Material offline service task scheduling method and system based on improved population algorithm

Abstract

The present invention relates to the field of medical material scheduling, and discloses a material offline service task scheduling method and system based on an improved population algorithm, including the following specific steps: constructing a task scheduling model; initializing the task scheduling model; applying an initial population; decoding the individuals in the population to obtain the distribution plan expressed by each individual, and calculating its fitness value; grouping according to each stallion and several colts as a group based on the fitness value; selecting the individual with the largest fitness value in the current population as the global optimal individual; sequentially performing wild horse grazing and colt mating operations, and team leader operations to update the individuals; applying 2-Opt, flipping and swapping subsequences to perform local search on wild horses, applying a leading function to optimize and update stallions, and applying a following function to optimize and update colts; outputting the optimal solution. The present invention solves the problems of low scheduling efficiency and easy falling into local optimum in the prior art, and has the characteristics of good convergence ability and high optimization efficiency.

Description

Technical Field

[0001] The present invention relates to the field of medical supplies scheduling, and more specifically, to a method and system for scheduling offline service tasks of supplies based on an improved population algorithm. Background Art

[0002] With the increasing severity of population aging, providing more efficient and high-quality medical services to patients in a timely manner is an urgent problem to be solved in current medical services. Therefore, researching and developing advanced and practical medical supplies service task scheduling methods has important economic significance and social value.

[0003] The medical supplies service task scheduling problem can be described as follows: providing medical supplies distribution services for remote offline service patients, using a distribution center, and the distribution center uses multiple identical types of distribution vehicles to distribute medical supplies to several patients; each vehicle departs from the distribution center, transports the medical supplies to all the patients in charge of the vehicle within the service time window required by the patients, and finally returns to the distribution center; it is required that each patient is exactly distributed medical supplies by one vehicle. For all the served patients, it is necessary to ensure that all patients can be served within their corresponding time windows, and it is required that the total driving distance of all vehicles in the distribution plan is the shortest.

[0004] There is a prior art method for scheduling multi-vehicle remote health monitoring offline service tasks. For the problem of scheduling multi-vehicle remote health monitoring offline service tasks, an improved adaptive symbiotic organism search algorithm is proposed to solve it. The specific steps are as follows: establishing a mathematical model for scheduling multi-vehicle remote health monitoring offline service tasks; defining a fitness function; setting the control parameters of the algorithm; initializing; performing gene fine-tuning and individual decoding on the initial population, and calculating the fitness value; determining the global optimal individual; sequentially performing adaptive mutualistic symbiosis operations and adaptive commensalism operations to update the population; performing gene fine-tuning and individual decoding on the updated population, and calculating the fitness value; updating the optimal solution; determining whether the maximum number of iterations is reached; and outputting the optimal solution.

[0005] However, the scheduling efficiency of the scheduling scheme in the prior art for logistics distribution is low and it is easy to fall into local optimality. Therefore, it is necessary to optimize and improve it. Inventing a method for scheduling offline service tasks of medical supplies is a technical problem that urgently needs to be solved in this technical field. Summary of the Invention

[0006] In order to solve the problems of low scheduling efficiency and easy falling into local optimality in the prior art, the present invention provides a method and system for scheduling offline service tasks of supplies based on an improved population algorithm, which has the characteristics of good convergence ability and high optimization efficiency.

[0007] In order to achieve the above object of the present invention, the following technical solutions are adopted:

[0008] Offline service task scheduling method based on improved population algorithm, including the following specific steps:

[0009] Construct a task scheduling model;

[0010] Set the fitness function of individuals; Initialize the task scheduling model;

[0011] Apply the logistic map to generate the initial population of the model, and assign individuals to studs and colts respectively;

[0012] Decode the population individuals, determine the values of each variable in the task scheduling model, determine that the service time window is not violated, obtain the delivery plan expressed by each individual, and calculate its fitness value;

[0013] Based on the fitness value, group by each stud and several colts as a group; Select the individual with the largest fitness value in the current population as the global optimal individual;

[0014] Update individuals by sequentially performing wild horse grazing, colt mating operations, and team leader operations;

[0015] Apply 2-Opt, flipping, and swapping subsequences to perform local search on wild horses;

[0016] Apply the leading function to optimize and update the studs, and apply the following function to optimize and update the colts;

[0017] Update the optimal solution;

[0018] Optimize the global optimal solution using a hybrid improved goose optimization algorithm;

[0019] Iteratively update the optimal solution;

[0020] Output the optimal solution as the optimal offline service task scheduling plan.

[0021] Preferably, the task scheduling model is specifically:

[0022]

[0023] Among them, Dist is the total driving distance of all vehicles; N is the number of patients, and the patients are numbered 1, 2,... in sequence N , and the distribution center number is 0; M is the number of vehicles owned by the distribution center; T ij is patient i , j the distance between i , j = 0, 1,... N ,i≠j , x ijm is a variable that can only be 0 or 1, i , j = 0, 1, ..., N , m = 0, 1, ..., M ; x ijm When = 1, it means the vehicle m travels from the patient i to the patient j ; g i is the demand of the patient i , i = 1, 2, ..., N ; z im is a variable that can only be 0 or 1, i = 0, 1, ..., N ; m = 1, 2, ..., M; z im When = 1, it means the patient i is delivered by the vehicle m ; cap represents the maximum load capacity of each vehicle; v represents the driving speed of the vehicle; t ij represents the time for the vehicle to travel from the patient i to j ; a i represents the left time window of the patient i , b i represents the right time window of the patient i ; s im represents the service time of the patient; Each patient can only be served by one vehicle; Each patient must be served within the given service time window. i Furthermore, the fitness function is specifically:

[0024] Initialize the task scheduling model, specifically:

[0025]

[0026] Set the population size

[0027] , pop_N the maximum number of iterations M_Iter , Iter, the current number of iterations PS, the stud ratio PC ; Calculate the number of studsSta_N = pop_N * PS and the number of ponies foal_N = pop_N *(1 - PS ).

[0028] Furthermore, the initial population of the application logistic mapping generation model is specifically as follows:

[0029] The generated initial population is in real - number coding. Denote the i ( i = 1, 2,..., pop_N ) - th individual as W i :

[0030] , j = 1, 2,..., b

[0031] Among them, w ij is the i - th gene of the j - th individual, .

[0032] Furthermore, decode the population individuals, specifically the steps are as follows:

[0033] By mapping the genes in each individual W i ( i = 1, 2,..., pop_N ) to the integer domain according to the index value, construct ; use a number greater than N to group the components of T i . Each group is an ordered set, corresponding to the delivery plan served by a vehicle.

[0034] Furthermore, update the individuals by sequentially performing wild - horse grazing and pony mating operations, and team - leader operations. The specific steps are as follows:

[0035] For each pony individual foal i ( i = 1, 2,..., pop_N *( 1 - PS )) for grazing:

[0036]

[0037] Among them, P is a vector composed of 0 and 1, , They are all random vectors uniformly distributed within the range of [0, 1]. R 2 is a random number within [0, 1]; SEQ is to satisfy the condition P == 0 of the random vector R 1 The returned index value, where Θ is the dot product; Y is the adaptive mechanism; TR represents the coefficient linearly decreasing from 1 to 0; is the position of the team member Pony j at present, is the position of the team leader Stallion at present;

[0038] Pony mating, specifically:

[0039]

[0040] Among them, is the group i the position of the individual q after leaving the group and re - entering the group i ; is the group j the position of the individual z after leaving the group and re - entering the group j ; is the group k the position of the individual p produced by mating i in the group q and the group j in the z offspring individual, and are their parental individuals respectively;

[0041] Team leader operation, specifically including the following steps:

[0042] Operate on the team leader Stallion Sta i ( i = 1, 2,..., pop_N * PS ) Calculate:

[0043]

[0044] Among them, is the next position of the group i leader, W best is the global optimal value, is the current position of the group i leader, Y is the calculated adaptive parameter, R is a random number in [-2, 2], and π = 3.14;

[0045] Perform communication and leadership selection operations:

[0046]

[0047] Among them, foal G,i is the best member in the group, Dist represents the path distance of the delivery plan; if the best member in the group is better than the group leader, then the group leader and the member are exchanged.

[0048] Furthermore, apply 2-Opt, flip, and swap subsequences to perform local search on the wild horses, specifically including the following steps:

[0049] Perform 2-Opt operation on the wild horse individual, select any two dimension values in the individual for exchange to obtain a new individual, compare the fitness value of the new individual with that of the original individual, if the fitness value of the new individual is larger than that of the original individual, then update the individual, specifically:

[0050] Select in i , j exchange the two dimensions, and after the exchange, it is ;

[0051] Perform a flip operation on the wild horse individual, select any two dimension values in the individual for flipping to obtain a new individual, compare the fitness value of the new individual with that of the original individual, if the fitness value of the new individual is larger than that of the original individual, then update the individual, specifically:

[0052] Select in i , j flip the two dimensions, and after the flip, it is .

[0053] Use the swap subsequence to perform local search on the wild horse individual, specifically:

[0054] Select in i, j, k, l exchange the four dimensions, the first two positions are used as subsequence 1, and the remaining positions are used as subsequence 2, and swap these two subsequences, and after the swap, it is .

[0055] Furthermore, apply the leading function to optimize and update the stud horses, and apply the following function to optimize and update the foal horses. The specific steps are:

[0056] Improve the stud horses by applying the leading function:

[0057]

[0058] For the position of the improved individual, For the global optimal value, For the position of the current value;

[0059] Apply the following function to the pony for improvement:

[0060]

[0061] foal i+1 For the position of the improved individual of the pony, θ For the included angle between the pony's position relative to the group leader and the global optimal individual, taking [0, 45°], ;

[0062] Continuously update the optimal solution.

[0063] Furthermore, use the hybrid improved goose optimization algorithm to optimize the global optimal solution. The specific steps are as follows:

[0064] ;

[0065]

[0066] Among them, randn () is a random function that follows the standard normal distribution; ; T Represents the time parameter; obtain the new W best Exchange with the worst stud horse, and at the same time compare with the current optimal value. If it is better than the optimal value, update the optimal value.

[0067] The offline material service task scheduling system based on the improved population algorithm includes a model construction module, an initialization module, a fitness value calculation module, an iterative optimization module, a local search module, a leading function and following function optimization module, a hybrid improved goose optimization algorithm module, and an output optimal result module;

[0068] The described model construction module is used to construct a task scheduling model;

[0069] The initialization module is used to set the fitness function of the individual; initialize the task scheduling model; apply the logistic mapping to generate the initial population of the model, and assign the individuals to the stud horses and ponies respectively;

[0070] The described fitness value calculation module is used to decode the individuals in the population, determine the values of each variable in the task scheduling model, determine that the service time window is not violated, obtain the distribution plan expressed by each individual, and calculate its fitness value;

[0071] The described iterative optimization module is used to group based on the fitness value, with each stallion and several ponies as a group; select the individual with the largest fitness value in the current population as the global optimal individual;

[0072] The described local search module is used to perform local search on the wild horses using 2-Opt, flipping, and swapping subsequences;

[0073] The described leading function and following function optimization module is used to optimize and update the stallions using the leading function and optimize and update the ponies using the following function;

[0074] The described hybrid improved goose optimization algorithm module is used to optimize the global optimal solution using the hybrid improved goose optimization algorithm;

[0075] The described output optimal result module is used to output the optimal solution as the optimal offline service task scheduling plan.

[0076] The beneficial effects of the present invention are as follows:

[0077] The present invention discloses a method for scheduling offline service tasks of materials based on an improved population algorithm. By constructing a task scheduling model, constructing a population, decoding the population and solving its fitness value, and iteratively solving the optimal fitness value, the optimal solution to the problem of scheduling offline service tasks of medical materials is obtained; the present invention applies 2-Opt, flipping, and swapping subsequences to perform local search on wild horses, applies the leading function to optimize and update stallions, and applies the following function to optimize and update ponies. Compared with the prior art, it can be optimized faster and more accurately, and has the characteristics of fast running speed, not easily falling into local optimum, and good convergence ability. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1 It is a flow diagram of the method for scheduling offline service tasks of materials based on the improved population algorithm of the present invention;

[0079] Figure 2 It is a schematic diagram of the optimal solution obtained by using the method for scheduling offline service tasks of materials based on the improved population algorithm of the present invention to solve the embodiment.

[0080] Figure 3 It is a schematic diagram comparing the convergence speed obtained by using the method for scheduling offline service tasks of materials based on the improved population algorithm of the present invention to solve the embodiment with the convergence speed of the genetic algorithm (GA) and the ivy algorithm (IVYA).

[0081] Figure 4 This is a schematic diagram of the offline material service task scheduling system based on the improved population algorithm of the present invention. Specific implementation manners

[0082] The present invention will be described in detail below with reference to the accompanying drawings and specific implementation manners.

[0083] Example 1

[0084] As Figure 1 shown, the offline material service task scheduling method based on the improved population algorithm includes the following specific steps:

[0085] Construct a task scheduling model;

[0086] Set the fitness function of the individual; initialize the task scheduling model;

[0087] Apply the logistic mapping to generate the initial population of the model, and assign individuals to studs and colts respectively;

[0088] Decode the population individuals, determine the values of the variables in the task scheduling model, determine that the service time window is not violated, obtain the delivery plan expressed by each individual, and calculate its fitness value;

[0089] Based on the fitness value, group them in groups of each stud and several colts; select the individual with the largest fitness value in the current population as the global optimal individual;

[0090] Update the individuals by sequentially performing wild horse grazing and colt mating operations, and team leader operations;

[0091] Apply 2-Opt, flipping, and swapping subsequences to perform local search on wild horses;

[0092] Apply the leading function to optimize and update the studs, and apply the following function to optimize and update the colts;

[0093] Update the optimal solution;

[0094] Optimize the global optimal solution using a hybrid improved goose optimization algorithm;

[0095] Iteratively update the optimal solution;

[0096] Output the optimal solution as the optimal offline service task scheduling plan.

[0097] Example 2

[0098] In a specific embodiment, the task scheduling model is specifically:

[0099]

[0100] Among them, Dist is the total driving distance of all vehicles; N is the number of patients, and the patients are numbered 1, 2, …, N , and the distribution center is numbered 0; M is the number of vehicles owned by the distribution center; T ij is for patient i , j the distance between, i , j = 0, 1, ..., N , i≠j ,; x ijm is a variable that is either 0 or 1, i , j = 0, 1, ..., N , m = 0, 1, ..., M ; x ijm When = 1, it means that vehicle m travels from patient i to patient j ; g i is the demand of patient i , i = 1, 2, ..., N ; z im is a variable that is either 0 or 1, i = 0, 1, ..., N ; m = 1, 2, ..., M; z im When = 1, it means that patient i is delivered by vehicle m ; cap represents the maximum load capacity of each vehicle; v represents the driving speed of the vehicle; t ij represents the time for the vehicle to travel from patient i to j ; a i represents the left time window of patient i , b i represents the right time window of patient i ; s im represents patient iService time; each patient can only be served by one vehicle; each patient must be served within the given service time window.

[0101] In a specific embodiment, the fitness function is specifically:

[0102]

[0103] In this embodiment, initializing the task scheduling model specifically includes: setting the population size popN = 50, the maximum number of iterations M_I = 300, the current iteration number Iter, elite ratio PS, mating probability PC the current iteration number Iter = 1, elite ratio PS = 0.2, mating probability PC = 0.13, calculating the number of elites S ta_N = popN * PS and the number of foals foal_N = popN *(1 - PS ).

[0104] Initializing the task scheduling model specifically includes:

[0105] Setting the population size pop_N 、the maximum number of iterations M_Iter 、the current iteration number Iter, elite ratio PS, mating probability PC ; calculating the number of elites Sta_N = pop_N * PS and the number of foals foal_N = pop_N *(1 - PS ).

[0106] In a specific embodiment, applying the logistic mapping to generate the initial population of the model, the specific steps are:

[0107] The generated initial population is in real - number encoding. Denote the i ( i = 1, 2,..., pop_N ) - th individual as W i :

[0108] , j = 1, 2,..., b

[0109] Wherein,w ij For the i th gene of the j th individual, .

[0110] In a specific embodiment, decoding the population individuals specifically includes the steps of:

[0111] By mapping the genes in each individual W i ( i = 1, 2,..., pop_N ) to the integer domain according to the index value, constructing ; using a number greater than N to group the T i components, and each group is an ordered set corresponding to the delivery plan served by a vehicle.

[0112] In a specific embodiment, updating the individuals by sequentially performing wild horse grazing, pony mating operations, and team leader operations, the specific steps are:

[0113] For each pony individual foal i ( i = 1, 2,..., pop_N * ( 1 - PS )) for grazing:

[0114]

[0115] Among them, P is a vector composed of 0 and 1, , are both random vectors uniformly distributed within the range of [0, 1], R 2 is a random number within [0, 1]; SEQ is a random vector that satisfies the condition P == 0, R 1 is the index value returned by Y is the adaptive mechanism; TR represents the coefficient that linearly decreases from 1 to 0; is the current position of the pony j in the group, is the current position of the stallion, the team leader;

[0116] Pony mating, specifically:

[0117]

[0118] Among them, For the group i The individuals in q After leaving the group and entering the group again i The individual positions of For the group j The individuals in z After leaving the group and entering the group again j The individual positions of; For the group k The individuals in p From the group i Among the q And the group j Among the z The offspring individuals produced by mating, And Are their parental individuals respectively;

[0119] The team leader operates, specifically including the following steps:

[0120] Operate on the team leader stallion Sta i ( i = 1, 2,..., pop_N * PS ) Calculate:

[0121]

[0122] Among them, Is the next position of the group i Leader, W best Is the global optimal value, Is the current position of the group i leader, Y is the calculated adaptive parameter, R is a random number in [-2, 2], π = 3.14;

[0123] Carry out communication and selection of the leader operation:

[0124]

[0125] Among them, foal G,i Is the optimal member in the group, Dist Represents the path distance of the delivery plan; if the optimal member in the group is better than the group leader, the group leader and the member are exchanged.

[0126] In a specific embodiment, 2-Opt, flipping and swapping subsequences are applied to perform local search on wild horses, specifically including the following steps:

[0127] Perform the 2-Opt operation on the wild horse individuals. Select any two-dimensional values in the individual for exchange to obtain a new individual. Compare the fitness value of the new individual with that of the original individual. If the fitness value of the new individual is greater than that of the original individual, update the individual. Specifically:

[0128] Select in i and j Exchange the two dimensions. After the exchange, it is ;

[0129] Perform the flip operation on the wild horse individuals. Select any two-dimensional values in the individual for flipping to obtain a new individual. Compare the fitness value of the new individual with that of the original individual. If the fitness value of the new individual is greater than that of the original individual, update the individual. Specifically:

[0130] Select in i and j Flip the two dimensions. After the flip, it is .

[0131] Perform local search on the wild horse individuals by swapping subsequences. Specifically:

[0132] Select in i, j, k, l Exchange the four dimensions. The first two positions are used as subsequence 1, and the remaining positions are used as subsequence 2. Then swap these two subsequences. After the swap, it is .

[0133] In a specific embodiment, apply the leading function to optimize and update the stud horses, and apply the following function to optimize and update the foal horses. The specific steps are:

[0134] Improve the stud horses by applying the leading function:

[0135]

[0136] is the position of the improved individual, is the global optimal value, is the position of the current value;

[0137] Improve the foal horses by applying the following function:

[0138]

[0139] foal i+1 is the position of the improved individual of the foal horses, θ is the included angle between the foal horses and the group leader and the global optimal individual, taking [0, 45°], ;

[0140] Continuously update the optimal solution.

[0141] In a specific embodiment, a hybrid improved goose optimization algorithm is used to optimize the global optimal solution. The specific steps are as follows:

[0142] ;

[0143]

[0144] Among them, randn () is a random function that follows a standard normal distribution; ; T represents the time parameter, taking values in [0.03, 0.5]; obtaining a new W best Exchange with the worst stallion, and at the same time compare with the current optimal value. If it is better than the optimal value, update the optimal value.

[0145] In this embodiment, the distribution center, patient information, and the output optimal solution are shown in Table 1 and Figure 2 shown as follows; The comparison of the convergence speeds of the method of the present invention with the genetic algorithm (GA) and the ivy algorithm (IVYA) is as Figure 3 shown as follows:

[0146] Table 1 Information of the distribution center and patients in the embodiment

[0147]

[0148] In summary, the present invention proposes a discrete wild horse optimization algorithm solution using a hybrid improved goose optimization algorithm. The present invention can effectively solve a medical material offline service task scheduling problem, and has the characteristics of fast running speed, not easy to fall into local optimum, good convergence ability, high optimization efficiency, etc.

[0149] Embodiment 3

[0150] As Figure 4 shown, the material offline service task scheduling system based on the improved population algorithm includes a model construction module, an initialization module, a fitness value calculation module, an iterative optimization module, a local search module, a leading function and following function optimization module, a hybrid improved goose optimization algorithm module, and an output optimal result module;

[0151] The described model construction module is used to construct a task scheduling model;

[0152] The initialization module is used to set the fitness function of the individual; initialize the task scheduling model; apply the logistic mapping to generate the initial population of the model, and assign the individuals to the stallions and ponies respectively;

[0153] The described fitness value calculation module is used to decode the individuals in the population, determine the values of each variable in the task scheduling model, determine that the service time window is not violated, obtain the distribution plan expressed by each individual, and calculate its fitness value;

[0154] The described iterative optimization module is used to group according to each stud horse and several foals based on the fitness value; select the individual with the largest fitness value in the current population as the global optimal individual;

[0155] The described local search module is used to perform local search on the wild horses by applying 2-Opt, flipping, and swapping subsequences;

[0156] The described leading function and following function optimization module is used to optimize and update the stud horses by applying the leading function, and optimize and update the foals by applying the following function;

[0157] The described hybrid improved goose optimization algorithm module is used to optimize the global optimal solution by using the hybrid improved goose optimization algorithm;

[0158] The described output optimal result module is used to output the optimal solution as the optimal offline service task scheduling plan.

[0159] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, rather than limiting the implementation manners of the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the claims of the present invention.

Claims

1. A material offline service task scheduling method based on an improved population algorithm, characterized by: The specific steps include: Build a task scheduling model; The task scheduling model is specifically as follows: in, Dist is the total distance travelled by all vehicles; N is the number of patients, and the patients are numbered 1, 2, ..., N , the distribution center number is 0; M The number of vehicles owned by the distribution center; T ij For patients i , j The distance between i , j =0,1,... , N , i ≠j ,; x ijm is a variable that is either 0 or 1. i , j =0,1,..., N , m =0,1,..., M ; x ijm =1 indicates vehicle m From the patient i Driving to the patient j ; g i For patients i The demand, i =1,2,..., N ; z im is a variable that is either 0 or 1. i =0,1,..., N ; m =1,2,..., M;z im =1 means the patient i By vehicle m delivery; cap Indicates the maximum cargo capacity of each vehicle; v Indicates the vehicle's speed; t ij Indicates that the vehicle is from the patient i arrive j time; a i Indicates patient i The left time window of b i Indicates patient i The right time window of s im Indicates patient i service time; each patient can only be served by one vehicle; each patient must be served within the given service time window; Set individual fitness function; initialize task scheduling model; The fitness function is specifically: Apply logistic mapping to generate the initial population of the model and assign individuals to stallions and ponies respectively; Decode the individuals in the population, determine the values ​​of the variables in the task scheduling model, make sure that the service time window is not violated, obtain the delivery plan expressed by each individual, and calculate its fitness value; Based on the fitness value, each stallion is grouped with a number of ponies; the individual with the largest fitness value in the current population is selected as the global optimal individual; Individuals are updated by sequentially performing wild horse grazing and pony mating operations, and team leader operations; Apply 2-Opt, flip and swap subsequences to perform local search on wild horses; Apply 2-Opt, flip and swap subsequences to perform local search for wild horses, which includes the following steps: Perform 2-Opt operation on wild horse individuals, select the values ​​of any two dimensions in the individual and exchange them to obtain a new individual, compare the fitness value of the new individual with the original individual, and if the fitness value of the new individual is greater than that of the original individual, update the individual, specifically: Select the initial population In i , j The individuals in the two dimensions are exchanged, and after the exchange, ;in, w ij For the i The individual j genes, , N is the number of patients, M The number of vehicles owned by the distribution center; Perform a flip operation on the wild horse individuals, select the values ​​of any two dimensions in the individuals and flip them to obtain new individuals. Compare the fitness values ​​of the new individuals with those of the original individuals. If the fitness value of the new individual is greater than that of the original individual, the individual will be updated. Specifically: Select In i , j The two dimensions are flipped and the flipped : The exchange subsequence performs a local search on the wild horse individuals, specifically: Select In i, j, k, l The four dimensions are exchanged, the first two positions are used as subsequence 1, and the remaining positions are used as subsequence 2, and the two subsequences are exchanged. ; Apply the leading function to optimize and update the stallion, and apply the following function to optimize and update the pony; Apply the leading function to optimize and update the stallion, and apply the following function to optimize and update the pony. The specific steps are: Improvements to the stallion application leading function: is the position of the individual after improvement, is the global optimal value, is the position of the current value; Improve the follow function applied to the pony: foal i+1 The improved individual position for the pony, θ is the angle between the pony's position relative to the group leader and the global optimal individual, which is [0,45°], ; Continuously update the optimal solution; Update the optimal solution; The hybrid improved goose optimization algorithm is used to optimize the global optimal solution; Iteratively update the optimal solution; Output the optimal solution as the optimal offline service task scheduling plan.

2. The method for scheduling offline material service tasks based on an improved population algorithm according to claim 1 is characterized in that: Initialize the task scheduling model, specifically: Setting the population size pop_N , maximum number of iterations M_Iter , current iteration number Iter, Stallion ratio PS, Mating probability PC ; Calculate the number of stallions Sta_N = pop_N * PS and the number of ponies foal_N = pop_N *(1- PS ).

3. The method for scheduling offline material service tasks based on an improved population algorithm according to claim 2 is characterized in that: Apply logistic mapping to generate the initial population of the model. The specific steps are: The generated initial population is a real number encoding method, record i ( i =1,2,..., pop_N ) individuals are W i : ， j =1,2,..., b in, w ij For the i The individual j genes, .

4. The method for scheduling offline material service tasks based on an improved population algorithm according to claim 3 is characterized in that: Decode the individuals in the population, specifically the steps: By making each individual W i ( i =1,2,..., pop_N ) are mapped to the integer domain according to the index value, constructing ; greater than N A few T i The components are grouped, each group is an ordered set, corresponding to a delivery plan for a vehicle service.

5. The method for scheduling offline material service tasks based on an improved population algorithm according to claim 4 is characterized in that: Individuals are updated by sequentially executing wild horse grazing, pony mating, and team leader operations. The specific steps are: For each individual pony foal i ( i =1,2,..., pop_N *( 1-PS )) For grazing: in, P is a vector consisting of 0s and 1s. , are all random vectors uniformly distributed in the range [0, 1]. R 2 is a random number in [0, 1]; SEQ To meet the conditions P ==0 random vector R 1 returns the index value, Θ is the dot product; Y It is an adaptive mechanism; TR Represents a coefficient that decreases linearly from 1 to 0; For group member Xiao Ma j Current location, Lead the stallion for the group in his current position; Pony mating, specifically: in, For Group i Medium q Rejoining a group after leaving the group i The individual position of For Group j Medium z Rejoining a group after leaving the group j individual location; For Group k Medium p By Group i In q and Group j In z The offspring produced by mating, and are their parent individuals respectively; Team leader operations include the following steps: Stallion Operations on Team Leaders Sta i ( i =1,2,..., pop_N * PS )calculate: in, Yes Group i The leader's next position, W best is the global optimal value, is the current position of the leader of group i, Y is the calculated adaptive parameter, R is a random number in [-2, 2], π=3.14; Conduct communication and select leadership operations: in, foal G,i The best member of the group, Dist Represents the path distance of the distribution plan; if the best member in the group is better than the group leader, the group leader and member will be exchanged.

6. The method for scheduling offline material service tasks based on an improved population algorithm according to claim 5 is characterized in that: The hybrid improved goose optimization algorithm is used to optimize the global optimal solution. The specific steps are as follows: ； in, randn () is a random function that obeys the standard normal distribution; T Represents the time parameter; get the new W best Exchange with the worst stallion and compare with the current optimal value. If it is better than the optimal value, update the optimal value.

7. The offline material service task scheduling system based on the improved population algorithm is characterized by: Used to implement the offline material service task scheduling method as described in any one of claims 1 to 6, comprising a model building module, an initialization module, a fitness value calculation module, an iterative optimization module, a local search module, a leading function and a following function optimization module, a hybrid improved goose optimization algorithm module, and an optimal result output module; The model building module is used to build a task scheduling model; The initialization module is used to set the fitness function of the individual; initialize the task scheduling model; apply logistic mapping to generate the initial population of the model, and assign the individuals to the stallion and the pony respectively; The fitness value calculation module is used to decode the population individuals, determine the values ​​of the variables in the task scheduling model, determine that the service time window is not violated, obtain the delivery plan expressed by each individual, and calculate its fitness value; The iterative optimization module is used to group each stallion and a number of ponies into a group based on the fitness value; select the individual with the largest fitness value in the current population as the global optimal individual; The local search module is used to perform local search on wild horses by applying 2-Opt, flipping and swapping subsequences; The leading function and following function optimization module is used to apply the leading function to optimize and update the stallion, and apply the following function to optimize and update the pony; The hybrid improved goose optimization algorithm module is used to optimize the global optimal solution using the hybrid improved goose optimization algorithm; The optimal result output module is used to output the optimal solution as the optimal offline service task scheduling solution.

Images