment encoding 1 equipment encoding 2 … equipment encoding L1

[0088] Wherein, L1 is the number of available equipment in the available resource set.

[0089] The encoding mode corresponding to each equipment encoding is:

[0090] Equipment encoding: [ability line 1 ability line 2 … ability line L2]

[0091] Wherein, L2 is the maximum number of targets that can be observed by the corresponding equipment in the available resource set at the same time.

[0092] The encoding mode corresponding to each ability line is:

[0093] Ability line: [resource id 1 resource id 2 … resource L3]

[0094] Wherein, L3 is the number of scheduled resources in a timeline in the scheduling scheme, and the specific meaning of the capability line is a single timeline (in the equipment coding, there is no time window overlap between the resource sets in the same capability line, and the resource data is not repeated between different capability lines).

[0095] In step S3 of the embodiment, the individual fitness value needs to be able to intuitively reflect the advantages and disadvantages of an individual (i.e. a scheduling scheme), and in the equipment scheduling, the task completion rate is the main improvement target, so the task completion rate is directly used as the individual fitness, i.e. the fitness is represented as the task completion rate in the batch demand.

[0096] In the embodiment, the fitness of the i-th scheduling scheme is calculated as follows: i

[0097]

[0098] In the formula, cot i is the number of completed tasks in the i-th scheduling scheme, X is the total number of valid tasks in the equipment scheduling demand data, the subscript i = 1, 2, 3, …, N, and N is the total number of scheduling schemes in the current scheduling scheme group.

[0099] In step S4 of the embodiment, the following is specifically performed:

[0100] Taking the fitness of the scheduling scheme as an index, the top 20% of the scheduling schemes are selected as the parent scheduling schemes in the current scheduling scheme group, and scheduling schemes are randomly selected to supplement the next generation of scheduling scheme group.

[0101] Wherein, according to the Pareto rule, i.e. 20% of the best individuals master most of the useful information, so the top 20% of the scheduling schemes are selected as the parent scheduling schemes in the current scheduling scheme group.

[0102] In step S5 of the embodiment, for the information exchange calculation process of the scheduling scheme, the exchange process in the genetic algorithm aims to combine the advantages of different individuals, and here, in order to ensure that the method complexity does not increase, two individuals in the population are randomly selected for exchange, and the exchange content is a certain capability line (i.e. a certain timeline) in the same equipment coding (i.e. the same equipment). Therefore, after the exchange, there is no resource conflict, but after the exchange, there may be a case of resource reuse, so after the exchange, the scheduling scheme needs to be normalized, and an extended solution space is obtained after the normalization, and other tasks are attempted to be scheduled. Based on this, as shown in the following formula, in step S5 of the embodiment, the method for scheduling scheme information exchange calculation is specifically: Figure 3

[0103] ​​S51-1, determine the i-th scheduling scheme and the j-th scheduling scheme participating in information exchange calculation according to a preset exchange probability;

[0104] S51-2, randomly select the same equipment code in the i-th scheduling scheme and the j-th scheduling scheme (i.e. the same equipment code in different scheduling schemes) as the information exchange subject;

[0105] S51-3, randomly select one capability line in each of the selected two equipment codes, and exchange them with each other;

[0106] S51-4, perform a collation operation on the scheduling scheme after the capability line exchange, and obtain an extended solution space;

[0107] S51-5, search for available resources from the extended solution space in a task-oriented manner to complete the scheduling task, and realize the information exchange calculation of the scheduling scheme;

[0108] The extended solution space is a set of available resources that can be scheduled after canceling the available resources of the scheduling part.

[0109] In step S5 of the embodiment of the application, for the mutation calculation process of the scheduling scheme, the mutation process in the genetic algorithm aims to jump out of the local optimal solution through randomness, and try to find a better individual. Therefore, in the method, first, a "solution space critical point" is defined, that is, a resource set at a certain moment when the number of observed targets of equipment reaches the maximum (a resource set with a maximum observed target number). The mutation process is to destroy the "critical point" to obtain an extended solution space. And through a task-driven manner, the resources in the solution space are searched to try to complete the tasks that have not been completed.

[0110] Based on this, as shown in Figure 4 , the scheduling scheme mutation calculation method in the embodiment is specifically:

[0111] S52-1, determine the i-th scheduling scheme and the equipment code participating in the mutation calculation according to a preset mutation probability;

[0112] S52-2, search for a set of solution space critical points existing in the equipment code participating in the mutation calculation according to the capability line thereof;

[0113] S52-3, randomly select a solution space critical point in the set of solution space critical points, and randomly cancel the scheduling of a resource to obtain an extended solution space;

[0114] S52-4, search for available resources from the extended solution space in a task-oriented manner to complete the scheduling task, and realize the mutation calculation of the scheduling scheme;

[0115] The solution space critical point is a resource set that makes the equipment observation times reach the maximum, that is, one solution space critical point is a resource set available with a size of the maximum observation number of the equipment.

[0116] In step S6 of the embodiment of the present application, the scheduling scheme with the highest fitness in the current scheduling scheme group is determined, and is decoded, and the decoded scheduling scheme is taken as the final multi-target equipment scheduling scheme.

[0117] Those skilled in the art will understand that the embodiments described herein are for the purpose of helping the reader to understand the principles of the present application, and should be understood as not limiting the protection scope of the present application to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations according to the technical inspirations disclosed in the present application without departing from the essence of the present application, and these modifications and combinations are still within the protection scope of the present application.

Claims

1. A multi-objective equipment scheduling scheme generation method based on a genetic algorithm, characterized by, The method comprises the following steps: S1, constructing a scheduling data set in equipment scheduling; S2, generating an initial scheduling scheme group according to the scheduling data set and encoding the initial scheduling scheme group; S3, calculating the fitness of the scheduling schemes in the encoded current scheduling scheme group; S4, determining the parent scheduling schemes in the current scheduling scheme group and supplementing the next generation scheduling scheme group according to the fitness of the scheduling schemes; S5, performing scheduling scheme information exchange calculation and single scheduling scheme mutation calculation on the scheduling schemes in the current scheduling scheme group, and updating the current scheduling scheme group; S6, repeating steps S3-S5 until a preset iteration number is reached, and determining the final multi-objective equipment scheduling scheme according to the fitness in the current scheduling scheme group; The data in the scheduling data set in step S1 comprises visible prediction data, equipment scheduling demand data and equipment capability data; The visible prediction data is single equipment resource data in the form of a five-tuple, which constitutes an available resource set and comprises a unique identifier, a start time, an end time, an observation target and an observation equipment; the equipment scheduling demand data comprises an observation target and an observation frequency; and the equipment capability data is the maximum number of simultaneously observable targets of the equipment; In step S2, the method for encoding each scheduling scheme in the initial scheduling scheme group is specifically as follows: Each scheduling scheme is encoded as an individual, each individual is encoded as an equipment code set, and the number of equipment codes in each individual is equal to the number of available equipment in the entire scheduling task; The encoding mode of the individual corresponding to a single scheduling scheme is as follows: L1 is the number of available equipment in the available resource set; The encoding mode of each equipment code is as follows: L2 is the maximum number of simultaneously observable targets of the corresponding equipment in the available resource set; The encoding mode of each capability line is as follows: wherein, is the number of scheduled resources in a timeline of the scheduling scheme; the capability line is a single timeline, and in equipment coding, there is no time window overlap between the resource sets in the same capability line, and the resource data does not repeat between different capability lines.

2. The multi-objective equipment scheduling scheme generation method based on genetic algorithm according to claim 1, characterized in that, In step S2, the method for generating the initial scheduling scheme group is specifically as follows: S2-1, calculating the number of time period conflict resources of each available resource in the available resource set; S2-2, sorting the available resources according to the number of time period conflict resources; S2-3, searching for available resources that meet the equipment scheduling demand and the equipment capability from the sorted available resources in a first-come-first-served manner for the tasks in the current task sequence until the tasks are completed or there is not enough available resource to meet the tasks; S2-4, judging whether there is still an unallocated task in the task sequence; If yes, returning to step S2-3; If no, proceeding to step S2-5; S2-5, judging whether the number of scheduling schemes reaches a preset number of scheduling schemes; If no, rearranging the tasks in the current task sequence and returning to step S2-2; If yes, generating all the scheduling schemes into the initial scheduling scheme group.

3. The genetic algorithm based multi-objective equipment scheduling scheme generation method of claim 1, wherein, In the step S3, the fitness of the first scheduling scheme is calculated according to the following formula: ​ In the formula, is the number of tasks completed in the th scheduling scheme, is the total number of effective tasks in the equipment scheduling demand data, and the subscript , N is the total number of scheduling schemes in the current scheduling scheme group.

4. The multi-objective equipment scheduling scheme generation method based on genetic algorithm according to claim 1, characterized in that, Step S4 is specifically as follows: Taking the fitness of the scheduling schemes as an index, selecting the top 20% of the scheduling schemes as the parent scheduling schemes in the current scheduling scheme group, and randomly selecting scheduling schemes to supplement the next generation scheduling scheme group.

5. The multi-objective equipment scheduling scheme generation method based on genetic algorithm according to claim 3, characterized in that, In step S5, the method for performing scheduling scheme information exchange calculation is specifically as follows: S51-1, according to the preset exchange probability, determine the first i dispatching scheme and the second j dispatching scheme participating in the information exchange calculation. S51-2, Randomly select the first i The scheduling scheme and the first j In each scheduling scheme, the equipment code with the same number serves as the main body for information exchange; S51-3, randomly select a capability line in each of the selected two equipment encodings, and exchange with each other; S51-4, perform arrangement operation on the scheduling scheme after the exchange of the capability line, and obtain an extended solution space; S51-5, search for available resources from the extended solution space in a task-oriented manner to complete the scheduling task, and realize scheduling scheme information exchange calculation; Wherein, the extended solution space is a set of available resources that can be scheduled after canceling the scheduling of part of the available resources.

6. The multi-objective equipment scheduling scheme generation method based on genetic algorithm according to claim 3, wherein, In the step S5, the scheduling scheme mutation calculation method is specifically: S52-1, according to the preset variation probability, determine the first i Scheduling scheme and the equipment code participating in the variation calculation S52-2, in the equipment encoding participating in the mutation calculation, search for the critical point set of the solution space according to the capability line thereof; S52-3, randomly select a solution space critical point in the solution space critical point set, and randomly cancel the scheduling of a resource, and then obtain an extended solution space; S52-4, search for available resources from the extended solution space in a task-oriented manner to complete the scheduling task, and realize scheduling scheme mutation calculation; Wherein, the solution space critical point is a set of resources that maximizes the number of equipment observations, that is, a solution space critical point is a set of available resources with a size of the maximum number of equipment observations.

7. The multi-objective equipment scheduling scheme generation method based on genetic algorithm according to claim 1, wherein, In the step S6, determine the scheduling scheme with the highest fitness in the current scheduling scheme group, and decode it, and take the decoded scheduling scheme as the final multi-objective equipment scheduling scheme.

Images