Collaborative Scheduling Method for High-End Equipment Prototyping and Testing Considering Process Uncertainty

Abstract

The present invention provides a collaborative scheduling method for the trial production and testing of high-end equipment considering process uncertainty, which relates to the technical field of task scheduling. The planned completion time of the test considered in the present invention is a time interval, that is, the process time is uncertain. In the case of uncertain process time, the collaborative scheduling of the high-end equipment trial production and testing process is carried out, and the part process requirements and the machining sequence of the parts are comprehensively considered to dynamically and accurately obtain an approximate optimal solution to the problem. It can effectively collaborate and optimize the resources in the two stages of high-end equipment trial production and testing under the condition of considering process uncertainty, so as to minimize the input cost of the trial production and testing process and improve the resource utilization efficiency and collaboration efficiency of high-end equipment enterprises.

Description

Technical Field

[0001] The present invention relates to the technical field of task scheduling, and particularly relates to a collaborative scheduling method and system for trial production and testing of high-end equipment considering process uncertainty. Background Art

[0002] After the design of high-end equipment is completed, a series of technical tests are required to assess the structure, load-bearing capacity, safety, etc. of the equipment before it can be officially put into use or mass-produced. Before the test, some samples need to be trial-produced according to the design scheme, and then the test is carried out. During the trial production stage, different tasks to be produced need to be assigned. After manufacturing, the test is carried out according to the requirements of different test parts. In the actual process of trial production and testing, the execution of trial production and testing of some parts often proceeds simultaneously, and due to the complex structure and strong professionalism of high-end equipment, it is often difficult to accurately estimate the execution cycle of the test.

[0003] The existing solutions to the two-stage optimization problem of trial production - testing mainly rely on simulation methods, heuristic methods or artificial intelligence algorithms. Most of the considered constraints focus on the production time of parts or the sequence of test tasks, and less attention is paid to other situations in the actual process, resulting in the inability to accurately guide the process of trial production and testing of high-end equipment, thus causing the occupation or waste of various resources. Summary of the Invention

[0004] (1) Technical Problems to be Solved

[0005] Aiming at the deficiencies of the prior art, the present invention provides a collaborative scheduling method and system for trial production and testing of high-end equipment considering process uncertainty, and solves the technical problem of collaborative scheduling optimization in the process of trial production and testing of high-end equipment considering process uncertainty.

[0006] (2) Technical Solutions

[0007] To achieve the above object, the present invention is realized through the following technical solutions:

[0008] In the first aspect, the present invention provides a collaborative scheduling method for trial production and testing of high-end equipment considering process uncertainty, including:

[0009] S1. Set the input parameters of the hybrid genetic algorithm based on neighborhood search according to the part data, factory data, task data during the test process, and technical personnel data during the test process in the trial production process of high-end equipment. Among them, the input parameters include: the planned completion time interval e of the test t , the manufacturing process C = {c1, c2,..., c s} of each part; the first-piece processing cycle and the subsequent processing cycle d i;

[0010] S2. Obtain the global optimal solution according to the input parameters and the hybrid genetic algorithm based on neighborhood search. Obtain the set of parts allocated to each factory in the trial production stage, the processing sequence of the parts within the factory, and the execution sequence of subsequent test tasks according to the global optimal solution, and assign technical personnel to each test task.

[0011] Preferably, obtaining the global optimal solution according to the input parameters and the hybrid genetic algorithm based on neighborhood search includes:

[0012] S21. Set the execution parameters of the hybrid genetic algorithm based on neighborhood search;

[0013] S22. Generate an initial population. The individuals in the initial population are represented by two-stage coding PS-FS and TS-PN. Decode the two-stage coding to obtain the fitness value of the individual, and obtain and record the current optimal solution;

[0014] S23. Based on the fitness values of the individuals in the population, perform a selection operation using the roulette wheel method;

[0015] S24. Based on the crossover probability, perform a crossover operation on the population individuals selected in S23 to obtain a new population with the same scale as the initial population;

[0016] S25. Decode the individuals in the current population to obtain the fitness value C max , record the best individual as π, select a neighborhood structure NS based on the roulette wheel method k , and generate a neighborhood solution π' of π according to this neighborhood structure NS k ;

[0017] S26. Use the neighborhood structure NS K to perform local search on the solution π' to obtain the local optimal solution π”. Compare the local optimal solution π” with the optimal solution π of the variable neighborhood search algorithm. If π” is better than π, then let π = π”, and at the same time record the successful times of the local optimal solution searched by this neighborhood structure. Otherwise, increase the failure times of the local optimal solution searched by this neighborhood structure, update the selection probability of the neighborhood structure NS K , and determine whether gen2 ≤ Gen2 holds. If so, assign gen2 + 1 to gen2, and return to step S25. Otherwise, execute step S27;

[0018] S27. Compare the optimal solution π of the variable neighborhood search algorithm with the worst individual in the initial population X0. If π is better than the worst individual in the worst initial population, then replace this solution with π, and compare π with the global optimal solution π best , if π is better than π best , then let πbest = π;

[0019] S28. Determine whether gen1 ≤ Gen1 holds. If so, assign gen1 + 1 to gen1, use the updated population as the input for the next iteration, and return to step S23. Otherwise, output the global optimal solution π best .

[0020] Preferably, the generating the initial population, calculating the fitness value of each individual, obtaining and recording the current optimal solution includes:

[0021] Based on the hybrid algorithm of genetic and variable neighborhood search, encode the part and factory data in the high-end equipment trial production process and the task and technician data in the test process to obtain an initial population X0 composed of PopSize initial solutions. Each solution in X0 is composed of two sequences PS-FS and TS-PN. Among them, the PS-FS encoding represents the corresponding relationship and sequence order between parts and factories in the trial production stage, and TS-PN represents the assignment order of test tasks and technicians in the test stage; decode each initial solution in the initial population to obtain the fitness value C max of the population individuals, and this value represents the total time span of the trial production and test processes; record the best individual among them as the current global optimal solution π best , and set the initial iteration number gen1 = 1 of the hybrid optimization algorithm, and set the initial iteration number gen2 = 1 of the variable neighborhood search.

[0022] Preferably, the hybrid genetic algorithm based on neighborhood search encodes the part and factory data in the high-end equipment trial production process and the task and technician data in the test process, including:

[0023] Based on the characteristics of the two key stages of part trial production and combined test, the solution X is composed of the solution X1 in the trial production stage and the solution X2 in the test stage. The solutions in each stage are represented by two-dimensional integer arrays; the solution X1 contains two parts, the part processing order sequence PS and the factory list FS, and the length of the sequence or list is The numbers in PS represent the part numbers, and the number of times the numbers appear represents the number of parts; FS represents the factory number, and each column in the array represents that the corresponding numbered part is assigned to the corresponding factory for trial production; the solution X2 is composed of the test execution order sequence TS and the personnel list PN, and the length of the sequence or list is k; TS and PN are respectively composed of test task and technician numbers, and each column in the array represents that the corresponding numbered test task is executed by the corresponding technician.

[0024] Preferably, the selection operation is performed by roulette wheel method based on the fitness value of each individual in the population, including:

[0025] S2301. Decode to obtain the total span time of the trial production and test process for each initial solution, and denote this value as f(x), i.e., let f(x) = C max , calculate each individual x in the initial population X0 i The probability of being inherited to the next generation population

[0026] S2302. Calculate the cumulative probability of each individual in X0

[0027] S2303. Define a variable Q, generate a random number in the interval (0,1), assign this random number to the variable Q, determine the interval to which the variable Q belongs. If Qx i-1 <Q<Qx i , then select the individual x in the initial population X0 i ;

[0028] S2304. Repeat step S2303 until PopSize population individuals are selected.

[0029] Preferably, use the neighborhood structure NS K Perform local search on the solution π' to obtain the local optimal solution π”, compare the local optimal solution π” with the variable neighborhood search algorithm optimal solution π. If π” is better than π, then let π = π”, and at the same time record the successful times of the local optimal solution searched by this neighborhood structure. Otherwise, increase the failure times of the local optimal solution searched by this neighborhood structure, and update the selection probability of the neighborhood structure NS K including:

[0030] S2601. The neighborhood structure NS of the variable neighborhood search algorithm K (K = 1, 2, …, K max ), in the stage of setting algorithm parameters, set the initial selection probability SP of each neighborhood structure K to be the same, where

[0031] S2602. During the local search process of the variable neighborhood search algorithm, record the number of times S that each neighborhood structure successfully searches to the local optimal solution K and the number of failures L K , update the selection probability of each neighborhood according to the following formula: where SUC K represents the local search success rate of the neighborhood structure NS K , and the calculation formula is Δ is a sufficiently small positive number;

[0032] S2603. Define a variable Q', generate a random number in the interval (0,1), assign this random number to the variable Q', determine the interval to which the variable Q' belongs. If SUCK-1 <Q' < SUC K , then select the neighborhood structure NS K as the type of neighborhood search.

[0033] In a second aspect, the present invention provides a collaborative scheduling system for high-end equipment trial production and testing considering process uncertainty, including:

[0034] An input parameter acquisition module, configured to set input parameters of a hybrid genetic algorithm based on neighborhood search according to part data, factory data, task data during the testing process, and technician data during the testing process in the high-end equipment trial production process, where the input parameters include: the planned completion time interval e of the test t , the manufacturing process C of each part = {c1, c2,..., c s}; the first-piece processing cycle of the part and the subsequent processing cycle d i ;

[0035] A solution module, configured to obtain a global optimal solution according to the input parameters and the hybrid genetic algorithm based on neighborhood search, obtain the set of parts allocated to each factory in the trial production stage, the processing order of the parts in the factory, the execution order of the subsequent test tasks according to the global optimal solution, and assign technicians to each test task.

[0036] Preferably, obtaining the global optimal solution according to the input parameters and the hybrid genetic algorithm based on neighborhood search includes:

[0037] S21. Set the execution parameters of the hybrid genetic algorithm based on neighborhood search;

[0038] S22. Generate an initial population. The individuals in the initial population are represented by two-stage encoding PS-FS and TS-PN. Decode the two-stage encoding to obtain the fitness value of the individual, and obtain and record the current optimal solution;

[0039] S23. Based on the fitness values of the individuals in the population, perform a selection operation using the roulette wheel method;

[0040] S24. Based on the crossover probability, perform a crossover operation on the population individuals selected in S23 to obtain a new population with the same scale as the initial population;

[0041] S25. Decode the individuals in the current population to obtain the fitness values C max of each individual, record the best individual as π, select a neighborhood structure NS based on the roulette wheel method k , and generate a neighborhood solution π' of π according to the neighborhood structure NS k ;

[0042] S26. Use the neighborhood structure NS K Perform a local search on the solution π' to obtain the local optimal solution π”. Compare the local optimal solution π” with the optimal solution π of the variable neighborhood search algorithm. If π” is better than π, then let π = π”, and record the number of successful times that the local optimal solution is found by this neighborhood structure. Otherwise, increase the number of failed times that the local optimal solution is found by this neighborhood structure, and update the neighborhood structure NS K of the selection probability, and determine whether gen2 ≤ Gen2 holds. If so, assign gen2 + 1 to gen2, and return to step S25. Otherwise, execute step S27;

[0043] S27. Compare the optimal solution π of the variable neighborhood search algorithm with the worst individual in the initial population X0. If π is better than the worst individual in the worst initial population, then replace this solution with π, and compare π with the global optimal solution π best for comparison. If π is better than π best , then let π best = π;

[0044] S28. Determine whether gen1 ≤ Gen1 holds. If so, assign gen1 + 1 to gen1, and use the updated population as the input for the next iteration, and return to step S23. Otherwise, output the global optimal solution π best .

[0045] (III) Beneficial effects

[0046] The present invention provides a collaborative scheduling method and system for the trial production and testing of high-end equipment considering process uncertainty. Compared with the prior art, it has the following beneficial effects:

[0047] The present invention considers that the planned completion time of the test is a time interval, that is, the process time is uncertain. In the case of uncertain process time, the collaborative scheduling of the high-end equipment trial production and testing process is carried out, and the part process requirements and the machining sequence of the parts are comprehensively considered, and the approximate optimal solution of the problem is dynamically and accurately obtained, which can effectively collaborate and optimize the resources in the two stages of high-end equipment trial production and testing, so as to minimize the input cost in the trial production and testing process and improve the resource utilization efficiency and collaboration efficiency of high-end equipment enterprises. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0049] Figure 1 It is a block diagram of the collaborative scheduling method for the trial production and testing of high-end equipment considering process uncertainty in the embodiments of the present invention;

[0050] Figure 2 It is a flowchart of the collaborative scheduling method for the trial production and testing of high-end equipment considering process uncertainty in the embodiments of the present invention;

[0051] Figure 3 It is a coding schematic diagram in the embodiments of the present invention. Specific embodiments

[0052] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0053] By providing a collaborative scheduling method and system for the trial production and testing of high-end equipment considering process uncertainty in the embodiments of the present application, the resources in the two stages of the trial production and testing of high-end equipment are collaboratively optimized and configured, thereby improving the resource utilization efficiency and collaboration efficiency of high-end equipment enterprises.

[0054] The technical solutions in the embodiments of the present application for solving the above technical problems are generally as follows:

[0055] The solution to the existing two-stage optimization problem of trial production - testing mainly relies on simulation methods, heuristic methods, or artificial intelligence algorithms, ignoring the process requirements in the actual process, the uncertainty of the machining sequence of parts, and the testing time, which will not be able to accurately and efficiently guide the process of high-end equipment trial production and testing, thus causing the occupation or waste of various resources. In the embodiments of the present invention, when considering the uncertainty of process time and performing collaborative scheduling on the process of high-end equipment trial production and testing, the part process requirements and the machining sequence of parts are comprehensively considered, and an approximate optimal solution to the problem is dynamically and accurately obtained, which can effectively collaboratively optimize and configure the resources in the two stages of high-end equipment trial production and testing, thereby minimizing the input cost of the trial production and testing process and improving the resource utilization efficiency and collaboration efficiency of high-end equipment enterprises.

[0056] To better understand the above technical solutions, the above technical solutions will be described in detail below in conjunction with the accompanying drawings of the specification and specific embodiments.

[0057] The embodiments of the present invention provide a collaborative scheduling method for the trial production and testing of high-end equipment considering process uncertainty, as Figure 1 shown, including:

[0058] S1. Set the input parameters of the hybrid genetic algorithm based on neighborhood search according to the part data, factory data, task data during the test process, and technician data during the test process in the high-end equipment trial production process. Among them, the input parameters include: the planned completion time interval e of the test t , the manufacturing process C of each part = {c1, c2, …, c s}; the first-piece processing cycle of the part and the subsequent processing cycle d i ;

[0059] S2. Obtain the global optimal solution according to the input parameters and the hybrid genetic algorithm based on neighborhood search. Obtain the set of parts allocated to each factory in the trial production stage, the processing order of the parts in the factory, and the execution order of the subsequent test tasks according to the global optimal solution, and assign technicians to each test task.

[0060] The embodiment of the present invention considers that the planned completion time of the test is a time interval, that is, the process time is uncertain. In the case of uncertain process time, the high-end equipment trial production and test processes are coordinated and scheduled, and the part process requirements and the processing order of the parts are comprehensively considered to dynamically and accurately obtain an approximate optimal solution to the problem, which can effectively coordinate and optimize the resources in the two stages of high-end equipment trial production and test, thereby minimizing the input cost of the trial production and test processes and improving the resource utilization efficiency and cooperation efficiency of high-end equipment enterprises.

[0061] The following describes each step in detail, as Figure 2 shown:

[0062] In step S1, set the input parameters of the hybrid genetic algorithm based on neighborhood search according to the part data, factory data, task data during the test process, and technician data during the test process in the high-end equipment trial production process. The specific implementation process is as follows:

[0063] Obtain the part data and factory data in the high-end equipment trial production stage, the test task data and R & D personnel data in the test stage through manual input and other methods, and set the input parameters of the genetic and variable neighborhood search hybrid algorithm based on the obtained data.

[0064] The input parameters of the hybrid genetic algorithm based on neighborhood search include: the high-end equipment part set J = {1, 2, …, n}; the manufacturing process C of the parts = {c1, c2, …, c s}, s ≤ n, each part corresponds to one of the manufacturing processes, and each process category c r (c r ∈C) needs to be in a specific factory Processed in; the set of manufacturing plants M = {1, 2, …, m} in the trial production stage; the quantity N required for each part i ; the processing cycle of the first piece of the workpiece and the subsequent processing time d in the same factory i ; the set of test tasks T = {1, 2, …, k} in the trial production stage; the planned completion time interval e of the test t ; the set of technical personnel P = {1, 2, …, l} in the test stage; the work efficiency w of technical personnel p p , set 0 < w p ≤1.

[0065] In step S2, according to the input parameters and the hybrid genetic algorithm based on neighborhood search, obtain the global optimal solution, and according to the global optimal solution, obtain the set of parts allocated to each factory in the trial production stage, the processing order of parts in the factory, the execution order of subsequent test tasks, and assign technical personnel to each test task. The specific implementation process is as follows:

[0066] S21. Set the execution parameters of the hybrid genetic algorithm based on neighborhood search. Specifically, it includes:

[0067] The maximum number of iterations Gen1 of the hybrid genetic algorithm based on neighborhood search;

[0068] The population size PopSize of the algorithm;

[0069] The crossover probability Pc;

[0070] The maximum number of iterations Gen2 of the local search;

[0071] The initial selection probability SP of each neighborhood structure NS in the neighborhood set {NS1, NS2, …, NS6} K = 1 / 6. K

[0072] S22. Generate the initial population. The individuals in the initial population are represented by two-stage encoding PS-FS and TS-PN. Decode the two-stage encoding to obtain the fitness value of the individual, and obtain and record the current optimal solution. Specifically, it includes:

[0073] Based on the hybrid algorithm of genetic and variable neighborhood search, encode the data of parts and factories in the high-end equipment trial production process and the data of tasks and technical personnel in the test process to obtain an initial population X0 composed of PopSize initial solutions. Each solution in X0 is composed of two sequences PS-FS and TS-PN. Among them, the PS-FS encoding represents the corresponding relationship and sequence of parts and factories in the trial production stage, and the TS-PN represents the assignment order of test tasks and technical personnel in the test stage. Decode each initial solution in the initial population to obtain the fitness value C of the population individuals​max , this value represents the total time span of the trial production and test process; the best individual among them is recorded as the current global optimal solution π best , and set the initial iteration number of the hybrid optimization algorithm gen1 = 1, and set the initial iteration number of the variable neighborhood search gen2 = 1. The specific implementation process is as follows:

[0074] The encoding process is as follows:

[0075] S2201. Based on the characteristics of the two key stages of part trial production and combined test, the solution X is composed of the solution X1 in the trial production stage and the solution X2 in the test stage. The solutions in each stage are represented by two-dimensional integer arrays. The solution X1 contains two parts, the part processing sequence PS and the factory list FS. The length of the sequence or list is The numbers in PS represent the part numbers, and the number of times the numbers appear represents the number of parts. FS represents the factory number, and each column in the array represents the parts with the corresponding number assigned to the corresponding factory for trial production. The solution X2 is composed of the test execution sequence TS and the personnel list PN. The length of the sequence or list is k. TS and PN are respectively composed of test tasks and technician numbers, and each column in the array represents the test task with the corresponding number executed by the corresponding technician. In particular, the order in TS is determined according to the following rules:

[0076] According to the planned completion time interval of each test task, take its expected value Calculate the completion time of each test task to obtain the earliest start time ES of each test k , and then arrange them in ascending order according to the ES k value. If the earliest start times of the test tasks are the same, the part availability time is calculated according to X1, and the test tasks with parts available first are given priority. The encoding method is as Figure 3 shown.

[0077] It should be understood that the number of parts to be processed in the trial production stage is actually calculated according to the requirements of subsequent tests for the types and quantities of parts and the reusable times of the parts themselves. Each test requires different parts for combination. In actual scenarios, high-end equipment manufacturing enterprises will put the reusable parts into the execution process of the next test after completing a certain test in order to avoid resource waste. In addition, the technicians that can be selected in the test stage are mainly to ensure the normal progress of the test, and the actual tasks are often greater than the number of personnel. This requires technicians to wait or immediately enter the next test process after completing a certain test. According to the differences in the work efficiency of each technician, when a certain test task t is assigned to technician p, the actual completion time formula of this test is where represents the expected value of the planned completion time of the test, wp Represents the efficiency of the technical staff, generally 0 < w p ≤ 1.

[0078] The decoding process is as follows:

[0079] S2202. The decoding process is to calculate the total span time of the entire trial production and test process based on the obtained feasible solution according to the actual scenario. It mainly includes the following operations:

[0080] S2202a. Generate the set of parts M to be processed in factory j from the solution X1 j . Let h = 1 and define the variable f h = 0, representing the completion time of the h-th part in the sequence PS.

[0081] S2202b. Derive the set of previous operations of the h-th part within the assigned factory from M j

[0082] S2202c. Calculate the latest completion time of the set of previous operations of the h-th part and determine whether it contains the part type of the h-th part. If so, execute step S2202d; otherwise, execute step S2202e.

[0083] S2202d. Assign h + 1 to h.

[0084] S2202e. Assign h + 1 to h.

[0085] S2202f. Determine whether holds. If it holds, define the variable t = 1 and execute step S2202g; otherwise, return to step S2202b.

[0086] S2202g. Assign the completion time of the set of immediate predecessor tasks T t of the t-th test to the variable F t * , assign the part availability time to the variable C t , and assign the latest completion time of the immediate predecessor test tasks of the corresponding technical staff in PN to the variable P t .

[0087] S2202h. Define the variable S t , S t represents the earliest start time of the t-th test. Let S t = max(F t * , C t , Pt )。

[0088] S2202i. Assign t + 1 to t. Determine whether t > k holds. If so, execute step S2202j; otherwise, return to the previous step S2202h.

[0089] S2202j. Select the maximum value from the earliest start times of each test task obtained in steps S2202g to S2202i and assign it to S max 。

[0090] S2202k. According to the formula C max ≥S max +a t Output C max ,where a t is the actual completion time of the test task corresponding to S max .

[0091] S23. Based on the fitness values of each individual in the population, perform a selection operation using the roulette wheel method. Specifically, it includes:

[0092] S2301. Decode to obtain the total span time of the trial production and test process of each initial solution, and record this value as f(x), that is, let f(x) = C max , calculate the probability that each individual x i in the initial population X0 is inherited to the next generation population

[0093] S2302. Calculate the cumulative probability of each individual in X0

[0094] S2303. Define a variable Q, generate a random number in the interval (0, 1), assign this random number to the variable Q, determine the interval to which the variable Q belongs. If Qx i-1 <Q<Qx i , then select the individual x i in the initial population X0.

[0095] S2304. Repeat step S2303 until PopSize population individuals are selected.

[0096] S24. Based on the crossover probability, perform a crossover operation on the population individuals selected in S23 to obtain a new population with the same scale as the initial population. The specific implementation process is as follows:

[0097] The crossover operation aims to reproduce and spread the excellent genes in the parent generation through the crossover combination of two individuals, thereby generating new excellent individuals and enabling the obtained solutions to evolve in a more optimal direction. To maintain the population size, the crossover operation needs to be performed PopSize / 2 times. Each time, it is judged whether random(1) > Pc holds. If it holds, these two parent individuals are crossed over; otherwise, two new parent individuals are reselected. The specific steps of the crossover operation are described as follows:

[0098] Select two individuals Pa1 and Pa2 from the parent generation, and randomly generate two integers Cut1 and Cut2 as crossover points in the interval where The PS-FS sequence in offspring 1 copies the encoding of the PS-FS between Cut1 and Cut2 in parent Pa2, and the remaining encodings are sequentially copied from the uncopied encodings in parent Pa2; randomly generate two integers Cut1 and Cut2 as crossover points in the interval (1, k), where 1 < Cut1 < Cut2 < k. The sequence PN in offspring 1 copies the encoding of the PN between Cut1 and Cut2 in parent Pa2, and the remaining encodings are directly copied from the remaining positions in parent Pa2. The generation method of offspring 2 is the same.

[0099] S25. Decode the individuals in the current population to obtain the fitness value C of each individual max , and denote the best individual among them as π. Select a neighborhood structure NS based on the roulette wheel method k , and generate a neighborhood solution π' of π according to this neighborhood structure NS k . The specific implementation process is as follows:

[0100] S2501. Decode the individuals in the current population to obtain the fitness value C of each individual max , and denote the best individual among them as π. The specific decoding process refers to S2202 and will not be elaborated here.

[0101] S2502. Select a neighborhood structure NS based on the roulette wheel method k , and generate a neighborhood solution π' of π according to this neighborhood structure NS k . Specifically, it includes:

[0102] Transform the feasible solution π in the selected neighborhood structure NS K so that the initial solution at this stage jumps to another solution in the feasible region, thereby avoiding repeated iteration under the same solution and preventing the algorithm from falling into a local optimum. Considering the complexity of the high-end equipment trial production and test process and the requirements of the constraint relationship, in order to ensure that each transformed solution is a feasible solution, the following 6 neighborhood structures are designed:

[0103] Neighborhood Structure 1: Define variables a and b. Randomly generate two integers within the interval and assign them to a and b, where a < b. Reverse the encoding between positions a and b in sequences PS and FS. Randomly generate two integers within the interval [1, k], assign these two integer values to variables a and b such that a < b, and reverse the encoding between positions a and b in sequence PN.

[0104] Neighborhood Structure 2: Define variables a and b. Randomly generate two integers within the interval and assign them to a and b, where a < b. Swap the encoding to the left of position a and to the right of position b in sequences PS and FS. Randomly generate two integers within the interval [1, k], assign these two integer values to variables a and b such that a < b, and swap the encoding to the left of position a and to the right of position b in sequence PN.

[0105] Neighborhood Structure 3: Define variables a and b. Arbitrarily select two part encodings with the same manufacturing process from PS, assign the encoding positions to variables a and b such that a < b, and swap the encoding at positions a and b in sequence FS. Randomly generate two integers within the interval [1, k], assign them to a and b, where a < b, and swap the encoding at positions a and b in PN.

[0106] Neighborhood Structure 4: Define variables a and b. Arbitrarily select two part encodings with the same manufacturing process from PS, assign the encoding positions to variables a and b such that a < b, and swap the encoding at positions a and b in sequence PS. Randomly generate two integers within the interval [1, k], assign them to a and b, where a < b, and swap the encoding at positions a and b in PN.

[0107] Neighborhood Structure 5: Define variables a, b, and c. Randomly generate three different integer values within the interval and assign them to variables a, b, and c such that a < b < c. Swap the encoding to the left of position a and between positions a and b in sequences PS and FS, and swap the encoding to the right of position c and between positions b and c. Randomly generate three different integer values within the interval [1, k], assign them to a, b, and c, where a < b < c, swap the encoding to the left of position a and between positions a and b in sequence PN, and swap the encoding to the right of position c and between positions b and c.

[0108] Neighborhood Structure 6: Define variables a, b, and c. Randomly generate three different integer values within the interval Randomly generate three different integer values in the range and assign them to variables a, b, and c such that a < b < c. Swap the encodings between positions a and b and between positions b and c in sequences PS and FS, and reverse the encodings to the left of a and to the right of c respectively. Randomly generate three different integer values in the interval [1, k] and assign them to a, b, and c, where a < b < c. Swap the encodings between positions a and b and between positions b and c in sequence PN, and reverse the encodings to the left of a and to the right of c respectively.

[0109] S26. Use neighborhood structure NS K Perform local search on solution π' to obtain a local optimal solution π". Compare the local optimal solution π" with the optimal solution π of the variable neighborhood search algorithm. If π" is better than π, then let π = π", and at the same time record the number of successful times that the local optimal solution is found by this neighborhood structure. Otherwise, increase the number of failed times that the local optimal solution is found by this neighborhood structure, and update the neighborhood structure NS. K the selection probability, and determine whether gen2 ≤ Gen2 holds. If so, assign gen2 + 1 to gen2 and return to step S25. Otherwise, execute step S27. The specific implementation process is as follows:

[0110] The traditional variable neighborhood search process has no requirements for the order or selection weights of neighborhood structures, and equally searches each neighborhood structure. Different neighborhood structures will bring different optimization performances. To select a better neighborhood structure and thus improve the optimization efficiency of local search, selection weights for neighborhood structures are introduced in the variable neighborhood search algorithm and continuously updated during the iteration process, including the following steps:

[0111] S2601. Neighborhood structure NS of the variable neighborhood search algorithm K (K = 1, 2,..., K max ), during the stage of setting algorithm parameters, set the initial selection probability SP of each neighborhood structure K to be the same, where

[0112] S2602. Update the selection probability of each neighborhood according to the adaptive selection mechanism, specifically including:

[0113] During the local search process of the variable neighborhood search algorithm, record the number of times S that each neighborhood structure successfully finds a local optimal solution K and the number of failed times L K , and update the selection probability of each neighborhood according to the following formula: where SUC K represents the local search success rate of neighborhood structure NS K , and the calculation formula is Δ is a sufficiently small positive number to avoid the situation where the selection probability of a certain neighborhood structure is 0.

[0114] S2603. Define a variable Q', generate a random number within the interval (0, 1), assign this random number to the variable Q', and determine the interval to which the variable Q' belongs. If SUC K-1 <Q'<SUC K , then select the neighborhood structure NS K as the type of neighborhood search.

[0115] S27. Compare the optimal solution π of the variable neighborhood search algorithm with the worst individual in the initial population X0. If π is better than the worst individual in the worst initial population, then replace that solution with π, and compare π with the global optimal solution π best . If π is better than π best , then let π best =π.

[0116] S28. Determine whether gen1≤Gen1 holds. If so, assign gen1 + 1 to gen1, use the updated population as the input for the next iteration, and return to step S23. Otherwise, output the global optimal solution π best . Specifically:

[0117] Based on the global optimal solution π output by the algorithm best , analyze the PS-FS sequence and TS-PN sequence in this solution according to the actual scenario of high-end equipment enterprises to obtain the set of parts allocated to each factory during the trial production stage and the processing order of the parts within the factory. On this basis, obtain the execution order of subsequent test tasks, and assign corresponding technical personnel to each test task according to PN.

[0118] The embodiment of the present invention also provides a high-end equipment trial production and test collaborative scheduling system considering process uncertainty, including:

[0119] An input parameter acquisition module, configured to set the input parameters of the hybrid genetic algorithm based on neighborhood search according to the part data, factory data, task data during the test, and technical personnel data during the test in the high-end equipment trial production process. Among them, the input parameters include: the planned completion time interval e of the test t , the manufacturing process C = {c1, c2,..., c s} of each part; the first-piece processing cycle and subsequent processing cycle d i ;

[0120] A solution module, configured to obtain a global optimal solution according to the input parameters and a hybrid genetic algorithm based on neighborhood search, and obtain a set of parts allocated to each factory in the trial production stage, the processing order of the parts in the factory, and the execution order of subsequent test tasks according to the global optimal solution, and assign technicians to each test task.

[0121] It can be understood that the high-end equipment trial production and test collaborative scheduling system considering process uncertainty provided by the embodiments of the present invention corresponds to the above-mentioned high-end equipment trial production and test collaborative scheduling method considering process uncertainty. For the explanations, examples, beneficial effects, etc. of the relevant content, reference can be made to the corresponding content in the high-end equipment trial production and test collaborative scheduling method considering process uncertainty, which will not be elaborated here.

[0122] In summary, compared with the prior art, the following beneficial effects are achieved:

[0123] 1. In the embodiments of the present invention, it is considered that the planned completion time of the test is a time interval, that is, the process time is uncertain. In the case of uncertain process time, the trial production and test processes of high-end equipment are collaboratively scheduled, and the part process requirements and the processing order of the parts are comprehensively considered, so as to dynamically and accurately obtain an approximate optimal solution to the problem, which can effectively collaborate and optimize the resources in the two stages of high-end equipment trial production and test, thereby minimizing the input cost in the trial production and test processes and improving the resource utilization efficiency and collaboration efficiency of high-end equipment enterprises.

[0124] 2. The embodiments of the present invention design a hybrid genetic algorithm based on neighborhood search, which combines the global optimization of the genetic algorithm and the local search ability of the variable neighborhood search algorithm. At the same time, an adaptive selection mechanism based on the search results is introduced into the variable neighborhood search algorithm, and a corresponding local search neighborhood structure is designed, effectively improving the optimization efficiency of the algorithm and the quality of the solution.

[0125] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "including a..." does not exclude the existence of additional identical elements in the process, method, article or device including the element.

[0126] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A collaborative scheduling method for the trial production and testing of high-end equipment considering process uncertainty, characterized in that, Including: S1. Set the input parameters of the hybrid genetic algorithm based on neighborhood search according to the part data, factory data, task data during the test, and technician data during the test in the high-end equipment trial production process. Among them, the input parameters include: the planned completion time interval e of the test t , the manufacturing process C of each part = {c1, c2, …, c s}; the first-piece processing cycle of the part and the subsequent processing cycle d i ; S2. Obtain the global optimal solution according to the input parameters and the hybrid genetic algorithm based on neighborhood search. Obtain the part sets assigned to each factory in the trial production stage, the machining sequence of parts in the factory, and the execution sequence of subsequent test tasks according to the global optimal solution, and assign technicians to each test task. Among them, the obtaining of the global optimal solution according to the input parameters and the hybrid genetic algorithm based on neighborhood search includes: S21. Set the execution parameters of the hybrid genetic algorithm based on neighborhood search. S22. Generate an initial population. The individuals in the initial population are represented by two-stage encoding PS-FS and TS-PN. Decode the two-stage encoding to obtain the fitness value of the individual, and obtain and record the current optimal solution. S23. Perform a selection operation based on the fitness values of the individuals in the population using the roulette wheel method. S24. Perform a crossover operation on the population individuals selected in S23 based on the crossover probability to obtain a new population with the same scale as the initial population. S25. Decode the individuals in the current population to obtain the fitness values C of each individual max , record the best individual among them as π, and select a neighborhood structure NS based on the roulette wheel method k , according to this neighborhood structure NS k generate a neighborhood solution π' of π; Considering the complexity of the high-end equipment trial production and test process and the requirements of the constraint relationship, in order to make the solutions generated by each transformation be feasible solutions, the following six neighborhood structures are designed: Neighborhood Structure 1: Define variables a and b, and randomly generate two integers in the interval Assign them to a and b, where a < b; reverse the encoding between positions a and b in sequences PS and FS; randomly generate two integers in the interval [1, k], and assign these two integer values to variables a and b such that a < b, and reverse the encoding between positions a and b in sequence PN; Neighborhood Structure 2: Define variables a and b, and randomly generate two integers in the interval Assign these two integers to a and b, where a < b; swap the encodings to the left of position a and to the right of position b in sequences PS and FS; randomly generate two integers in the interval [1, k], assign these two integer values to variables a and b such that a < b, and swap the encodings to the left of position a and to the right of position b in sequence PN; Neighborhood structure 3: Define variables a and b. Arbitrarily select two part codes with the same manufacturing process from PS, assign the coding positions to variables a and b so that a < b, and swap the codes at positions a and b in the sequence FS. Randomly generate two integers in the interval [1, k], assign them to a and b, where a < b, and swap the codes at positions a and b in PN. Neighborhood structure 4: Define variables a and b. Arbitrarily select two part codes with the same manufacturing process from PS, assign the coding positions to variables a and b so that a < b, and swap the codes at positions a and b in the sequence PS. Randomly generate two integers in the interval [1, k], assign them to a and b, where a < b, and swap the codes at positions a and b in PN. Neighborhood structure 5: Define variables a, b, and c, and randomly generate three different integer values in the interval to assign them to variables a, b, and c, such that a < b < c. Swap the codes between the left side of position a and between a and b in sequences PS and FS, and swap the codes between the right side of position c and between b and c. Randomly generate three different integer values in the interval [1, k] and assign them to a, b, and c, where a < b < c. Swap the codes between the left side of position a and between a and b in sequence PN, and swap the codes between the right side of position c and between b and c. Neighborhood structure 6: Define variables a, b, and c, and randomly generate three different integer values in the interval to assign them to variables a, b, and c such that a < b < c. Swap the encodings between positions a and b and between positions b and c in sequences PS and FS, and reverse the encodings to the left of a and to the right of c respectively. Randomly generate three different integer values in the interval [1, k] and assign them to a, b, and c, where a < b < c. Swap the encodings between positions a and b and between positions b and c in sequence PN, and reverse the encodings to the left of a and to the right of c respectively. S26. Use the neighborhood structure NS K Perform a local search on the solution π' to obtain the local optimal solution π”. Compare the local optimal solution π” with the optimal solution π of the variable neighborhood search algorithm. If π” is better than π, then let π = π”, and at the same time record the successful number of times that the local optimal solution is found by this neighborhood structure. Otherwise, increase the failed number of times that the local optimal solution is found by this neighborhood structure, and update the neighborhood structure NS K The selection probability, determine whether gen2 ≤ Gen2 holds. If so, assign gen2 + 1 to gen2, and return to step S25. Otherwise, execute step S27; S27. Compare the optimal solution π of the variable neighborhood search algorithm with the worst individual in the initial population X0. If π is better than the worst individual in the worst initial population, then replace the solution with π, and compare π with the global optimal solution π best . If π is better than π best , then let π best = π; S28. Determine whether gen1 ≤ Gen1 holds. If so, assign gen1 + 1 to gen1, use the updated population as the input for the next iteration, and return to step S23. Otherwise, output the global optimal solution π best ; The generating of the initial population, calculating the fitness value of each individual, and obtaining and recording the current optimal solution include: Based on the hybrid algorithm of genetic and variable neighborhood search, the part and factory data in the high-end equipment trial production process and the task and technician data in the test process are encoded to obtain an initial population X0 composed of PopSize initial solutions. Each solution in X0 is composed of two sequences, PS-FS and TS-PN. Among them, the PS-FS encoding represents the corresponding relationship and sequence order between parts and factories in the trial production stage, and TS-PN represents the assignment order of test tasks and technicians in the test stage. Decode each initial solution in the initial population to obtain the fitness value C of the population individuals max , which represents the total time span of the trial production and test processes; record the best individual among them as the current global optimal solution π best , and set the initial iteration number of the hybrid optimization algorithm gen1 = 1, and set the initial iteration number of the variable neighborhood search gen2 = 1.

2. The collaborative scheduling method for high-end equipment trial production and testing considering process uncertainty according to claim 1, characterized in that The hybrid genetic algorithm based on neighborhood search encodes the part and factory data in the high-end equipment trial production process and the task and technician data in the test process, including: Based on the characteristics of the two key stages of part trial production and combined testing, the solution X consists of the solution X1 in the trial production stage and the solution X2 in the testing stage. The solutions in each stage are represented by two-dimensional integer arrays; the solution X1 contains two parts, the part processing sequence PS and the factory list FS, and the length of the sequence or list is The numbers in PS represent the part numbers, and the number of times the numbers appear represents the number of parts; FS represents the factory numbers, and each column in the array represents the parts with the corresponding numbers assigned to the corresponding factories for trial production; the solution X2 consists of the test execution sequence TS and the personnel list PN, and the length of the sequence or list is k; TS and PN are respectively composed of test tasks and technician numbers, and each column in the array represents the test tasks with the corresponding numbers executed by the corresponding technicians.

3. A collaborative scheduling method for high-end equipment trial production and testing considering process uncertainty as described in claim 1, characterized in that The performing of the selection operation based on the fitness values of the individuals in the population using the roulette wheel method includes: S2301. Decode to obtain the total span time of the trial production and test process for each initial solution, and record this value as f(x), that is, let f(x) = C max , calculate the probability that each individual xi in the initial population X0 is inherited to the next generation population S2302. Calculate the cumulative probability of each individual in X0 S2303. Define a variable Q, generate a random number in the interval (0, 1), assign the random number to the variable Q, and determine the interval to which the variable Q belongs. If Qx i-1 <Q<Qx i , then select the individual x in the initial population X0 i ; S2304. Repeat step S2303 until PopSize population individuals are selected.

4. The collaborative scheduling method for high-end equipment trial production and testing considering process uncertainty according to claim 1, characterized in that, The use of the neighborhood structure NS K Perform a local search on the solution π' to obtain the local optimal solution π”. Compare the local optimal solution π” with the optimal solution π of the variable neighborhood search algorithm. If π” is better than π, then let π = π”, and at the same time record the number of successful times that the local optimal solution is found by this neighborhood structure. Otherwise, increase the number of failure times that the local optimal solution is found by this neighborhood structure, and update the neighborhood structure NS K selection probability, including: S2601, Neighborhood Structure NS of Variable Neighborhood Search Algorithm K (K = 1, 2, …, K max ), in the stage of setting algorithm parameters, set the initial selection probability SP of each neighborhood structure K to be the same, where S2602. During the local search process of the variable neighborhood search algorithm, record the number of times S that each neighborhood structure successfully searches for the local optimal solution K and the number of failures L K , and update the selection probability of each neighborhood according to the following formula: where SUC K represents the local search success rate of the neighborhood structure NS K , and the calculation formula is Δ is a sufficiently small positive number; S2603. Define a variable Q', generate a random number in the interval (0, 1), assign the random number to the variable Q', determine the interval to which the variable Q' belongs. If K-1 <Q' < K , then select the neighborhood structure NS K as the type of neighborhood search.

5. A high-end equipment trial production and test collaborative scheduling system considering process uncertainty, characterized in that, Including: An input parameter acquisition module is used to set the input parameters of a hybrid genetic algorithm based on neighborhood search according to part data, factory data, task data during the test process, and technician data during the high-end equipment trial production process. Among them, the input parameters include: the planned completion time interval e of the test t , the manufacturing process C = {c1, c2, …, cs} of each part; the first-piece processing cycle of the part and the subsequent processing cycle d i ; A solution module, configured to obtain the global optimal solution according to the input parameters and the hybrid genetic algorithm based on neighborhood search, obtain the part sets assigned to each factory in the trial production stage, the machining sequence of parts in the factory, and the execution sequence of subsequent test tasks according to the global optimal solution, and assign technicians to each test task. Among them, the obtaining of the global optimal solution according to the input parameters and the hybrid genetic algorithm based on neighborhood search includes: S21. Set the execution parameters of the hybrid genetic algorithm based on neighborhood search. S22. Generate an initial population. The individuals in the initial population are represented by two-stage encoding PS-PN and TS-PN. Decode the two-stage encoding to obtain the fitness value of the individual, and obtain and record the current optimal solution. S23. Perform a selection operation using the roulette wheel method based on the fitness values of each individual in the population; S24. Perform a crossover operation on the population individuals selected in S23 based on the crossover probability to obtain a new population with the same scale as the initial population; S25. Decode the individuals in the current population to obtain the fitness value C of each individual max , record the best individual among them as π, and select a neighborhood structure NS based on the roulette wheel method k , according to this neighborhood structure NS k Generate a neighborhood solution π' of π; Considering the complexity of the trial production and test process of high-end equipment and the requirements of constraint relationships, in order to ensure that the solutions generated by each transformation are feasible solutions, the following six neighborhood structures are designed: Neighborhood Structure 1: Define variables a and b, and randomly generate two integers within the interval and assign them to a and b, where a < b; reverse the encoding between positions a and b in sequences PS and FS; randomly generate two integers within the interval [1, k], assign these two integer values to variables a and b such that a < b, and reverse the encoding between positions a and b in sequence PN; Neighborhood Structure 2: Define variables a and b, and randomly generate two integers in the interval Assign these two integers to a and b, where a < b; swap the encodings to the left of position a and to the right of position b in sequences PS and FS; randomly generate two integers in the interval [1, k], assign these two integer values to variables a and b such that a < b, and swap the encodings to the left of position a and to the right of position b in sequence PN; Neighborhood structure 3: Define variables a and b. Arbitrarily select two part codes with the same manufacturing process from PS, assign the coding positions to variables a and b such that a < b, and swap the codes at positions a and b in the sequence FS; randomly generate two integers in the interval [1, k], assign them to a and b, where a < b, and swap the codes at positions a and b in PN; Neighborhood structure 4: Define variables a and b. Arbitrarily select two part codes with the same manufacturing process from PS, assign the coding positions to variables a and b such that a < b, and swap the codes at positions a and b in the sequence PS; randomly generate two integers in the interval [1, k], assign them to a and b, where a < b, and swap the codes at positions a and b in PN; Neighborhood structure 5: Define variables a, b, and c, and randomly generate three different integer values in the interval to assign them to variables a, b, and c such that a < b < c. Swap the encodings between the left of position a and between a and b in sequences PS and FS, and swap the encodings between the right of position c and between b and c. Randomly generate three different integer values in the interval [1, k] and assign them to a, b, and c, where a < b < c. Swap the encodings between the left of position a and between a and b in sequence PN, and swap the encodings between the right of position c and between b and c. Neighborhood structure 6: Define variables a, b, and c, and randomly generate three different integer values in the interval , assign them to variables a, b, and c such that a < b < c, swap the encodings between positions a and b and between positions b and c in sequences PS and FS, and reverse the encodings to the left of a and to the right of c respectively; randomly generate three different integer values in the interval [1, k], assign them to a, b, and c, where a < b < c, swap the encodings between positions a and b and between positions b and c in sequence PN, and reverse the encodings to the left of a and to the right of c respectively; S26. Use the neighborhood structure NS K to perform a local search on the solution π' to obtain the local optimal solution π", compare the local optimal solution π" with the variable neighborhood search algorithm optimal solution π. If π" is better than π, then let π = π", and at the same time record the successful number of times the local optimal solution is found by this neighborhood structure. Otherwise, increase the failed number of times the local optimal solution is found by this neighborhood structure, update the selection probability of the neighborhood structure NS K , and determine whether gen2 ≤ Gen2 holds. If so, assign gen2 + 1 to gen2 and return to step S25. Otherwise, execute step S27; S27. Compare the optimal solution π of the variable neighborhood search algorithm with the worst individual in the initial population X0. If π is better than the worst individual in the worst initial population, replace the solution with π, and compare π with the global optimal solution π best . If π is better than π best , then let π best = π; S28. Determine whether gen1 ≤ Gen1 holds. If so, assign gen1 + 1 to gen1, use the updated population as the input for the next iteration, and return to step S23. Otherwise, output the global optimal solution π best ; The generation of the initial population, calculation of the fitness value of each individual, and obtaining and recording the current optimal solution include: Based on the hybrid algorithm of genetic and variable neighborhood search, encode the part and factory data in the high-end equipment trial production process and the task and technician data in the test process to obtain an initial population X0 composed of PopSize initial solutions. Each solution in X0 is composed of two sequences, PS-FS and TS-PN. Among them, the PS-FS encoding represents the corresponding relationship and sequence of parts and factories in the trial production stage, and TS-PN represents the assignment sequence of test tasks and technicians in the test stage; decode each initial solution in the initial population to obtain the fitness value C of the population individuals max , which represents the total time span of the trial production and test processes; record the best individual among them as the current global optimal solution π best , and set the initial iteration number of the hybrid optimization algorithm gen1 = 1, and set the initial iteration number of the variable neighborhood search gen2 = 1.

Images