CN-122022273-A - Energy-saving multi-objective scheduling optimization method for battery module two-stage manufacturing

CN122022273ACN 122022273 ACN122022273 ACN 122022273ACN-122022273-A

Abstract

The invention provides an energy-saving multi-objective scheduling optimization method for battery module two-stage manufacturing, and belongs to the technical field of intelligent manufacturing and production scheduling optimization. The method is used for optimizing the maximum finishing time and the total cost under the condition of considering the time-of-use electricity price, job cluster compatibility, batch processing machine capacity and sequence related set time. The method adopts a two-stage scheduling multi-objective evolutionary optimization algorithm based on hierarchical ECDF-PDS environment selection, represents the operation priority and the insertion behavior through double-layer hybrid coding, sequentially executes two-stage collaborative decoding to generate feasible scheduling, combines mask hybrid crossover and self-adaptive variation to carry out population evolution, utilizes ECDF mapping and hierarchical PDS strategy to carry out environment selection, and finally outputs a group of non-dominant scheduling schemes balanced in time and cost targets. The invention can effectively solve the complex scheduling problem of two stages of laser welding and heat treatment curing in the manufacturing of the new energy battery module, and improves the production efficiency and the economic benefit.

Inventors

ZHANG HAN
WANG YILAN
CHEN WENHE
WANG WEIZHONG
TAN WEIMIN
KONG MIN
GAO YUTONG
TAO YATING

Assignees

安徽师范大学

Dates

Publication Date: 20260512
Application Date: 20260106

Claims (10)

1. The energy-saving multi-objective scheduling optimization method for the two-stage manufacturing of the battery module is characterized by comprising the following steps of: initializing algorithm parameters including population scale Maximum algebra of evolution Environmental selection parameters Let the current iteration number Randomly generating an initial population Each individual in the population is represented by a double-layered mixed chromosome, which is represented by a job priority vector Inserting behavior vectors Constructing; step two, for population Each individual in the list performs decoding operation to obtain a corresponding scheduling scheme and a target value; Step three, if Step eleven, otherwise, initializing the offspring population Make the child individual counter Turning to the fourth step; Step four, if Turning to step eight, otherwise, selecting the population from Selecting two father individuals and recording the chromosome as And Turning to a step five; Fifth step, pair And Performing a mask mix crossover operation to generate two child chromosomes And ; Step six, pairing And Respectively executing adaptive mutation operation to obtain mutated chromosomes; Step seven, will be Adding in Order-making If (1) Will then Adding in Order-making Turning to the fourth step; step eight, population of child generations Performing decoding operation to obtain target value, and collecting parent population And offspring populations Combining to obtain a combined population ; Step nine, combining the population Performing hierarchical PDS environment selection operation based on ECDF mapping to obtain a new generation population ; Step ten, order Turning to the third step; Step eleven, outputting the population Is a non-dominant solution set in (a), the algorithm ends.
2. The method for optimizing energy-saving multi-objective scheduling for battery module two-stage manufacturing of claim 1, wherein the operation priority vector in the first step is Length is Is used to determine the vector of the real number, The numerical value is used for determining the ordering sequence of the operation and is divided into two sections structurally, wherein the first section is the first section A gene for determining the job-ordering priority of the first stage, the second stage being the first stage A gene for determining job sequencing priority of the second stage, inserting a behavior vector Length is Is used to determine the binary vector of (c), Wherein Indicating that the device or lot with the earliest predicted completion time is preferentially selected within the feasible set, Indicating that the lowest energy cost equipment or process time period is preferred under the TOU price curve.
3. The method for optimizing energy-saving multi-objective schedule for battery module two-stage manufacturing according to claim 1, wherein the decoding operation in the second step is chromosomal Comprises the following sub-steps: step 1, chromosome is subjected to Represented as a2× Wherein the first behavior is job priority Second behavior insert behavior Dividing it into two sub-matrices for part of the first stage From the front Column composition, part for the second stage From behind A column composition; Step 2, executing the first stage decoding to obtain a first stage scheduling scheme Time set for finishing operation And machine allocation map ; Step 3, executing the second stage decoding to obtain a second stage scheduling scheme Time set for finishing operation And machine allocation map ; Step 4, based on complete dispatch Calculating maximum finishing time And total cost of Constructing a target vector 。
4. The energy-saving multi-target scheduling optimization method for battery module two-stage manufacturing according to claim 3, wherein the first stage decoding in the step 2 comprises the following sub-steps: step a, according to A kind of electronic device Gene value will work together Arranged in ascending order to obtain a job sequence list ; Initializing each machine Is a scheduling sequence of (b) Time for completion of each job ; Step c, pressing Sequentially processing each job : Step i-for each machine Calculating the basic start time Wherein Is a machine The finishing time of the last operation is 0 if the last operation is empty. Step ii-calculating the machining on the machine Is to be completed in the time of completion of (a) Wherein Setting time for sequence correlation, if the operation Belonging to a different cluster than the previous job on the machine Then is Otherwise, 0. Step iii, calculating the energy consumption cost 。 Step iv according to A kind of electronic device Gene value selection machine If (if) then Then select to make Minimum machine, tie time selection Smaller machines, if Then select to make Minimum machine, tie time selection Smaller machines, if And the machine On which there is a feasible idle time window Then the time point for minimizing the energy consumption cost is selected in the time window As a start time; step v, determining the operation Is to be completed in the time of completion of (a) If the start time is adjusted, it is Otherwise, it is Will work Joining machine Is a scheduling sequence of (b) 。
5. The energy-saving multi-target scheduling optimization method for battery module two-stage manufacturing according to claim 3, wherein the second stage decoding in the step 3 comprises the following sub-steps: step a, according to A kind of electronic device Gene value will work together Arranged in ascending order to obtain a job sequence list ; Initializing each machine Is a lot dispatch sequence of (2) Time for completion of each job ; Step c, pressing Sequentially processing each job : Step i, calculating the operation Reach each machine Time of (2) ; Step ii for each machine If (if) Time to finish If not, the first part of the first part is connected with the second part, ; Step iii building a non-dominated set of machines Included in the two-dimensional evaluation vector A machine that is not governed by other machines; Step iv, giving priority to energy to be combined The machines inserted into the same cluster and having less than full capacity in the existing batch are denoted as a collection If (if) Order in principle ; Step v-from Is selected to cause Minimal machine as At tie time selection Smaller machines, then randomly selected in the tie; step vi in a machine Updating the batch, if the same cluster is not full Will then Adding and updating the batch processing time If not, creating new batch Processing time of it ; Step vii-determination of batch start time If (if) Then To meet the earliest time of job arrival and machine availability, if Selecting a start time for minimizing the processing energy consumption cost of the batch in a feasible range; Step viii computing the job Is to be completed in the time of completion of (a) And dispensing machine 。
6. The energy-saving multi-target scheduling optimization method for battery module two-stage manufacturing according to claim 1, wherein the mask mixing and crossing operation in the fifth step comprises the following sub-steps: step 1, randomly generating a length of Is a binary mask vector of (1) And And is in front of Bit and postamble Within the bits, the number of 0 and 1 are each equal; Step 2 for Layer, generation of offspring And : For the following Will be In (a) A position of value 1 Direct copying of values to At the same position, and then In (a) A value of 1 The values are filled in their original order Is a blank position of (1); For the following Will be In (a) Position with value 0 Direct copying of values to At the same position, and then In (a) A value of 0 The values are filled in their original order Is a blank position of (1); Step 3 for A layer using mask vector using the same flow as step 2 For a pair of The values are crossed to obtain And A kind of electronic device Part(s).
7. The energy-saving multi-objective scheduling optimization method for battery module two-stage manufacturing according to claim 1, wherein the adaptive mutation operation in the step six specifically comprises the following sub-steps: Step 1, calculating the current variation probability Wherein , ; Step 2, generating random number If (if) If not, turning to the step 3; Step 3, pair Each real gene of the layer Adopts the distribution index as Polynomial variation of (2) to generate And calculate the disturbance term Pressing again Updating the gene value and cutting off the gene value to a feasible interval; Step 4, for Layer each binary gene Performing bit flipping to make it at And (3) with And exchanging the two.
8. The energy-saving multi-objective scheduling optimization method for battery module two-stage manufacturing according to claim 1, wherein the hierarchical PDS environment selection operation based on ECDF mapping in step nine specifically comprises the following sub-steps: step 1, combining population Non-dominant ordering to obtain ordered front layer list ; Step 2, initializing a new generation population Leading edge layer index When (when) When it will All individuals in (1) add Order-making Repeating the steps; Step 3, if Then directly output Otherwise, make the candidate set Number of individuals in need of supplementation ; Step 4, for population ECDF mapping is performed for each target Ranking individual target values in a population Normalizing to obtain mapping coordinates Wherein ; Step 5, from Medium screening Personal addition Firstly, selecting the extreme point with the minimum mapping value of each target direction, if If the score is null, supplementing the knee point closest to the ideal point, and selecting the individual with the highest comprehensive score by greedy iteration through the residual scores Wherein At the level of the minimum distance to be reached, In order to converge the weights, Is a density penalty term.
9. The method for optimizing energy-saving multi-objective scheduling for battery module two-stage manufacturing according to claim 8, wherein the density penalty term in step 5 Wherein Is that At the position of In space to the first The average distance of the neighbors is calculated, Is a very small constant.
10. The method for optimizing energy-saving multi-objective scheduling for battery module two-stage manufacturing according to claim 1, wherein the total cost is Wherein For the total energy consumption of the first stage, For the total energy consumption of the second stage, Representing a job Whether or not from Is transported to , Is a transportation cost.

Description

Energy-saving multi-objective scheduling optimization method for battery module two-stage manufacturing Technical Field The invention relates to the technical field of scheduling optimization in a manufacturing process, in particular to an energy-saving multi-objective scheduling optimization method for two-stage manufacturing of a battery module. Background With the rapid development of the new energy automobile industry, the manufacturing process of the battery module sets higher requirements on equipment coordination and process stability. Typical module manufacturing processes include multiple links such as cell housing, structural assembly, laser welding, thermal curing, etc. Wherein, laser welding and heat treatment curing are key links with strict process control and tight beat engagement. The two phases exhibit different scheduling characteristics, respectively. The laser welding stage often considers the parallel machine scheduling problem of the sequence related set time, while the heat treatment curing stage usually considers the parallel batch machine scheduling problem with cluster compatibility constraint, and needs to consider batch processing and capacity limitation. More importantly, a tight timing and process linkage relationship exists between the two stages. The finishing time of the preceding operation determines the arrival time of the following operation, and the batch processing efficiency of the subsequent stage depends on the synchronous finishing of the preceding operation. The tight connection and the mutual restriction enable the scheduling decisions of the two stages to mutually influence, and the segmentation optimization cannot be realized. In addition, modern green manufacturing also requires a mechanism to take into account the price of electricity in scheduling, further increasing the complexity of the trade-off between minimizing maximum finishing time and minimizing total energy costs. The existing scheduling method is difficult to effectively solve the complex problems of multi-stage, multi-constraint and multi-target cooperative scheduling. Therefore, a scheduling method capable of integrally co-optimizing two-stage scheduling to simultaneously optimize the completion time and the total cost is needed. Disclosure of Invention The invention aims to solve the technical problems that the scheduling scheme of two stages of laser welding and heat treatment curing in the existing battery module manufacturing is complex, and the maximum finishing time and total cost are difficult to cooperatively optimize while the time-of-use electricity price and the compatibility of an operation cluster are difficult to consider, so that the production efficiency and the economic benefit are difficult to balance. The technical problem is solved by the technical means that a two-stage scheduling multi-objective evolution optimization method HEP-MOEA based on hierarchical ECDF-PDS environment selection is provided. The method is used for solving a method to minimize the maximum finishing timeAnd minimizing the total costScheduling problems for the goal, the total costIncluding the energy costs of both processing stages and the transportation costs between the stages. The problem is limited by job cluster compatibility, batch machine capacity, sequence dependent setup time, and time-of-use electricity pricing mechanisms. The method comprises the following steps: initializing algorithm parameters including population scale Maximum algebra of evolutionEnvironmental selection parametersLet the current iteration number. Randomly generating an initial populationEach individual in the population is represented by a double-layered mixed chromosome, which is represented by a job priority vectorInserting behavior vectorsThe composition is formed. Step two, for populationEach individual of (3)Performing decoding operation to obtain corresponding scheduling schemeAnd the target vector。 Step three, ifStep eleven, otherwise, initializing the offspring populationMake the child individual counterGo to step four. Step four, ifTurning to step eight, otherwise, selecting the population fromSelecting two father individuals and recording the chromosome asAndTurning to step five. Fifth step, pairAndPerforming a mask mix crossover operation to generate two child chromosomesAnd。 Step six, pairingAndAnd respectively executing adaptive mutation operation to obtain mutated chromosomes. Step seven, will beAdding inOrder-makingIf (1)Will thenAdding inOrder-making. Turning to step four. Step eight, population of child generationsThe decoding operation is performed as in step two to obtain the target value thereof. Will father and mother populationAnd offspring populationsCombining to obtain a combined populationThe size is。 Step nine, combining the populationPerforming hierarchical PDS environment selection operation based on ECDF mapping to obtain a new generation populationThe size is。 Step ten, order,Turning to step t