Multi-strategy flexible job shop energy-saving scheduling method

Abstract

The invention relates to the technical field of flexible job shop scheduling control, in particular to a multi-strategy flexible job shop energy-saving scheduling method, which comprehensively considers quality requirements, manufacturing span and energy consumption factors of job processing, constructs a multi-strategy energy-saving scheduling problem model, defines a job set, a process set, a machine set and a processing speed set, and provides a multi-strategy energy-saving scheduling scheme. And the maximum manufacturing span and the total energy consumption are taken as optimization objectives. According to the method, two optimization objectives of the maximum manufacturing span and the total energy consumption are comprehensively considered by constructing a multi-strategy energy-saving scheduling problem model, the solving efficiency and precision are effectively improved by simulating distribution and energy consumption setting of double-layer coding definition operation, the heuristic energy-saving strategy is further applied, and the method is suitable for large-scale popularization and application. Comprising a machine on / off strategy, a processing speed adjusting strategy and a processing speed combination changing strategy, the energy consumption of the job shop is further reduced, and in conclusion, the method has important application value in the aspects of improving the scheduling efficiency of the job shop and saving energy and reducing emission.

Description

Technical Field

[0001] The invention relates to the technical field of flexible job shop scheduling control, and in particular to a multi-strategy flexible job shop energy-saving scheduling method. Background Art

[0002] The flexible job shop scheduling problem is a generalization of the classic job shop scheduling problem. It is a type of NP-hard problem. In the flexible job shop scheduling problem, a process is allowed to be processed by multiple machines. Therefore, it is necessary to assign appropriate machines to each process of each job and determine the processing order of the processes on each machine to achieve the shortest completion time of the job and load balancing of each machine under the premise of meeting various constraints.

[0003] For example, announcement number CN105929690B discloses a flexible workshop robust scheduling method based on a decomposed multi-objective evolutionary algorithm, which includes: (1) reading input information such as the job and machine attributes of the flexible workshop; defining the optimization goal and setting constraints; (2) initializing the algorithm parameters; (3) determining the neighborhood of each subproblem, generating an initial parent population, and determining all Pareto non-dominated solutions from the initial population to form an external memory; (4) generating a child population. Perform mating selection, use an adaptive mutation operator and a repair-based crossover operator to reproduce the offspring individuals, and update the external memory; (5) use the generated child population to update the current optimal individuals of each subproblem to form a new parent population; (6) if the number of individual target evaluations reaches the maximum, output the external memory, that is, a set of Pareto non-dominated flexible workshop scheduling solutions; if not reached, jump to (4). The present invention quickly and efficiently implements scheduling tasks in flexible workshops;

[0004] The method for solving the workshop scheduling problem based on the NSGA-III algorithm provided by the announcement number CN116360355A first constructs a multi-objective flexible workshop scheduling problem model: including the description of the multi-objective flexible workshop scheduling problem and the construction of the model; generates a set of evenly distributed reference points; uses an equal-length three-segment encoding method for encoding, performs crossover and mutation operations based on adaptive operators on the individuals in the parent population Pt, performs approximate dominance sorting based on the reference points on the temporary population Rt, and determines whether the termination condition is met. Finally, a weighted method is used to select a solution from the optimal solution set as the optimal compromise solution and output it. The present invention solves the problem that different rotation speeds have an impact on the machining effect of the machine in the prior art;

[0005] The announcement number CN114819353A provides a machine multi-speed workshop energy-saving scheduling method based on the improved SPEA2 algorithm. First, a flexible workshop energy-saving scheduling problem model is constructed: including the description of the flexible workshop energy-saving scheduling problem and model construction; a three-segment coding based on natural numbers is adopted, and the three-segment coding is respectively a process code, an equipment code and a speed code; fitness assignment is performed, a non-dominated solution set is constructed through fast non-dominated sorting, and then an environment selection is performed, and the optimal compromise solution is selected using a weighted method; termination judgment, if t is greater than or equal to t max , output the optimal compromise solution X * ; On the contrary, if t is less than t max , perform genetic selection; divide populations A, B, C, and D by special crowding distance, average fitness value and other indicators; use simulated annealing selection mechanism to select individuals in populations A, B, C, and D for reverse learning, spiral position update, Gaussian mutation or random perturbation, and multi-bit transformation operations, respectively. The present invention solves the problem of excessive energy consumption in workshop production processes in the prior art.

[0006] However, when the above technology is actually used:

[0007] First, although CN105929690B uses MOEA / D, it does not consider the variable speed principle of machine tools, which is easy to implement for flexible workshops, and does not formulate relevant energy-saving strategies to improve energy utilization;

[0008] CN116360355A and CN114819353A take into account the situation of multiple speeds, but do not formulate energy-saving strategies, nor do they consider the processing quality. In the actual processing process, if the processing speed is adjusted arbitrarily, the processing quality will vary. Summary of the invention

[0009] The purpose of the present invention is to provide a multi-strategy flexible job shop energy-saving scheduling method to solve the problem of not considering the variable speed principle of processing and not formulating relevant energy-saving strategies to improve energy utilization.

[0010] To achieve the above object, the present invention provides the following technical solution: a multi-strategy flexible job shop energy-saving scheduling method, comprising the following steps:

[0011] S1. Comprehensively consider the quality requirements, manufacturing span and energy consumption factors of job processing, build a multi-strategy energy-saving scheduling problem model, define the job set, process set, machine set and processing speed set, and take the maximum manufacturing span and total energy consumption as the optimization goals, set the problem constraints, and the multi-strategy energy-saving scheduling problem model is:

[0012]

[0013] S2. Using a decomposition-based multi-objective evolutionary algorithm to solve the multi-strategy energy-saving scheduling problem model in step S1, specifically including:

[0014] S21, using simulated double-layer coding to define the allocation of jobs and the energy consumption setting of the jobs;

[0015] S22, using three methods of random initialization, local load minimum initialization and global load minimum initialization to initialize the population non-dominated resettable;

[0016] S23, by constructing neighborhoods, adaptive crossover and mutation operations, the population is continuously updated iteratively to obtain the optimal Pareto frontier solution;

[0017] S3. In the solution process of step S2, a heuristic energy-saving strategy is applied to reduce the energy consumption of the job shop, wherein the heuristic energy-saving strategy includes a machine on / off strategy, a processing speed adjustment strategy, a processing speed combination change strategy, and a low-speed strategy under no-load state.

[0018] Preferably, the multi-strategy energy-saving scheduling problem model in step S1 is defined as follows:

[0019] There exists a job J = {J1, J2, ..., J n}, where each J p There is a set of independent processes O = {O p1 ,O p2 ,…,O pj},O i,j represents the jth process of the ith job, in the machine set M = {M1, M2, ..., M k} selects a machine to process, any machine from V = {V1, V2, ..., V s Select speed processing O in i,j , usually V→V i,j , that is, the processing speed on the machine is mapped to O i,j Processing speed (O i,j ∈O), considering the control of processing speed, according to O i,j Q i,j Get V i,j (Q i,j ∈Q,V i,j ∈V), in machine M k Upper processing O i,j Used time T i,j,k The energy consumption caused is calculated as follows: EO i,j,k *T i,j,k .

[0020] Preferably, the constraint condition in step S1 is:

[0021] The initial available time of each machine starts from 0;

[0022] Each machine processes only one process at a time;

[0023] Each machine's on / off is subject to a safety time limit, i.e. the interval between shutting down and starting up. The machine's processing speed is determined by the processing speed of the workpiece;

[0024] The processing speed of the workpiece is limited by the processing quality requirements of the workpiece;

[0025] The sequence of different processes in the same operation is fixed.

[0026] Preferably, the simulated double-layer coding in step S21 includes a job allocation code and an energy consumption setting code. The job allocation code is used to determine which machine each process is assigned to for processing, and the energy consumption setting code is used to determine the processing speed of each process on the selected machine, thereby further determining the energy consumption.

[0027] Preferably, the adaptive crossover and mutation operation steps in step S23 are based on simulated two-layer vector coding, and adopt a job-based crossover operator and an elite retention strategy.

[0028] Preferably, the machine on / off strategy in step S3 is to cut off the power supply of the machine according to a set time threshold to save energy when the machine is idle for too long.

[0029] Preferably, the processing speed strategy is adjusted in step S3, including appropriately slowing down the processing speed of a certain process while satisfying load balancing, and considering the processing speed combination of the previous and next processes to form the best energy scheduling.

[0030] Compared with the prior art, the present invention has the following beneficial effects:

[0031] 1. The present invention achieves comprehensive consideration of the two optimization objectives of maximum manufacturing span and total energy consumption by constructing a multi-strategy energy-saving scheduling problem model. By imitating double-layer coding to define the allocation and energy consumption settings of jobs, the solution efficiency and accuracy are effectively improved. The present invention also applies heuristic energy-saving strategies, including machine on / off strategies, processing speed adjustment strategies, and processing speed combination changes. The strategy further reduces the energy consumption of the job shop. In summary, this method has important application value in improving the scheduling efficiency of the job shop and energy conservation and emission reduction. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is an overall flow chart of the multi-strategy flexible job shop energy-saving scheduling method of the present invention;

[0033] Figure 2 It is a schematic diagram of the non-dominated resettable initialization of the population by the multi-strategy flexible job shop energy-saving scheduling method of the present invention;

[0034] Figure 3 It is a schematic diagram of a crossover operator based on location in the multi-strategy flexible job shop energy-saving scheduling method of the present invention;

[0035] Figure 4 Qs in the multi-strategy flexible job shop energy-saving scheduling method of the present invention min With SS max Schematic diagram of the corresponding selection relationship;

[0036] Figure 5 Schematic diagram of machine on / off strategy of the multi-strategy flexible job shop energy-saving scheduling method of the present invention;

[0037] Figure 6 A schematic diagram of a strategy for accelerating processing speed in adjusting the processing speed strategy in the multi-strategy flexible job shop energy-saving scheduling method of the present invention;

[0038] Figure 7 A schematic diagram of a strategy for slowing down the processing speed in a strategy for adjusting the processing speed in a multi-strategy flexible job shop energy-saving scheduling method of the present invention;

[0039] Figure 8 The multi-strategy flexible job shop energy-saving scheduling method of the present invention changes the depth-first search algorithm in the processing speed combination strategy to find the optimal speed combination;

[0040] Fig. 9 A schematic diagram of a combination strategy for changing processing speeds in a multi-strategy flexible job shop energy-saving scheduling method of the present invention;

[0041] Fig.10 It is a schematic diagram of the low-speed strategy in the no-load state of the multi-strategy flexible job shop energy-saving scheduling method of the present invention;

[0042] Fig.11 A comparison chart of total energy consumption between no strategy and machine on / off strategy in the embodiment;

[0043] Fig.12 This is a comparison chart of total energy consumption between the machine on / off strategy and the processing speed adjustment strategy in this embodiment;

[0044] Fig.13 This is a TEC comparison diagram of the multi-strategy 2-1 and the combination strategy of changing the processing speed in this embodiment;

[0045] Fig.14 This is a comparison chart of TEC without strategy and executing multiple strategies in this embodiment. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] See also Figure 1-14 The present invention provides a technical solution: a multi-strategy flexible job shop energy-saving scheduling method, comprising the following steps:

[0048] S1. Comprehensively consider the quality requirements, manufacturing span and energy consumption factors of job processing, build a multi-strategy energy-saving scheduling problem model, define the job set, process set, machine set and processing speed set, and take the maximum manufacturing span (MS) and total energy consumption (TEC) as the optimization goals, set the problem constraints, and the multi-strategy energy-saving scheduling problem model is:

[0049]

[0050] The multi-strategy energy-saving scheduling problem model is defined as follows:

[0051] There exists a job J = {J1, J2, ..., J n}, where each J p There is a set of independent processes O = {O p1 ,O p2 ,…,O pj},O i,j The jth process of the i-th job can be found in the machine set M = {M1, M2, ..., M k} selects a machine to process. Any machine can be selected from V = {V1, V2, ..., V s Select speed processing O in i,j , and usually V→V i,j , that is, the processing speed on the machine can be mapped to O i,j Processing speed (O i,j ∈O). Considering the control of processing speed, according to O i,j Q i,j You can get V i,j (Q i,j ∈Q,V i,j ∈V). It can be used on machine M k Upper processing O i,j Used time T i,j,k The resulting energy consumption can be calculated as follows: EO i,j,k *T i,j,k ;

[0052] And the constraints set are:

[0053] The initial available time of each machine starts from 0;

[0054] Each machine processes only one process at a time;

[0055] Each machine's on / off is subject to a safety time limit, i.e. the interval between shutting down and starting up. The machine's processing speed is determined by the processing speed of the workpiece;

[0056] The processing speed of the workpiece is limited by the processing quality requirements of the workpiece;

[0057] The sequence of different processes in the same operation is fixed;

[0058] in:

[0059] n: number of jobs;

[0060] i: current job number, i=1,2,…,n..;

[0061] j: the number of processes in each job, j = 1, 2, ..., n i ..;

[0062] k: number of machines, k = 1, 2, ..., m..;

[0063] q: processing quality code of each process, q = 1, 2, ..., n..;

[0064] s: processing speed code of the machine where the process is located, s = 0.5, 0.6, ..., n..;

[0065] u: idle time period number on machine k, u = 1, 2, ..., n..;

[0066] p: the processing sequence number on machine k, p = 1, 2, ..., n..;

[0067] L k :machine machine k load;

[0068] J n : Code for job n;

[0069] M k : Code of machine k;

[0070] O i,j : The jth process of the i-th job;

[0071] V i,j :Process O i,j The set of available processing speeds;

[0072] Q: A collection of processing quality;

[0073] Qi,j :Process O i,j Processing quality requirements;

[0074] MS i,j :Process O i,j Manufacturing span;

[0075] T i,j,k :Process O i， Processing time at machine k and default processing speed;

[0076] TI k,u : The uth idle time of machine k;

[0077] TM k,p : The pth processing time of machine k;

[0078] TMS k : Safety time interval on machine k;

[0079] MSM k : Processing rate on machine k;

[0080] MIP k : Unit idle power on machine k;

[0081] MSP k : Unit starting power on machine k;

[0082] MWP k : Unit preheating power on machine k;

[0083] MSBP k : Unit standby power on machine k;

[0084] MSTP k : The unit setting power on machine k;

[0085] EO i,j,k :Process O i,j The default unit energy consumption on machine k;

[0086] EO i,j,k,s :Process O i,j The actual unit energy consumption on machine k running at speed s;

[0087] V i,j,k :O i,j Running speed on Mk

[0088] V i,j,k ∈(V i,j = {V1, V2, …, V s})

[0089]

[0090] X i,j,k : is defined as a binary set as follows:

[0091]

[0092] BO i,j : is defined as a binary set as follows:

[0093]

[0094] BI u,k : is defined as a binary set as follows:

[0095]

[0096] In the above formula, X i,j,k Determine the process to be processed on the machine, BO i,j Check the current O i,j Whether the acceleration and deceleration operation has been completed. u,k Check whether the time slot has been changed, and finally build the multi-strategy energy-saving scheduling problem model considering the processing quality mentioned above.

[0097] S2. Using a decomposition-based multi-objective evolutionary algorithm (MOEA / D) to solve the multi-strategy energy-saving scheduling problem model in step S1, specifically including:

[0098] S21. Use pseudo-double-layer coding to define job allocation coding and energy consumption setting coding. The job allocation coding is used to determine which machine each process is assigned to for processing. The energy consumption setting coding is used to determine the processing speed of each process on the selected machine, thereby further determining the energy consumption. In the coding design of the traditional variable-speed flexible job shop scheduling problem, two problems need to be solved: process sorting and equipment selection. Because in addition to the problems of process sorting and processing equipment selection, process quality is also required. Among them, the process quality determines the processing speed range, thereby affecting the processing energy consumption of the process. Therefore, the present invention designs a pseudo-double-layer vector coding composed of process allocation, process energy consumption setting, process quality setting and processing equipment selection, which are respectively: a tuple (OQ) of process sorting and process quality combination, and a tuple (ME) of processing equipment and process basic processing energy consumption selection. The OQ vector indicates the configuration of all operations in order through a tuple array, and each tuple contains two elements. The first element and the second element suggest process selection (OS) and quality selection (QS), respectively. Taking job 1 in Table 1 as an example, {〈1,1〉, 〈1,7〉, 〈1,1〉}, the assigned processes represent the configurations of O {O1,1,O1,2,O1,3} respectively. The specific values ​​of the selected QS are obtained from Table 2.

[0099] The ME used in this embodiment is expressed as 〈MS, OE〉, where OE represents the basic processing energy consumption of the process when the processing speed is 1. The length is the same as the length of the OQ vector composed of 〈OS, QS〉. For example, to obtain some elements of ME, the machine sub-element OE allocated to job 1 is obtained from Table 2, and the first, first and third available machines are recommended for each process of job 1. The bold part of Table 1 is an example used for the heuristic multi-strategy.

[0100] Table 1: Example artifacts and assignable machines

[0101]

[0102]

[0103] Table 2: Quality requirement selection, processing speed range and corresponding unit energy consumption range for example workpieces

[0104]

[0105]

[0106] S22, using three methods of random initialization, local load minimum initialization and global load minimum initialization to initialize the population non-dominated resettable, specifically:

[0107] There are three ways to initialize the population, such as Figure 2 As shown in the figure, in order to ensure the diversity and superiority of the initial population, three initialization methods are designed, and the selection principles are introduced as follows:

[0108] Random selection principle: Random selection is conducive to the distribution range of the initial population and reduces the possibility of falling into the local optimum. n Randomly select, the corresponding O i,j Randomly select a processable machine M k .

[0109] Principle of minimum local load: This principle is conducive to the initial population to produce high-quality solutions. n Processing selection, O i,j Select the machine M that can be processed k It should be the one with the smallest current machine load.

[0110] Global minimum load principle: Similar to the previous principle. For all job processing selections, O i,j Select the machine M that can be processed k It should be the one with the smallest current machine load;

[0111] S23, by constructing neighborhoods, adaptive crossover and mutation operations, the population is continuously updated iteratively to obtain the optimal Pareto front solution, wherein the adaptive crossover and mutation operation steps are based on the simulated two-layer vector encoding, using the job-based crossover operator and elite retention strategy, and the steps of constructing the neighborhood are specifically as follows:

[0112] Generate a uniformly distributed weight vector λ = {λ1,λ2,…,λ i}, calculate the Euclidean distance between the weight vectors λ to construct the neighborhood {λ T1 ,λ T2 ,…,λ Ti}, the Euclidean calculation is as shown in formula (3).

[0113]

[0114] The specific steps of adaptive crossover and mutation operation are as follows:

[0115] Adaptive mutation and crossover operators produce corresponding probability changes as the population evolves. In the later stages of population evolution, the crossover and mutation probabilities gradually decrease, thereby accelerating the convergence of the population. The crossover rate should satisfy the following equation (4), and the mutation rate is similar.

[0116]

[0117] Among them, P cmax represents the upper limit of the crossover rate, P cmin Represents the lower limit of the crossover rate, L Gen Represents the chromosome length.

[0118] In this embodiment, individuals are randomly selected for adaptive crossover and mutation. The crossover adopts a job-based crossover operator. The parents come from the elite reserved set. According to the simulated two-layer vector encoding, for the OQ vector, a job-based crossover operator (PBX) is adopted, such as Figure 3 As shown. Randomly select two jobs J1 and J2 (randomly select from the initial population) J1 and J2 from the elite reserved set, copy the process in p1 that belongs to the job set J1 to c1, and copy the process in p2 that belongs to the job set J2 to c2, and keep their positions; other positions are postponed;

[0119] The minimum requirement for the second element QS of the tuple can meet the acceleration requirement, such as Figure 4 As shown. min The larger the value, the lower the required processing quality and the faster the selectable processing speed (SS). The default slowest processing speed is 1, which represents the original speed designed by the operator to meet the processing requirements;

[0120] Decompose the multi-objective optimization problem of each weight vector into a single-objective optimization problem. Use the Chebyshev method in formula (5) to find the maximum distance of each dimension and optimize the maximum distance during the optimization process. That is, decompose MS and TEC into Max MS or Max TEC .

[0121]

[0122] The parent population P i and the offspring population C i Merge into R i , the population size is N m ={N,N+1,…,2N}. i Perform fast non-dominated sorting to obtain non-dominated solution N p , if N p > N, then according to the sorting level, the ones with higher rankings are placed in the elite reserved set P first i , until N is satisfied Pi = N. At the same time, save the optimal point z*;

[0123] If the maximum number of iterations Max is met p , then the iteration cycle ends and the optimal Pareto front solution is obtained, otherwise Gen i =Gen i +1, continue calculating non-dominated solutions.

[0124] S3. In the solution process of step S2, a heuristic energy-saving strategy is applied to reduce the energy consumption of the job shop, wherein the heuristic energy-saving strategy includes a machine on / off strategy, a processing speed adjustment strategy, a processing speed combination change strategy, and a low-speed strategy under no-load state.

[0125] The machine on / off strategy is to cut off the power supply of the machine according to the set time threshold when the machine is idle for too long to save energy. Specifically:

[0126] Cutting off the power supply of the machine when the idle time is too long can save energy. Mainstream research only considers the energy consumed by turning on and off the machine and the energy consumed by standby to obtain the energy threshold. However, frequent switching on and off will degrade the state of the machine. How to consider the time threshold for cutting off the power supply should be considered. Formula (6) can define the shutdown criteria of a machine: if it is satisfied, shut down the machine k according to the strategy; otherwise, keep it idle.

[0127]

[0128] In this embodiment, Table 3 shows the safety time and state power of the machine. Figure 5 In machine 3, according to the information in Table 3, 4,2 With O1,3 The time interval can be used to compare the energy consumption generated by the power on and off state with the energy consumption generated by the standby state. It can be seen that shutting down the machine can consume less energy. Then, the safety time conditions can be compared. If they are met, the machine can be shut down.

[0129] Table 3: Machine related information

[0130]

[0131] The strategy for adjusting the processing speed includes appropriately slowing down the processing speed of a certain process while satisfying load balancing, and considering the combination of processing speeds of the previous and next processes to form the best energy scheduling, specifically:

[0132] Under the premise of satisfying the processing quality, the processing speed of a certain process can be appropriately accelerated. Considering that the change of processing speed will produce different energy consumption, the process energy consumption-processing speed model is required to satisfy formula (7).

[0133]

[0134] In this embodiment, Table 2 shows the maximum values ​​of the processing quality of the process and the corresponding MSM, wherein the maximum value is determined by the processing quality, and different quality requirements determine different MSMs.

[0135] According to O 2,2 MSM max It can be seen that the processing time can be shortened to 1.8 (rounded to one decimal place). The results after optimization are as follows Figure 6 As shown. Formula (6) can be used to determine whether the time threshold can be met. Formulas (7) and (8) show that when the processing speed is increased, the energy threshold needs to be recalculated. After calculation, it is known that the energy threshold is met, so the processing speed can be increased to increase the idle time to meet the machine on / off strategy, thereby realizing shutdown to save energy. Formula (8) is shown as follows:

[0136] MSP k +MWP k +MSBP k +MSTP k +OP i,j,k,s +OP i',j',k,s <MIP k *TMS k +OP i,j,k +OP i',j',k (8).

[0137] Under the condition of satisfying load balancing, the processing speed of a certain process can be appropriately slowed down.

[0138] When it is possible to reduce speed and energy consumption to increase the processing time of the machine, the choice of acceleration needs to be constrained. Otherwise, all processing time can be reduced by reducing processing energy consumption, and the idle time can be compressed to a minimum to reduce idle energy consumption, while also avoiding power on and off energy consumption. However, the load of the machine will increase infinitely, so this strategy is considered based on load balancing. First, determine whether the current machine meets the acceleration conditions according to formula (9):

[0139]

[0140] If J is a positive value, it means that the current load of machine k does not meet the load balancing condition. If the condition is met, the processing time of the process can be calculated by substituting it into formula (8). It should be noted that the result of formula (10) needs to be updated every time a change occurs.

[0141]

[0142] Under the premise of satisfying the scheduling strategy, Figure 7 Is there a process that can be slowed down? 1,1 Under the above conditions, the processing speed can be slowed down to extend the processing time and set the idle energy consumption between processes to zero. However, because the idle time on machine 1 does not satisfy equation (9), the processing speed cannot be slowed down and the original idle state is temporarily maintained.

[0143] The specific strategy for changing the processing speed combination is:

[0144] The best energy scheduling can be achieved by using appropriate acceleration combinations for two non-adjacent processes. When both the front and rear processes in the idle time window can be accelerated, accelerating both processes simultaneously can optimize the idle energy consumption to the power-on / off energy consumption. However, if the front and rear processes both choose the fastest processing speed, energy waste will also occur, because the fastest processing speed means the maximum processing energy consumption. Therefore, it is necessary to consider the optimal processing speed for the front and rear processes, rather than the fastest processing speed.

[0145] In this embodiment, considering the scheduling environment, the processing speed adjustment strategy is used to slow down the processing speed. Then, under the premise of satisfying the machine on / off strategy after changing the processing speed, TI needs to be considered. k,u The choice of increasing the processing speed at both ends. According to Table 2, TI 1,1 Take MSM for the previous and next processes max The time threshold and energy consumption threshold of equations (6) and (8) can be satisfied respectively. Considering that there may be a large number of selectable processing speeds and the depth is fixed at 3, the idea of ​​the Deep-First Search (DFS) algorithm is adopted to find the optimal processing speed combination using less computer memory, such as Figure 8As shown. Under the premise of meeting the time threshold, the fixed energy consumption of the switch is used as the initial node, and the first layer is O 1,2 Energy consumption after changing the processing speed, the second layer is O 5,4 The energy consumption after changing the processing speed is to find the processing speed combination that can meet the energy threshold from the initial node. Here we can make the following simple assumptions:

[0146] 1.O 1,2 The processing speed and energy consumption model is a linear model that passes through (1,8) and (1.6,10) under the premise of satisfying formula (4);

[0147] 2.O 5,4 The processing speed and energy consumption model is a linear model that passes through (1,6) and (2.0,8) under the premise of satisfying formula (4);

[0148] 3. The step size of the processing speed step is 0.1;

[0149] It can be seen that the best processing speed combination results are as follows Fig. 9 shown.

[0150] The specific low-speed strategy under no-load state is:

[0151] There are three situations when the machine is idle: the process from the machine starting to the start of active processing, the process from the completion of active processing to the machine shutting down, and the process of being idle between the active processing of one process and the next process. When the machine is in an idle state, it should be kept at a low speed to reduce energy loss.

[0152] In this embodiment, considering the standard machine running speed, a fixed minimum speed in no-load state should be set for the machine. Therefore, when the machine is in no-load state, the specified minimum speed should be maintained as much as possible. For example Fig.10 In the process, on machine 4, when the machine is turned on and ready to process O 2,2 When the machine is running, it should be kept at the lowest speed, which needs to meet the state specified by the machine on / off strategy. When active processing is required, the speed should be increased to the speed required for active processing. 2,2 After processing is completed, it needs to remain idle and wait for the next process O 5,2 Start processing. The idle state belongs to the no-load state of this strategy, so it is necessary to reduce the speed and keep it running at the lowest speed. Start processing O 5,2 When the machine is turned on, the speed is increased to the speed required for active processing. When the machine needs to be shut down after the final processing, the speed is reduced to the minimum speed and then shut down.

[0153] Comparative Example 1: Algorithm Index Comparison

[0154] Compare other classic multi-objective algorithms with MOEA / D, which considers the multi-strategy of processing quality. Because there is no benchmark data set containing data such as processing speed and processing quality, the relevant international benchmark data set is used for reference, and the relevant parameter vectors are taken for verification. When verifying the superiority of MOGWO, NSGAⅡ and MOEA / D, because it is necessary to verify the mean and standard deviation multiple times, the relevant vectors of the first verified algorithm will be saved to ensure that the other verified algorithms use the same vectors. For example, if MOGWO is run first and the standard energy consumption of O2,3 is 10, the vector will be saved in the configuration file, and the same vector will be used for subsequent NSGAⅡ and MOEA / D algorithm verification. The machine vector will also be saved and used in the same way. Among them, the population and crossover and mutation probabilities of each algorithm are the same.

[0155] Hypervolume (HV) indicator: The volume of the area in the target space enclosed by the non-dominated solution set obtained by the algorithm and the reference point. The larger the HV value, the better the overall performance of the algorithm. HV can evaluate convergence and diversity at the same time without knowing the advantages of the true PF or reference set. HV is expressed as:

[0156]

[0157] Among them, δ represents the Lebesgue measure, L represents the length of the data set, which is used to measure the volume. |S| represents the number of non-dominated solution sets, v i represents the hypervolume composed of the reference point and the ith solution in the solution set. The results are shown in Table 4.

[0158] The generation distance (GD) index measures the distance from the optimal solution set obtained by the multi-objective optimization algorithm to the true PF. This distance indicates the degree of deviation from the true boundary. The larger the GD value, the farther it deviates from the true optimal boundary, the worse the convergence performance, and the smaller the GD value, the better the convergence performance of the algorithm. GD is expressed as formula (13).

[0159]

[0160] Among them, M represents the set of PF vectors to be verified obtained by the algorithm {m1,m2,m3,...,m M}, d i Represents m i The euclidean distance to the nearest point in the true PF set. Because the true PF is usually not accurately obtained, the true PF of this study is taken from the comprehensive optimal PF of all algorithms. The GD comparison results are recorded in Table 5.

[0161] Table 4: HV index test results

[0162]

[0163]

[0164]

[0165]

[0166] Table 5: GD index test results

[0167]

[0168]

[0169]

[0170] Comparative Example 2: Energy Saving Comparison

[0171] The machine on / off strategy adopts the idle on / off strategy. When the idle time of the machine is longer than the safe shutdown time of the current machine, and the execution of the machine on / off strategy can reduce the idle energy consumption, the machine on / off strategy can be executed. Fig.11 It can be seen that the total energy consumption of the same data set using the machine on / off strategy is less than the total energy consumption without the strategy. The idle energy consumption under the machine on / off strategy is much less than the idle energy consumption without the strategy, and the optimization effect is obvious for large-scale benchmark data sets. Different processes will produce different energy consumption on different machines, so the change in the processing energy consumption is relatively small.

[0172] For variable speed workshops, the speed of the process can be changed under the premise of ensuring the processing quality. Accelerated processing has a negative impact on the processing quality, so only some processes in the benchmark test set can be accelerated, and the quality requirements of these processes are lower, that is, strategy 2-1. Slowing down processing can improve the processing quality, so all processes can be slowed down, but the load balancing principle of the machine must be met, that is, strategy 2-2. Using strategy 2-1 and strategy 2-2 and the machine on / off strategy independently, the bar chart results are as follows Fig.12 The results show that executing strategy 2-1 and strategy 2-2 separately can produce lower energy consumption than executing only the machine on / off strategy. It is not certain which of the two strategies can achieve the best energy-saving scheduling solution in different problems in the dataset.

[0173] In order to further optimize the optimal energy-saving scheduling brought by speed change, a combination strategy for changing processing speed was developed, including a combination of deceleration and acceleration, and a combination of different accelerations. Strategy 2-1 is compared with the combination strategy for changing processing speed, such as Fig.13As shown in the figure, different strategies can be implemented for different job situations. If the job only allows slowing down the processing speed, it is recommended to implement the strategy of adjusting the processing speed. If the job allows speeding up, it is not recommended to implement the strategy of adjusting the processing speed alone compared with the strategy of changing the processing speed combination, because the solution of changing the processing speed combination strategy is more energy-efficient.

[0174] Considering that the machine still has a lot of non-processing time in the idle state, the low-speed strategy in the idle state can minimize the energy waste of non-processing time. It is easy to understand that this strategy, as a supplement to other strategies, can definitely reduce energy consumption. Therefore, the energy consumption of the non-processing period when executing this strategy is set to 95% of the original energy consumption. Combining the first three strategies, the final workshop energy-saving scheduling scheme is compared with the traditional non-strategy workshop scheduling scheme. Fig.14 In the 10 examples, the average total energy consumption saving is 30%, and the highest total energy consumption saving in MK08 is 51%.

[0175] The present invention comprehensively considers the two optimization objectives of maximum manufacturing span (MS) and total energy consumption (TEC), and realizes comprehensive optimization of job shop scheduling by constructing a multi-strategy energy-saving scheduling problem model. It clarifies constraints including the initial available time of the machine, the uniqueness of simultaneous processing procedures, the safe time limit for machine power on and off, and the processing speed determined by the workpiece quality requirements, thereby ensuring the feasibility and effectiveness of the scheduling plan.

[0176] At the same time, the MOEA / D algorithm is used to solve the multi-strategy energy-saving scheduling problem model. The allocation and energy consumption settings of jobs are defined by simulating two-layer coding, which improves the solution efficiency and accuracy. The three methods of random initialization, local load minimum initialization and global load minimum initialization are combined to perform non-dominated resettable initialization on the population, which enhances the diversity and convergence of the population. According to the simulating two-layer vector coding, the job-based crossover operator and elite retention strategy are adopted. By constructing neighborhoods, adaptive crossover and mutation operations, the population is continuously iterated and updated to obtain the optimal Pareto frontier solution.

[0177] Heuristic energy-saving strategy: Machine on / off strategy: When the machine is idle for too long, the power supply of the machine is cut off according to the set time threshold to save energy, effectively reducing the energy consumption of the workshop.

[0178] Adjust processing speed strategy: appropriately slow down the processing speed of a certain process while satisfying load balancing, and consider the combination of processing speeds of previous and subsequent processes to form the best energy scheduling, further achieving energy-saving effects.

[0179] Changing the processing speed combination strategy: Using the idea of ​​the depth-first search (DFS) algorithm to find the optimal processing speed combination, the energy scheduling scheme is optimized;

[0180] Combining the above-mentioned multiple heuristic energy-saving strategies, comprehensive optimization of job shop scheduling and energy saving and consumption reduction are achieved, which has significant technical innovation and practical application value.

[0181] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0182] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the invention, and the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A multi-strategy flexible job shop energy-saving scheduling method, characterized by: The following steps are involved: S1. Comprehensively consider the quality requirements, manufacturing span and energy consumption factors of job processing, build a multi-strategy energy-saving scheduling problem model, define the job set, process set, machine set and processing speed set, and take the maximum manufacturing span and total energy consumption as the optimization goals, set the problem constraints, and the multi-strategy energy-saving scheduling problem model is: S2. Using a decomposition-based multi-objective evolutionary algorithm to solve the multi-strategy energy-saving scheduling problem model in step S1, specifically including: S21, using simulated double-layer coding to define the allocation of jobs and the energy consumption setting of the jobs; S22, using three methods of random initialization, local load minimum initialization and global load minimum initialization to initialize the population non-dominated resettable; S23, by constructing neighborhoods, adaptive crossover and mutation operations, the population is continuously updated iteratively to obtain the optimal Pareto frontier solution; S3. In the solution process of step S2, a heuristic energy-saving strategy is applied to reduce the energy consumption of the job shop, wherein the heuristic energy-saving strategy includes a machine on / off strategy, a processing speed adjustment strategy, a processing speed combination change strategy, and a low-speed strategy under no-load state.

2. The multi-strategy flexible job shop energy-saving scheduling method according to claim 1 is characterized by: The multi-strategy energy-saving scheduling problem model in step S1 is defined as follows: There exists a job J = {J1, J2, ..., J n }, where each J p There is a set of independent processes O = {O p1 ,O p2 ,…,O pj },O i,j represents the jth process of the ith job, in the machine set M = {M1, M2, ..., M k } selects a machine to process, any machine from V = {V1, V2, ..., V s Select speed processing O in i,j , usually V→V i,j , that is, the processing speed on the machine is mapped to O i,j Processing speed (O i,j ∈O), considering the control of processing speed, according to O i,j Q i,j Get V i,j (Q i,j ∈Q,V i,j ∈V), in machine M k Upper processing O i,j Used time T i,j,k The energy consumption caused is calculated as follows: EO i,j,k *T i,j,k .

3. The multi-strategy flexible job shop energy-saving scheduling method according to claim 1 is characterized by: The constraint conditions in step S1 are: The initial available time of each machine starts from 0; Each machine processes only one process at a time; Each machine's on / off is subject to a safety time limit, i.e. the interval between shutting down and starting up. The machine's processing speed is determined by the processing speed of the workpiece; The processing speed of the workpiece is limited by the processing quality requirements of the workpiece; The sequence of different processes in the same operation is fixed.

4. The multi-strategy flexible job shop energy-saving scheduling method according to claim 1 is characterized by: The simulated double-layer coding in step S21 includes a job allocation code and an energy consumption setting code. The job allocation code is used to determine which machine each process is assigned to for processing, and the energy consumption setting code is used to determine the processing speed of each process on the selected machine, thereby further determining the energy consumption.

5. The multi-strategy flexible job shop energy-saving scheduling method according to claim 1 is characterized by: The adaptive crossover and mutation operation steps in step S23 are based on simulated double-layer vector coding, and adopt a job-based crossover operator and an elite retention strategy.

6. The multi-strategy flexible job shop energy-saving scheduling method according to claim 1 is characterized by: The machine on / off strategy in step S3 is to cut off the power supply of the machine according to a set time threshold to save energy when the machine is idle for too long.

7. The multi-strategy flexible job shop energy-saving scheduling method according to claim 1 is characterized by: The processing speed strategy is adjusted in step S3, including appropriately slowing down the processing speed of a certain process while satisfying load balancing, and considering the processing speed combination of the previous and next processes to form the best energy scheduling.

Images