Overhead transmission line personnel scheduling process optimization method based on improved genetic algorithm

Abstract

The invention relates to the field of multi-target resource limited scheduling, and discloses an overhead transmission line personnel scheduling process optimization method based on an improved genetic algorithm. Comprising the steps that the positions of all towers needing overhead transmission line construction, existing road information in the construction range where all the towers are located and connection road information between all the towers are obtained; obtaining task loads corresponding to four procedures required in the construction range and the number of constructors and construction efficiency of the corresponding procedures; constructing a minimum construction period solving model which takes minimization of the maximum total construction period as an objective function and takes construction resources, the immediate preceding relation of each working procedure and the completion degree of the working procedure as constraint conditions; and solving the minimum construction period solving model based on a genetic algorithm so as to obtain a construction sequence of the minimum and maximum total construction period of four working procedure construction of overhead power transmission line construction of all towers and shift arrangement of corresponding personnel. According to the invention, the transparency and traceability of the scheduling and workflow are improved.

Description

Technical Field

[0001] The present invention relates to the field of multi-objective resource-constrained scheduling, and in particular to a method for optimizing the personnel scheduling process of overhead power transmission lines based on an improved genetic algorithm. Background Art

[0002] In the process of building line towers in mountainous areas, it is necessary to ensure that road engineering, foundation engineering, tower erection engineering and line stringing engineering (four processes, or four projects) are carried out quickly and effectively, especially in the southwest region where the proportion of mountains is large, line towers are mostly distributed in high mountains, and the tower location has complex terrain. These adverse factors increase the construction period and construction cost of line towers. Therefore, in order to quickly and efficiently complete the construction of power transmission lines in mountainous areas, reasonable planning of construction sequence and reasonable arrangement of construction personnel are the inevitable conditions to ensure the efficient completion of the project. The core of the planning of construction sequence and the arrangement of construction personnel is the exploration of personnel scheduling and process optimization methods for overhead transmission line construction.

[0003] At present, the construction sequence and personnel arrangement of high-altitude power transmission lines mainly rely on the areas in charge of each contractor and the construction experience of on-site technicians. However, due to the lack of corresponding scientific guidance, it is very easy for some construction personnel to be idle, resulting in waste of resources and low construction efficiency during the construction process. This construction method lacks overall coordination and has a large degree of randomness, resulting in too many human experience factors affecting the scientific nature of the construction, which is not conducive to future mechanized construction guidance. Summary of the invention

[0004] In order to solve the problem that the personnel scheduling and process arrangement of overhead power transmission line construction in traditional technology completely depend on the experience and judgment of technical workers, resulting in waste of construction resources and excessively long construction period, the present invention provides an overhead power transmission line personnel scheduling process optimization method based on an improved genetic algorithm, according to the engineering quantity information of the four processes of each line tower in the construction area, the construction efficiency information, and the construction relationship matrix of each line tower, a construction flow chart of each line tower constrained by the immediate predecessor relationship is constructed, and the improved genetic algorithm fused with machine learning and known construction information is used to search for the corresponding construction sequence and personnel arrangement method with the shortest construction period as the objective function. Construction personnel can refer to this construction sequence and personnel arrangement method according to actual conditions to improve the scientificity and reliability of personnel scheduling and process arrangement of overhead power transmission line construction, and facilitate the formation of standards and specifications for personnel arrangement and process arrangement to guide subsequent engineering construction.

[0005] The present invention is achieved through the following technical solutions:

[0006] A method for optimizing the personnel scheduling process of overhead power transmission lines based on an improved genetic algorithm, comprising:

[0007] Obtain the location of each tower where overhead transmission line construction is required, the existing road information within the construction range of each tower, and the connecting road information between the towers;

[0008] Obtain the task volume corresponding to the four processes required within the construction scope, as well as the number of construction personnel and construction efficiency of the corresponding processes;

[0009] Based on the task volume, the number of construction personnel and the construction efficiency, a minimum construction period solution model is constructed with minimization of the maximum total construction period as the objective function and construction resources, the preceding relationship of each process and the completion degree of the process as the constraint conditions;

[0010] The minimum construction period solution model is solved based on a genetic algorithm, thereby obtaining a construction sequence that minimizes the maximum total construction period of the four construction processes for overhead power transmission line construction on all towers and corresponding personnel scheduling.

[0011] As an optimization, if the number of pole towers exceeds the set number threshold, the total construction area corresponding to all the pole towers will be divided into several tensioning sections, and different contractors will carry out independent road engineering, foundation engineering, and tower erection engineering construction in the construction process of the overhead transmission line.

[0012] As an optimization, the objective function is:

[0013] minC max ;

[0014] C max =max{E 1,h ,E 2,h ,E 3,h ,…,E n,h};

[0015] Among them, C max is the maximum completion time set of all towers, E i,h represents the completion time of tower i in process h, i∈[1,n], n is the number of towers in the tensioning section, and h is the total number of construction processes in the traction section.

[0016] As an optimization, the constraints are:

[0017]

[0018] S i,l+1 ≥E i,l ;

[0019]

[0020] Among them, M l represents the number of construction teams in the lth construction process, Xi,l,k Indicates that tower i is constructed by construction team k in process l, I = {l, 2, 3, ..., n}, is the set of towers in a tensioning section, n is the number of towers in the tensioning section, K is the set of construction teams responsible for the traction section, E i,l represents the completion time of tower i in process l, S i,l represents the starting time of tower i in process l, T i,l,k represents the construction time of process l of construction team k at tower i, S i,l+1 represents the starting time of tower i in process l+1, D i,l represents the release time of tower i in process l, L represents the process set, Y i,t,l indicates that tower i is constructed in process l, T is the number of time periods, L is the process set, T i,l,k represents the construction time of process l of construction team k at tower i, S i,立塔 Indicates the start time of the tower erection process for tower i, B i,基础,60 represents the time when the erection process of tower i is 60% complete; S i,架线 Indicates the start time of the wiring process for tower i, B i,立塔,100 Indicates the time when the completion rate of the tower erection process of tower i is 100%.

[0021] As an optimization, the minimum construction period solution model based on the genetic algorithm is specifically to optimize the construction period for all tower construction procedures in a tensioning section corresponding to a contractor. The specific process is:

[0022] A1. Set the number of iterations of the genetic algorithm and the number of chromosomes N in each generation of the population, wherein a fixed number of n' towers are randomly selected from all n towers contracted by a contractor to form a chromosome, and the selection is random and non-repetitive in sequence to form an initial population P, wherein n = n'*N, and the code corresponding to the gene position in the chromosome is the tower number, and the order of the gene position is the order of the construction process of the corresponding tower;

[0023] A2. Divide N chromosomes into k clusters according to the K-means clustering algorithm, and each cluster is a sub-dataset;

[0024] A3, merging the k sub-datasets to form a new data set P′;

[0025] A4. Select chromosomes from the new data set P′ according to the selection rule, so as to divide the new data set P′ into k sub-populations. The selection rule is as follows: the elements in the sub-population a are obtained in an increasing manner with a step size of k based on the element at the a-th position in the data set P′;

[0026] A5. Perform fitness evaluation on the chromosomes in each of the sub-populations, and sort the chromosomes in each of the sub-populations according to the results of the fitness evaluation, and let each of the sub-populations in this iteration process be the parent population, and the chromosomes in the parent population be the parent individuals;

[0027] A6, save the parent individual with the largest result of the first R fitness evaluations in the parent population, select the parent individual from all the parent individuals except the parent individual with the largest result of the first R fitness evaluations by roulette wheel to perform crossover mutation operation to obtain the offspring individual, calculate the fitness evaluation results of the offspring individuals after the crossover mutation and sort them, reinsert the offspring individuals into the parent population according to the fitness evaluation results, select the set number of chromosomes to form a new parent population, and then return to A5;

[0028] A7. Repeat A5-A6 until the termination condition is met, and integrate the optimal solution in each subpopulation.

[0029] As an optimization, when the chromosome performs a crossover operation, if the number of iterations meets the number of self-learning strategies, the following judgment is made:

[0030] Randomly generate a real number in the interval (0,1) and determine whether the real number is <0.5. If it is less than 0.5, use the self-learning strategy to calculate the crossover rate. If it is greater than 0.5, use the initial crossover rate to perform the crossover operation directly.

[0031] If the number of iterations does not meet the number of self-learning strategies, the initial crossover rate is directly used for crossover operation.

[0032] As an optimization, in the self-learning strategy, the learning rate α, discount factor γ, and ε are obtained by combining the Sarsa algorithm with the Q-Learning algorithm. The switching conditions between the Sarsa algorithm and the Q-Learning algorithm are as follows:

[0033] when When , the Sarsa algorithm is used to obtain the learning rate α, discount factors γ, ε;

[0034] when When , the Q-Learning algorithm is used to obtain the learning rate α, discount factors γ, ε;

[0035] Among them, N i Indicates the number of current iterations, N s Indicates the total number of states, N a Indicates the total number of actions.

[0036] As an optimization, in the self-learning strategy, the learning rate α=0.05, the discount factor γ=0.85, and ε=0.80.

[0037] As an optimization, the mutation operation is implemented using a two-point exchange strategy.

[0038] As an optimization, the fitness function for fitness evaluation is Among them, T max (i) represents the maximum total construction period of tower i.

[0039] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0040] Through the method of the present invention, the construction plan can be modeled and optimized to improve the construction efficiency. First, the construction process is converted into an optimization model based on comprehensive consideration of the process sequence and resource constraints, and a heuristic algorithm is designed according to the model requirements to find the optimal construction plan under a given line network layout to shorten the construction period and achieve balanced resource allocation, while estimating the construction period.

[0041] The present invention integrates the optimization genetic algorithm of machine learning, and shows significant improvements and advantages in the field of personnel scheduling and process optimization of overhead power transmission lines. The present invention combines the advantages of global search and local search, maintains the global search capability of the genetic algorithm, and enhances the local search capability at the same time, more effectively explores the solution space, and reduces unnecessary waiting time by optimizing the execution order of the process. The present invention improves the consistency and fluency of the workflow, and uses data analysis and machine learning technology to make the scheduling decision based on data drive, which improves the scientificity and accuracy of the decision. Finally, the present invention improves the transparency and traceability of the scheduling and workflow through digital scheduling and process management, which is convenient for auditing and quality control. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0043] Figure 1 is a flow chart of an embodiment of the present invention;

[0044] Figure 2 This is a flow chart of an improved genetic algorithm integrating machine learning in an embodiment of the present invention;

[0045] Figure 3 It is a Gantt chart of the process of a certain contractor's bid section in an embodiment of the present invention. DETAILED DESCRIPTION

[0046] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.

[0047] As described in the background technology, the construction sequence and personnel arrangement of high-altitude transmission lines mainly rely on the areas in charge of each contractor and the construction experience of on-site technicians. However, due to the lack of corresponding scientific guidance, it is very easy for some construction personnel to be idle, resulting in waste of resources and low construction efficiency during the construction process. This construction method lacks overall coordination and has a large degree of randomness, resulting in excessive human experience factors affecting the scientific nature of the construction, which is not conducive to future mechanized construction guidance.

[0048] In order to solve the defects described in the background technology, an embodiment of the present invention provides an optimization method for the personnel scheduling process of overhead power transmission lines based on an improved genetic algorithm. The method constructs a construction flow chart of each line tower with a preceding relationship as a constraint based on the engineering quantity information of the four processes of each line tower in the construction area (i.e., four projects, including road engineering, foundation engineering, tower erection engineering and wire stringing engineering or called road process, foundation process, tower erection process and wire stringing process), construction efficiency information, and the construction relationship matrix of each line tower, and uses the improved genetic algorithm integrated with machine learning and known construction information, with the shortest construction period as the objective function, to search for the corresponding construction sequence and personnel arrangement method. Construction personnel can refer to this construction sequence and personnel arrangement method according to actual conditions to improve the scientificity and reliability of personnel scheduling and process arrangement of overhead power transmission line construction, and facilitate the formation of standards and specifications for personnel arrangement and process arrangement to guide subsequent engineering construction.

[0049] Furthermore, the personnel arrangement scheme obtained by the optimization method of the embodiment of the present invention can reduce idle personnel in the construction process of the transmission line, make personnel arrangement more scientific and cost-saving, and thus improve the efficiency of the construction scheme. At the same time, by taking the minimum construction time of the entire project as the objective function, the personnel arrangement scheme obtained under this optimal goal can also achieve the effect of shortening the construction period, saving costs, and generating economic benefits in advance.

[0050] Before describing the specific solution, the following parameters are explained first.

[0051] In order to facilitate the description and discussion of this problem, the following mathematical symbols and decision variables are defined:

[0052] I: Tower set, I = {l, 2, 3, ..., n}, n is the total number of towers;

[0053] i: tower number, i∈{1,2,3,…,n};

[0054] K: construction team set, K = {1, 2, 3, ..., m}, m is the total number of construction teams;

[0055] k: construction team number, k∈{1,2,3,…,m};

[0056] L: process set, L = {1, 2, 3, ..., h}, h is the total number of construction processes;

[0057] l: construction process number, l∈{1,2,…,h};

[0058] M l : The number of construction teams in the lth construction process;

[0059] t:: time sequence number, t∈{1,2,3,…,T};

[0060] T i,l,k : Construction time of process l of construction team k at tower i;

[0061] S i,l : The starting time of tower i in process l;

[0062] E i,l : Completion time of tower i in process l;

[0063] D i,l : Release time of tower i in process l;

[0064] B i,l,p : The time when the completion percentage of process l of tower i is p;

[0065] C max : Maximum completion time;

[0066] Decision variables:

[0067]

[0068] The method of the present invention will be described in detail below with reference to specific embodiments. Figure 1 , Figure 1 The present application embodiment provides a schematic flow chart of a method for optimizing the personnel scheduling process of overhead transmission lines based on an improved genetic algorithm, such as Figure 1 As shown, the method comprises the following steps:

[0069] S1. Obtain the location of each tower where overhead transmission line construction is required, the existing road information within the construction range of each tower, and the connecting road information between the towers;

[0070] Specifically, the user inputs the longitude and latitude location data of each transmission tower, the location data of the roads that have been built in the mountainous area, and the connection road information between the towers.

[0071] It should be noted that the connection road information between towers may include three situations: built roads, roads to be built, and cableways to be built. The connection road information needs to be marked and classified for input.

[0072] Different types of connecting road information have no effect on the iteration of the genetic algorithm.

[0073] In addition, the location of the tower can also be regarded as a waypoint, which is obtained by dividing the existing road into a series of continuous points.

[0074] In some embodiments, if the number of pole towers exceeds a set number threshold, the total construction area corresponding to all the pole towers is divided into several tensioning sections, and different contractors carry out independent road engineering, foundation engineering, and tower erection engineering construction in the construction process of the overhead transmission line.

[0075] Within the construction scope, the bus tower is divided into a fixed number of tensioning sections. Different contractors are responsible for the construction of overhead transmission lines in some of the tensioning sections. The contractors do not cooperate with each other during the road engineering, foundation engineering, and tower erection engineering. Only unified wiring is carried out during the wiring engineering stage.

[0076] S2. Obtain the task amount corresponding to the four processes required within the construction scope and the number of construction personnel and construction efficiency of the corresponding processes;

[0077] In some embodiments, the excavation volume of the pole tower road project within the construction scope, the construction time of the foundation project, the tower erection project, the line stringing project, the number of existing construction personnel and the construction efficiency are obtained.

[0078] Specifically, the four processes (road engineering, foundation engineering, tower erection engineering, and line stringing engineering) required for the construction of each pole tower and the corresponding construction time are stored in a list. In this embodiment, the construction time of the road engineering of each pole tower is equal to the ratio of the excavation volume of the pole tower road engineering to the construction efficiency of the road workers. The construction time of the foundation engineering of each pole tower is 7 days, the construction time of the tower erection engineering of each pole tower is 3 days, and the line stringing engineering time of all pole towers in a tension section is 10 days. According to the actual situation, each contractor has a construction team for the four processes of road engineering, foundation engineering, tower erection engineering, and line stringing engineering. Each of the contractors includes four construction teams with full construction capabilities (capable of independently completing the construction content), wherein the workers in the four construction teams with full construction capabilities are road engineering workers, foundation engineering workers, tower erection engineering workers, and line stringing engineering workers, and one process of the same pole tower is carried out by the same construction team. Specifically, in this embodiment, each contractor has a team of road engineering construction workers, and two teams of construction workers for the three types of foundation engineering, tower erection engineering, and line stringing engineering.

[0079] Each process of the tower construction cannot be carried out by multiple construction teams at the same time, but can only be carried out by one construction team.

[0080] The construction of pole towers is not carried out in sequence according to the construction of a single pole tower. It is required that after the completion of road engineering, transportation preparation must be completed before continuing with the foundation engineering. After the completion of the foundation engineering, the tower erection project can only be carried out after the concrete solidifies and the materials are transported. After the tower erection project is completed, preparation and layout must be carried out before the line stringing project can be carried out. Moreover, the line stringing project in a tensioning section can only be carried out after the tower erection projects of all pole towers in the tensioning section are completed.

[0081] The calculation method of each construction period is different. The construction period of the road project depends on the amount of excavation of the temporary road built between two pole towers or between the existing road and the pole tower. The construction period is the ratio of the road excavation construction volume to the efficiency of the road construction team. The construction period of the pole tower foundation project is set at a fixed value of 7 days, and the construction period of the tower erection project is set at a fixed value of 3 days. The wiring project is carried out in tensioning sections. Each tensioning section requires 10 days to carry out the wiring project of all pole towers.

[0082] S3. Based on the task volume, the number of construction personnel and the construction efficiency, a minimum construction period solution model is constructed with minimization of the maximum total construction period as the objective function and construction resources, the preceding relationship of each process and the completion degree of the process as the constraint conditions;

[0083] In each small section (tension and traction section) contracted by each contractor, an improved genetic algorithm that integrates machine learning and the existing construction information of each tower is used to search for the construction sequence with the minimum construction period to complete the four construction processes of all towers.

[0084] Specifically, the information of each tower, including the tower number, the tower process name, the task volume of each tower process, the preceding activity, the waiting time, and the current construction status information are stored and encapsulated in a list to construct a tower information table for each contractor's tensioning section.

[0085] The genetic algorithm integrated with machine learning is used to search the tower information table based on the task volume information of the tower and the sequential relationship of the activities with the purpose of shortest construction period, and the construction process with the shortest construction period in the entire construction process is obtained.

[0086] Specifically, the pole towers to be constructed are numbered, and the construction time of each process in S120 is added to each pole tower. The connection status of each pole tower in S110 can be used to obtain the before and after relationship of the road engineering construction, and the pole tower information table of each contractor section is obtained in combination with the input information.

[0087] The input data types are shown in Table 1 below

[0088] Table 1 Input data name and type

[0089]

[0090]

[0091] Each process of each tower is assigned a time variable. The time variables here specifically include the tower process waiting time variable, the tower process start time variable, and the tower process end time variable. Then, the goal of solving the problem of personnel scheduling and process optimization scheduling of transmission lines is to minimize the maximum completion time. The maximum completion time refers to the time after the last construction process of the last tower is completed. Therefore, the objective function is:

[0092] The objective function is:

[0093] minC max ;

[0094] C max =max{E 1,h ,E 2,h ,E 3,h ,…,E n,h}; Among them, C max is the maximum completion time set of all towers, E i,h represents the completion time of tower i in process h, i∈[1,n], n is the number of towers in the tensioning section, and h is the total number of construction processes in the traction section.

[0095] In this embodiment, since there are 4 processes in total, h is 4. i,h It is the completion time of the fourth process of the i tower, that is, to find the maximum end time of the fourth process of all towers, that is, the end time of the last stage of the last tower is the maximum completion time. The objective function is to minimize the maximum completion time.

[0096] The constraints are:

[0097]

[0098] S i,l+1 ≥E i,l #(3);

[0099]

[0100]

[0101] Among them, M l represents the number of construction teams in the lth construction process, X i,l,kIndicates that tower i is constructed by construction team k in process l, I = {l, 2, 3, ..., n}, is the set of towers in a tensioning section, n is the number of towers in the tensioning section, K is the set of construction teams responsible for the traction section, E i,l represents the completion time of tower i in process l, S i,l represents the starting time of tower i in process l, T i,l,k represents the construction time of process l of construction team k at tower i, S i,l+1 represents the starting time of tower i in process l+1, D i,l represents the release time of tower i in process l, L represents the process set, Y i,t,l indicates that tower i is constructed in process l, T is the number of time periods, L is the process set, T i,l,k represents the construction time of process l of construction team k at tower i, S i,立塔 Indicates the start time of the tower erection process for tower i, B i,基础,60 represents the time when the erection process of tower i is 60% complete; S i,架线 Indicates the start time of the wiring process for tower i, B i,立塔,100 Indicates the time when the completion rate of the tower erection process of tower i is 100%.

[0102] In the above constraints, constraint (1) indicates that in any construction process, each tower to be constructed can only be constructed under the work of one construction team; constraint (2) indicates the constraint relationship between the start time and completion time of each tower in any construction process, that is, the tower process completion time is equal to the tower process start time plus the process construction time, and the construction cannot be interrupted. Constraint (3) indicates the constraint relationship between the various processes of the same tower, i.e., the previous process of the same tower always starts before the next process; Constraint (4) indicates the constraint relationship between the release time and the start time of the same tower in different processes, i.e., the start time of the next process of the tower is equal to the release time of the previous process; Constraint (5) indicates the constraint relationship between the release time and the start time of the same tower in the same construction process, i.e., the start time of the same process of the same tower is always less than the release time, and the tower can only be released after the construction is completed; Constraint (6) indicates that the number of towers constructed at any time in any construction process cannot exceed the number of construction teams available for that process; Constraint (7) indicates the blocking relationship of the towers during construction; Constraint (8) indicates that the start time of any tower must be greater than or equal to 0; Constraint (9) indicates that the construction can only be completed after the construction task of the last process is completed; Constraint (10) indicates that the tower erection project can only be carried out after 60% of the foundation project is completed; Constraint (11) indicates that the line stringing project can only be carried out after 100% of the tower erection project is completed.

[0103] Constraint (1) refers to the constraints between contractors. Specifically, each contractor is responsible for a fixed number of tensioning sections, and contractors do not assist each other in work other than stringing. Contractors can only perform construction work within their own contracted scope.

[0104] Constraint (10) means that there is a constraint relationship between each tower and each tower process. The specific constraint is: when constructing the same tower (pole tower), the road project must be completed before the foundation project, and the tower erection project can be carried out after the foundation project is completed. After the tower erection project is completed, the line stringing project can be carried out. 60% of all tower foundation projects must be completed before the tower erection project can be carried out. After all tower erection projects in a tensioning section are completed, the line stringing project in the tensioning section can be carried out.

[0105] It should be noted that in the actual construction process, due to time and cost considerations, the tower erection project can only be carried out after most of the tower foundation projects are completed. The line stringing project can only be carried out after all the tower erection projects are completed within a tensioning section. In this embodiment, the tower erection project can only be carried out after 60% of the tower foundation projects are completed.

[0106] In addition to the above constraints, the following constraints are also included:

[0107] The construction personnel are regarded as engineering resources. The constraints between the construction personnel are as follows: the road engineering construction personnel are only responsible for the road engineering, the foundation engineering personnel are responsible for the foundation engineering, the tower erection engineering personnel are responsible for the tower erection engineering, and the line stringing engineering personnel are responsible for the line stringing engineering, and they are not interoperable.

[0108] The constraints between construction resources are as follows: each contractor has four construction teams, each of which has one team responsible for road engineering, one team responsible for tower erection engineering, and one team responsible for line stringing engineering. In any time period, the number of construction personnel for one process in the tensioning section of the same contractor shall not exceed the number of existing construction resources, and different construction resources can be allocated in different tensioning sections.

[0109] There are constraints on the construction period of each tower process. The specific constraints are: the construction period of the road project is determined by the excavation volume of the road from the previous tower or the existing road to the current tower, and the construction period of the foundation project is set to a fixed time of 7 days, the construction period of the tower erection project is set to a fixed value of 3 days, and the construction period of the line stringing project is based on the tensioning section. When the tower erection projects of all towers in a tensioning section are completed, the line stringing project in the next tensioning section can be carried out. The construction personnel of the line stringing project are jointly responsible by all contractors. This project uses the line stringing construction team of all contractors. The construction period of the line stringing project in each tensioning section is 10 days.

[0110] Since the actual situation needs to be taken into consideration, the time for transporting materials during the workers' rest time and the solidification time of the construction of concrete, etc., the waiting relationship constraint between the processes is proposed. In this embodiment, after the road project is completed, it is necessary to wait for 1 day of material transportation time before the foundation project is constructed. After the foundation project is completed, considering the solidification time of the concrete and the transportation time of the materials required for the tower project, it is necessary to wait for 7 days of concrete solidification time and 1 day of material transportation time before the tower project is completed. Since the material transportation is independent of the construction team, if the time from the release time of the resources of the previous construction team to the tower exceeds 1 day, there is no need to wait and the construction can be carried out directly. After completing all the tower projects, it takes 1 day to prepare for the site before the line construction project can be carried out. That is to say, since the construction speed of the line construction project is fast and the time cost is large, it will only be prepared when the tower can be used for line construction. In actual construction, the line construction project will definitely require an additional 1 day of preparation time, which can be understood here as the construction time of the line construction project will be extended by 1 day.

[0111] Since some special areas in the construction area need to be completed as soon as possible, there are constraints on the completion time of some tensioning sections. In this embodiment, there is a tensioning section that is completed within the first 100 days of the start of the construction.

[0112] S4. Solving the minimum construction period solution model based on a genetic algorithm, thereby obtaining a construction sequence that minimizes the maximum total construction period for the four construction processes of the overhead transmission line construction on all towers and corresponding personnel scheduling.

[0113] It should be noted that once the construction sequence is obtained, the personnel schedule can be obtained. This is because the construction sequence is arranged according to the existing human resource constraints. Each process will occupy a group of worker resources at the same time, so the final construction sequence also means that the personnel arrangement has also been made.

[0114] It should be noted that, in the present invention, a contractor undertakes multiple continuous tensioning sections. A contractor can construct pole towers in multiple tensioning sections. The genetic algorithm is used to optimize the pole towers in multiple tensioning sections of the same contractor.

[0115] Specifically, the improved genetic algorithm that integrates machine learning and the existing construction information of each tower is used to search for the construction sequence and personnel scheduling that can complete the construction of all four processes of the towers with the minimum construction period. In the process of searching for the minimum construction period and construction sequence, the genetic algorithm is improved by combining the k-means clustering algorithm and the Q-Learning algorithm and Sarsa algorithm in reinforcement learning.

[0116] The resource-constrained project scheduling problem of determining the construction sequence is solved by the minimum construction period model established to solve the minimum construction period required to complete the project and the construction sequence of the tower. The allocation method of construction personnel at the time nodes is obtained.

[0117] In this embodiment, the minimum construction period solution model is solved based on the genetic algorithm to optimize the construction period for all the tower construction procedures in a tensioning section corresponding to a contractor.

[0118] By initializing a population P with N chromosomes, each chromosome becomes a data set separately, and the population P is a data set. The number of centers is set to k, and P is clustered into k clusters according to clustering. That is, the k sub-data sets are merged into a new data set P′, and data are selected from the new data set P′ according to certain selection rules, divided into k sub-populations, and then respectively from.

[0119] The specific process is:

[0120] A1. Set the number of iterations of the genetic algorithm, the number of chromosomes N in each generation population, and the number of clusters k. In the tower group contracted by each contractor, a fixed number of n' towers are randomly selected from all n towers contracted by a contractor to form a chromosome, and the selection is random and non-repetitive in sequence to form an initial population P, where n = n'*N, and the code corresponding to the gene position in the chromosome is the tower number, and the order of the gene position is the order of the construction process of the corresponding tower;

[0121] That is, in the initialization phase, a population P is set, which contains N chromosomes, each chromosome represents a data point, and the entire population P constitutes the data set. At the same time, the number of clusters k is determined

[0122] The present invention adopts the method of initializing the code in the form of integers. A group of codes represents a group of individual pole towers, and each gene position in the code represents the number of the pole tower to be constructed. The group of codes formed represents the construction sequence of this group of pole towers. The length of the code represents the total number of pole towers to be constructed in this group. For example, a group of gene codes is [11, 5, 9, 6, 3, 7, 13, 17], and each digit represents the number of the pole tower to be constructed, that is, n′=8 here. For example, this group of codes represents that the pole towers 11, 5, 9, 6, 3, 7, 13, 17 will be constructed, and the construction order is as shown in the above gene code: 11, 5, 9, 6, 3, 7, 13, 17.

[0123] Fitness evaluation requires decoding of the encoded chromosomes.

[0124] In the process of decoding using the improved genetic algorithm, the decoding method based on integer coding is adopted. All construction teams in each stage are cycled in turn according to the coding sequence, and the construction team that finished the previous work the earliest is selected as the construction team of the current line tower to be constructed. The constraints in any stage can be divided into three parts, namely the completion time of the previous stage, the idle time of the selected construction team, and the waiting time for the construction of the line tower within the scope of responsibility of the entire contractor when all construction teams participate in the work. After entering the next stage, the start time of each stage is the release time of the construction team in the previous stage. This process is to calculate the total construction period.

[0125] The next step is to improve the genetic algorithm, specifically by combining the k-means clustering algorithm with the Q-Learning algorithm and Sarsa algorithm in reinforcement learning.

[0126] A2. Divide N chromosomes into k clusters according to the K-means clustering algorithm, and each cluster is a sub-dataset;

[0127] According to k preset cluster centers, the data set P is clustered and divided into k clusters to form k sub-data sets X(1) to X(k).

[0128] The k-means clustering algorithm is used to classify the initial towers into different clusters, and then the sub-populations with more obvious individual differences are adjusted according to the characteristics of different clusters. When performing cluster analysis, the number of clusters, that is, the number of cluster centers k, must be determined first. In the initial stage, k samples are randomly selected from the data set as the initial cluster centers, denoted as M i , where i ranges from 1 to k. Then, for the remaining samples in the dataset, the Euclidean distance between them and each cluster center Mi is calculated. Based on these distances, each sample is assigned to the nearest cluster center M. i , and classify them into corresponding clusters. Next, the cluster center is updated by calculating the average value of samples in each cluster. Finally, the cluster center is continuously updated through an iterative process until the preset algorithm termination condition is met. The Euclidean distance calculation formula between points is as follows:

[0129]

[0130] Euclidean distance is used to calculate the distance between two data samples in clustering algorithms. Therefore, in the K-means algorithm, the formula for calculating the Euclidean distance is as follows:

[0131]

[0132] Among them, X is the data object in the data set; Mi is the i-th cluster center; m is the dimension of the data object; X j Represents the jth attribute of the data object; C i,j represents the jth attribute of the i-th cluster center.

[0133] One x or y is the sum of the distances of a chromosome (multiple towers).

[0134] The sum of squared errors (SSE), also known as the total squared error, absolute error, residual sum of squares or within-group sum of squares, is the sum of the squares of the errors of all data points. In cluster analysis, SSE is used to evaluate the compactness of clusters, thereby reflecting the quality of clustering. In an ideal clustering process, SSE is inversely proportional to the number of iterations. As the number of iterations increases, SSE should gradually decrease until it reaches a stable state, at which point the error no longer changes significantly. In the K-means algorithm, the formula for calculating the sum of squared errors is as follows:

[0135]

[0136] Where X is the data object; Ci is the center of the i-th cluster; k is the number of clusters.

[0137] A3, merging the k sub-datasets to form a new data set P′;

[0138] The above k sub-datasets are merged to form a new data set P′, which contains all sub-datasets X(1) to X(k).

[0139] When different workers merge sub-datasets into data set P′, the order of the sub-datasets may be different.

[0140] For example, if the k clusters are abcd, then the data set P′ can be case 1: {a, b, c, d}, or case 2: {d, b, c, a}. A4. Select chromosomes from the new data set P′ according to the selection rule, so as to divide the new data set P′ into k sub-populations. The selection rule is as follows: the elements in the sub-population a are obtained based on the (a-1)th position in the data set P′ and in an increasing manner with k as the step size;

[0141] If the order of elements in the data set P′ is different, that is, the previous case 1 and case 2, then the chromosomes in the k subpopulations divided by case 1 and case 2 are also different. Select data from the new data set P′ according to specific rules and divide it into k subpopulations. The selection rules are as follows: Subpopulation 1 contains the first (corresponding to the first position P′[0] in the data set P′, that is, the first element), k+1 (corresponding to the k+1 position P′[k] in the data set P′), 2k+1, etc. elements of P′, until the last element; subpopulation 2 contains the second, k+2, 2k+2, etc. elements of P′, until the second to last element; and so on, until subpopulation k, which contains the (k-1)th, 2k-1th, 3k-1th, etc. elements of P′, until the last element.

[0142] The K-means clustering algorithm is used to divide the population. The initial towers are divided into different clusters according to categories, and then adjusted into sub-populations with more obvious individual differences according to the characteristics of different clusters.

[0143] A5. Perform fitness evaluation on the chromosomes in each of the sub-populations, and sort the chromosomes in each of the sub-populations according to the results of the fitness evaluation, and let each of the sub-populations in this iteration process be the parent population, and the chromosomes in the parent population be the parent individuals;

[0144] In the fitness evaluation phase, the principle adopted is that the higher the fitness score, the greater the possibility of the individual being selected, and thus the closer to the optimal solution of the problem. Given that the scheduling problem aims to minimize the maximum completion time or the maximum total construction period (minT max ), the fitness function is designed to be the inverse of the maximum completion time. This design makes the fitness function inversely proportional to the objective function, that is, the smaller the maximum completion time, the higher the fitness score.

[0145] That is, for tower i, the fitness function is Among them, T max (i) represents the maximum total construction period of tower i.

[0146] A6, save the parent individual with the largest result of the first R fitness evaluations in the parent population, select the parent individual from all the parent individuals except the parent individual with the largest result of the first R fitness evaluations by roulette wheel to perform crossover mutation operation to obtain the offspring individual, calculate the fitness evaluation results of the offspring individuals after the crossover mutation and sort them, reinsert the offspring individuals into the parent population according to the fitness evaluation results, select the set number of chromosomes to form a new parent population, and then return to A5;

[0147] It should be noted that the chromosomes in each subpopulation undergo crossover mutation independently, that is, for example, the chromosomes in subpopulation one will not crossover with the chromosomes in subpopulation two.

[0148] A7. Repeat A5-A6 until the termination condition is met, and integrate the optimal solution in each subpopulation.

[0149] In the population selection stage, a roulette selection mechanism is adopted to determine the probability of an individual being selected as a parent based on his / her fitness. The possibility of an individual being selected as a parent is deduced based on the fitness of individuals in the population. The level of individual fitness directly determines the probability of whether the individual will be selected.

[0150] In this embodiment, in the crossover operation, the OX crossover operator is mainly used, and a crossover rate setting strategy based on improved reinforcement learning is proposed. As for the mutation operation, a two-point exchange strategy is mainly implemented, that is, two gene positions are randomly selected in an individual for exchange to generate a new individual.

[0151] In this embodiment, if Figure 2 As shown, when the chromosome performs a crossover operation, if the number of iterations meets the number of self-learning strategies, the following judgment is made:

[0152] Randomly generate a real number in the interval (0,1) and determine whether the real number is <0.5. If it is less than 0.5, use the self-learning strategy to calculate the crossover rate. If it is greater than 0.5, use the initial crossover rate to perform the crossover operation directly.

[0153] If the number of iterations does not meet the number of self-learning strategies, the initial crossover rate is directly used for crossover operation.

[0154] In the self-learning strategy, the learning rate α, discount factors γ and ε are obtained by combining the Sarsa algorithm with the Q-Learning algorithm. The Sarsa algorithm is used in the early stage of improvement, and the Q-learning algorithm is used in the later stage of improvement. In the early stage of improvement, the Sarsa algorithm is first combined with the crossover of the genetic algorithm, and then the Q value is updated by the update method of the Sarsa algorithm and recorded in a table Q-table. This Q value will be used for Q learning and training.

[0155] The switching conditions between the Sarsa algorithm and the Q-Learning algorithm are as follows:

[0156] when When , the Sarsa algorithm is used to obtain the learning rate α, discount factors γ, ε;

[0157] when When , the Q-Learning algorithm is used to obtain the learning rate α, discount factors γ, ε;

[0158] Among them, N i Indicates the number of current iterations, N s Indicates the total number of states, N a Indicates the total number of actions.

[0159] The total number of states and actions can be obtained as follows:

[0160] State: Each specific crossover rate value in units of 0.01 in [0.70, 0.99] constitutes the entire state set S; the total number of states is the number of all possible crossover rate values ​​in this interval. The calculation method is as follows: Ns = (0.99-0.70) / 0.01+1 = 30. The addition of 1 here is because it includes the two endpoints of 0.70 and 0.99.

[0161] Action: The actions that can be taken in the self-learning crossover rate are to reduce, keep the crossover rate unchanged, or increase it by 0.01. Therefore, the corresponding action set A is {-1, 0, 1}, which means there are three possible actions: reduce by 0.01 (-1), keep it unchanged (0), and increase by 0.01 (1). So the total number of actions is 3.

[0162] The Q-Learning algorithm selects the action with the highest expected Q value in each state to update the Q value. Only through the accumulation of a large amount of learning experience can Q-Learning play a stronger role. However, there is not much experience accumulation in the early stage, which will lead to low learning effect and poor quality. The Sarsa algorithm has a higher learning effect and convergence in the early stage of the algorithm, but due to its conservative learning characteristics, it is easy to fall into the local optimal state and it is difficult to obtain the optimal strategy. Therefore, based on the different characteristics of the two algorithms, the two algorithms are combined, and the Sarsa algorithm and the Q-Learning algorithm are applied at different stages. In line with the idea of ​​​​strengthening the strengths and avoiding the weaknesses, the improved algorithm gives full play to the advantages of the two algorithms and improves the learning efficiency of the algorithm.

[0163] When the initial period is over, the Q-learning algorithm will replace the Sarsa algorithm combined with the genetic method.

[0164] In general, the optimization ideas of the algorithm include: using the Sarsa algorithm in the early stage and switching to the Q-Learning algorithm in the later stage.

[0165] The specific algorithm flow is as follows:

[0166] 1. Combine the Sarsa algorithm with the crossover operation of the genetic algorithm.

[0167] 2. Use the update formula of the Sarsa algorithm to adjust the Q value and store it in the Q table. These Q values ​​will be used in the pre-training stage of the Q learning algorithm.

[0168] 3. Determine whether the algorithm switching requirements are met based on specific transition conditions. Once the conditions are met, the algorithm switches from the Sarsa algorithm to the Q-Learning algorithm. The Q-Learning algorithm can be switched after accumulating enough sample data. The switching condition depends on the number of Q values, which is determined by the number of states and actions. The switching condition is as follows:

[0169]

[0170] Among them, N i Indicates the number of current iterations, N s Indicates the total number of states, N a Indicates the total number of actions;

[0171] After the initial stage, the Q-Learning algorithm replaces the Sarsa algorithm and is combined with the genetic algorithm, and the Q value is updated by the update formula of the Q-Learning algorithm.

[0172] When the Sarsa algorithm updates the action-value function, it relies on five key parameters: the current state (s), the action selected in the current state (a), the feedback obtained (r), the new state reached after executing action a (s'), and the action to be taken in the new state s' (a'). The update formula of the Sarsa algorithm is as follows:

[0173] Q(s,a):=Q(s,a)+α[R(s,a)+γQ(s′,a′)-Q(s,a)]

[0174] In the Sarsa algorithm, when updating the action value function, instead of the traditional γ multiplied by the maximum Q value (γmaxQ(s′,a′)), the algorithm uses γ multiplied by the Q value (γQ(s′,a′)). This update method allows the algorithm to iterate based on the exact state transition results instead of relying on the expected maximum value.

[0175] Q(s,a) represents the expected return that can be obtained by performing action a (a belongs to action set A) in a given state s (s belongs to state set S). During the execution of the Q-Learning algorithm, decisions are made based on the Q table. The Q table consists of state-action pairs, and the Q value in the table reflects the maximum expected future return that can be obtained by taking a specific action in a specific state. When solving the optimal strategy for the Markov decision process, Q-Learning is different from the Bellman equation. The latter uses the expected value of the next state to estimate the current Q value, while Q-Learning evaluates the Q value of the current state based on the maximum Q value of the next state. Therefore, the update formula of the Q table is as follows:

[0176] Q(s,a):=Q(s,a)+α[R(s,a)+γmaxQ(s′,a′)-Q(s,a)]

[0177] α represents the learning rate, which determines the step size of each Q value update; and γ represents the decay factor of the reward. R(s,a) refers to the immediate reward, that is, the immediate reward obtained by performing action a in state s; and γ multiplied by the maximum value of the Q value of all possible actions a′ in the next state s′ represents the expected long-term reward. The Q-Learning algorithm calculates the maximum value of the Q value of all possible actions a′ in the next state s′, multiplies it by the decay coefficient γ, and adds the immediate reward R(s,a) to form a realistic estimate of the Q value of action a in the current state s. The difference between this realistic estimate and the Q(s,a) value (i.e., Q estimate) already in the Q table is called the error, which is represented by △Q(s,a). The error is the difference between the Q reality and the Q estimate, reflecting the accuracy of the algorithm's prediction. When △Q(s,a) gradually approaches 0, it means that the algorithm has stabilized, that is, converged.

[0178] That is, the process and parameter settings of the improved algorithm are as follows:

[0179] The general process of the improved genetic algorithm integrated with machine learning proposed in this embodiment is as follows:

[0180] ① Initialize the population, the number of chromosomes in the population is P = 50, the initial crossover rate Pc = 0.80, and the mutation rate Pm = 0.05;

[0181] ②Evaluate the overall fitness of the initial population, and the evaluation function is fitness(i)=1 / Tmax(i);

[0182] ③ Perform population division based on K-means clustering, with the number of clusters k = 3, that is, divided into 3 sub-populations;

[0183] ④Evaluate the fitness of each sub-population, and the fitness function remains unchanged;

[0184] ⑤Perform selection operation on each sub-population;

[0185] ⑥ Determine whether the number of iteration steps is the number of steps of the self-learning crossover rate, learning rate α = 0.05, discount factor γ = 0.85, ε = 0.80,

[0186] If the self-learning strategy is satisfied, it is updated according to the cycle C, C = 30, that is, it is updated every 30 steps. If it is not satisfied, the initial crossover rate is directly used for crossover operation;

[0187] ⑦ Determine whether to perform self-learning crossover rate, randomly generate a real number in the interval (0,1), and determine whether it is <0.5. If it is less than 0.5, perform self-learning strategy. If it is greater than 0.5, directly use the initial crossover rate for crossover operation;

[0188] ⑧ Each sub-population performs a crossover operation;

[0189] ⑨Each sub-population undergoes mutation operation;

[0190] ⑩ Determine whether the termination condition is met and set the termination iteration number to 200. If it is met, integrate the optimal solution in each sub-population and then terminate the algorithm. If it is not met, continue to evaluate the fitness of the sub-population and continue iterating.

[0191] In terms of the setting of the crossover rate and mutation rate, this improved algorithm proposes a self-learning strategy based on improved reinforcement learning. This algorithm can not only intelligently predict parameters based on the current population state, but also based on the past state and appropriate future state, and realize the self-learning of parameters. In different improved genetic algorithms, this self-learning strategy can improve the parameters from learning by learning data samples in different environments, so this strategy has good generalization. This strategy is not only suitable for crossover operations, but also for mutation operations. This article only uses this strategy in crossover operations.

[0192] The reinforcement learning model consists of multiple parts, namely, agent, action, state, reward, strategy and environment. The interdependence between these parts constitutes the interaction between the agent and the environment. ①Agent: It is the subject of reinforcement learning training, that is, the subject that takes action; ②Environment: The environment is where the agent exists and is the background of training. The agent takes action under this background; ③Strategy: The strategy is the way the agent takes action. It is not a specific action, but it guides the occurrence of the action. For example, when abstracted as a function, the action of changing the input state to the output is the strategy; ④State: The state refers to the situation in which the agent is in after taking action. It includes the state of the agent and the state of the environment, but the main focus is on the state of the environment, because the state obtained by the agent from the environment is used to reflect the environment; ⑤Action: The action is some operations performed by the agent in the environment. It is the actual behavior of the agent, which will be reflected in the environment and change the state of the environment; ⑥Reward: The reward is the feedback given by the environment after the agent takes action in the current state according to a certain strategy. Feedback can be a reward or a punishment. Among these six elements, the first three parts describe the macroscopic interactive process of reinforcement learning, where the agent obtains results through strategies in the environment; the last three parts are the specific interactive process of reinforcement learning, where an action is taken in a certain state to obtain a reward and then reach the next state, and then take the next action and repeat. The state is obtained from the environment, and the action is taken through the strategy. In this way, the macroscopic and specific processes are linked together to form a complete interactive process of reinforcement learning. In reinforcement learning, the action is the selected crossover rate.

[0193] The mutation process mainly adopts the mutation strategy of two-gene exchange, that is, randomly selecting two unused gene points in the same individual for exchange, thereby producing a new individual.

[0194] Finally, the improved genetic algorithm integrating machine learning and the existing tower task information and process sequence relationship are used to search the tower information table with the shortest construction period as the goal, and the construction process with the shortest construction period in the entire construction process is searched to generate the construction time Gantt chart of all tower processes, such as Figure 3 shown.

[0195] Specifically, the genetic algorithm is used to determine the best order for the construction of line towers within the scope of a contractor's jurisdiction, and the personnel division of labor is arranged according to this order, and the start and end time of each process of each tower is calculated. A Gantt chart of each process in terms of construction time is drawn to intuitively display the construction progress and personnel arrangement.

[0196] This embodiment finally visualizes the final project scheduling results with the help of a Gantt chart. The visualization results of the personnel schedule and process plan in the three tensioning sections that one contractor is responsible for are shown in the figure below: Figure 3As shown in the figure, the vertical axis represents the tower position, and the horizontal axis represents the time axis. The orange block is the road project, the red block is the foundation project, the green block is the tower erection project, and the blue block is the line stringing project. The three tensioning sections are: tower position 1-tower position 12, tower position 13-tower position 26, and tower position 27-tower position 40. It can be seen that the scheduling results meet all constraints.

[0197] This implementation method significantly improves the application effect of genetic algorithms in the scheduling and process optimization of overhead power transmission line personnel by introducing advanced clustering strategies and adaptive learning mechanisms, and provides a new solution to the optimization problems in this field. By optimizing the execution order of the process and reducing unnecessary waiting time, the present invention improves the consistency and fluency of the workflow. By using data analysis and machine learning technology, the scheduling decision is based on data-driven, which improves the scientificity and accuracy of the decision. Finally, the present invention improves the transparency and traceability of the scheduling and workflow through digital scheduling and process management, which is convenient for auditing and quality control. The specific implementation methods described above further explain the purpose, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not used to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for optimizing the personnel scheduling process of overhead power transmission lines based on an improved genetic algorithm, characterized in that: include: Obtain the location of each tower where overhead transmission line construction is required, the existing road information within the construction range of each tower, and the connecting road information between the towers; Obtain the task volume corresponding to the four processes required within the construction scope, as well as the number of construction personnel and construction efficiency of the corresponding processes; Based on the task volume, the number of construction personnel and the construction efficiency, a minimum construction period solution model is constructed with minimization of the maximum total construction period as the objective function and construction resources, the preceding relationship of each process and the completion degree of the process as the constraint conditions; The minimum construction period solution model is solved based on a genetic algorithm, thereby obtaining a construction sequence that minimizes the maximum total construction period of the four construction processes for overhead power transmission line construction on all towers and corresponding personnel scheduling.

2. The method for optimizing the personnel scheduling process of overhead power transmission lines based on an improved genetic algorithm according to claim 1, characterized in that: If the number of poles and towers exceeds the set number threshold, the total construction area corresponding to all the poles and towers will be divided into several tensioning sections, and different contractors will carry out independent road engineering, foundation engineering, and tower erection engineering construction in the construction process of the overhead transmission line.

3. The method for optimizing the personnel scheduling process of overhead power transmission lines based on an improved genetic algorithm according to claim 2, characterized in that: The objective function is: minC max ; W max =max{E 1,h ,AND 2,h ,AND 3,h ,…,AND n,h }; Among them, C max is the maximum completion time set of all towers, E i,h It represents the completion time of tower i in process h, i∈[1,n], n is the number of towers in the tensioning section, and h is the total number of construction processes in the traction section.

4. The method for optimizing the personnel scheduling process of overhead power transmission lines based on an improved genetic algorithm according to claim 1, characterized in that: The constraints are: S i,l+1 ≥E i,l ; Among them, M l represents the number of construction teams in the lth construction process, X i,l,k Indicates that tower i is constructed by construction team k in process l, I = {l, 2, 3, ..., n}, is the set of towers in a tensioning section, n is the number of towers in the tensioning section, K is the set of construction teams responsible for the traction section, E i,l represents the completion time of tower i in process l, S i,l represents the starting time of tower i in process l, T i,l,k represents the construction time of process l of construction team k at tower i, S i,l+1 represents the starting time of tower i in process l+1, D i,l represents the release time of tower i in process l, L represents the process set, Y i,t,l indicates that tower i is constructed in process l, T is the number of time periods, L is the process set, T i,l,k represents the construction time of process l of construction team k at tower i, S i,立塔 Indicates the start time of the tower erection process for tower i, B i,基础,60 represents the time when the erection process of tower i is 60% complete; S i,架线 Indicates the start time of the wiring process for tower i, B i,立塔,100 Indicates the time when the completion rate of the tower erection process of tower i is 100%.

5. The method for optimizing the personnel scheduling process of overhead power transmission lines based on an improved genetic algorithm according to claim 2, characterized in that: The specific method of solving the minimum construction period solution model based on the genetic algorithm is to optimize the construction period for all the tower construction procedures in a tensioning section corresponding to a contractor. The specific process is as follows: A1. Set the number of iterations of the genetic algorithm and the number of chromosomes in each generation, N. A fixed number of n towers are randomly selected from all n towers contracted by a contractor. ′ The pole towers form a chromosome, and are randomly selected without repetition in sequence, thus forming the initial population P, where n = n ′ *N, and the code corresponding to the gene position in the chromosome is the tower number, and the order of the gene position is the order of the construction process of the corresponding tower; A2. Divide N chromosomes into k clusters according to the K-means clustering algorithm, and each cluster is a sub-dataset; A3. Merge the k sub-datasets to form a new data set P ′ ; A4. According to the selection rules, select the new data set P ′ Select chromosomes from the new dataset P ′ Divide into k sub-populations, the selection rule is as follows: the elements in sub-population a are selected according to the data set P ′ The element at the ath position in is used as the basis, and is obtained in an increasing manner with k as the step size; A5. Perform fitness evaluation on the chromosomes in each of the sub-populations, and sort the chromosomes in each of the sub-populations according to the results of the fitness evaluation, and let each of the sub-populations in this iteration process be the parent population, and the chromosomes in the parent population be the parent individuals; A6, save the parent individual with the largest result of the first R fitness evaluations in the parent population, select the parent individual from all the parent individuals except the parent individual with the largest result of the first R fitness evaluations by roulette wheel to perform crossover mutation operation to obtain the offspring individual, calculate the fitness evaluation results of the offspring individuals after the crossover mutation and sort them, reinsert the offspring individuals into the parent population according to the fitness evaluation results, select the set number of chromosomes to form a new parent population, and then return to A5; A7. Repeat A5-A6 until the termination condition is met, and integrate the optimal solution in each subpopulation.

6. The method for optimizing the personnel scheduling process of overhead power transmission lines based on an improved genetic algorithm according to claim 5, characterized in that: When the chromosome performs a crossover operation, if the number of iterations meets the number of self-learning strategies, the following judgment is made: Randomly generate a real number in the interval (0,1) and determine whether the real number is <0.

5. If it is less than 0.5, use the self-learning strategy to calculate the crossover rate. If it is greater than 0.5, use the initial crossover rate to perform the crossover operation directly. If the number of iterations does not meet the number of self-learning strategies, the initial crossover rate is directly used for crossover operation.

7. The method for optimizing the personnel scheduling process of overhead power transmission lines based on an improved genetic algorithm according to claim 6, characterized in that: In the self-learning strategy, the learning rate α, discount factors γ and ε are obtained by combining the Sarsa algorithm with the Q-Learning algorithm. The switching conditions between the Sarsa algorithm and the Q-Learning algorithm are as follows: when When , the Sarsa algorithm is used to obtain the learning rate α, discount factors γ, ε; when When , the Q-Learning algorithm is used to obtain the learning rate α, discount factors γ, ε; Among them, N i Indicates the number of current iterations, N s Indicates the total number of states, N a Indicates the total number of actions.

8. The method for optimizing the personnel scheduling process of overhead power transmission lines based on an improved genetic algorithm according to claim 6 or 7, characterized in that: In the self-learning strategy, the learning rate α=0.05, the discount factor γ=0.85, and ε=0.

80.

9. The method for optimizing the personnel scheduling process of overhead power transmission lines based on an improved genetic algorithm according to claim 1, characterized in that: The mutation operation is implemented using a two-point exchange strategy.

10. The method for optimizing the personnel scheduling process of overhead power transmission lines based on improved genetic algorithm according to claim 5, characterized in that: The fitness function for fitness evaluation is Among them, T max (i) represents the maximum total construction period of tower i.

Images