CN114415615A - Mixed-flow assembly line balance distribution method and device under uncertain demand - Google Patents
Mixed-flow assembly line balance distribution method and device under uncertain demand Download PDFInfo
- Publication number
- CN114415615A CN114415615A CN202210060798.3A CN202210060798A CN114415615A CN 114415615 A CN114415615 A CN 114415615A CN 202210060798 A CN202210060798 A CN 202210060798A CN 114415615 A CN114415615 A CN 114415615A
- Authority
- CN
- China
- Prior art keywords
- target
- distribution
- robot
- distributed
- mutation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 310
- 238000009826 distribution Methods 0.000 title claims abstract description 129
- 230000008569 process Effects 0.000 claims abstract description 192
- 238000012545 processing Methods 0.000 claims abstract description 112
- 230000035772 mutation Effects 0.000 claims abstract description 84
- 230000009471 action Effects 0.000 claims description 26
- 238000004590 computer program Methods 0.000 claims description 11
- 238000003780 insertion Methods 0.000 claims description 10
- 230000037431 insertion Effects 0.000 claims description 10
- 238000003860 storage Methods 0.000 claims description 8
- 125000004122 cyclic group Chemical group 0.000 claims description 5
- 238000004519 manufacturing process Methods 0.000 description 16
- 230000006870 function Effects 0.000 description 8
- 238000011156 evaluation Methods 0.000 description 6
- 238000005457 optimization Methods 0.000 description 6
- 230000008859 change Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 239000011159 matrix material Substances 0.000 description 4
- NUFBIAUZAMHTSP-UHFFFAOYSA-N 3-(n-morpholino)-2-hydroxypropanesulfonic acid Chemical compound OS(=O)(=O)CC(O)CN1CCOCC1 NUFBIAUZAMHTSP-UHFFFAOYSA-N 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000006872 improvement Effects 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 2
- 238000013459 approach Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000002787 reinforcement Effects 0.000 description 2
- 230000000717 retained effect Effects 0.000 description 2
- 241001522296 Erithacus rubecula Species 0.000 description 1
- 241001415395 Spea Species 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000008094 contradictory effect Effects 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B19/00—Programme-control systems
- G05B19/02—Programme-control systems electric
- G05B19/418—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM]
- G05B19/41885—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM] characterised by modeling, simulation of the manufacturing system
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/30—Nc systems
- G05B2219/32—Operator till task planning
- G05B2219/32339—Object oriented modeling, design, analysis, implementation, simulation language
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Landscapes
- Engineering & Computer Science (AREA)
- Manufacturing & Machinery (AREA)
- General Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a balanced distribution method and a balanced distribution device for a mixed flow assembly line under uncertain requirements, wherein the method comprises the following steps: acquiring a process set to be distributed, a priority value of each process to be distributed and a candidate robot set; randomly generating distribution coding groups with target quantity based on the priority value of each procedure to be distributed; determining a crossover operator, and performing crossover processing on the distribution code groups of the target number to obtain reference code groups of the target number; determining a mutation operator, and carrying out mutation processing on the allocated code groups of the target number to obtain alternative code groups of the target number; and obtaining an updated distribution coding group based on the distribution coding group, the reference coding group and the alternative coding group, iterating the cross processing and the variation processing, changing the cross probability and the variation probability in each iteration process until the iteration duration exceeds a duration threshold, and outputting the distribution coding group updated by the last iteration. The assembly cost can be reduced, the assembly requirement is met, the application scene is enlarged, and the assembly work efficiency is improved.
Description
Technical Field
The invention relates to the technical field of factory scheduling, in particular to a balanced distribution method and device for a mixed flow assembly line under uncertain requirements.
Background
With the emergence of large-scale customized requirements, modern manufacturing enterprises design a flexible multi-robot assembly line layout mode on the basis of an intelligent assembly technology and a new-generation information technology to meet the requirement of multi-product large-scale assembly. In real-world assembly, due to the dynamics and uncertainty of product demand, conventional assembly line configurations cannot guarantee that all the required products are completed within a cycle time, thereby greatly affecting production schedules and creating economic losses. Therefore, there is a need to develop a balanced distribution method for a mixed flow assembly line under uncertain requirements to meet the requirement of large-scale customization of products.
The prior multi-robot mixed-flow assembly line balanced distribution method neglects that the demand of the product is uncertain and can be influenced by daily change. In a mixed robotic assembly line, all tasks are fixed and assigned to specific operators and workstations. If the requirement changes, the task time also changes, the assembly requirement cannot be met, the application scene is limited, and the working efficiency is low.
Disclosure of Invention
The invention provides a balanced distribution method and a balanced distribution device for a mixed flow assembly line under uncertain requirements, which are used for overcoming the defects that the assembly requirements cannot be met, the application scenes are limited and the working efficiency is lower in the prior art, reducing the assembly cost, meeting the assembly requirements, expanding the application scenes and improving the working efficiency of assembly.
The invention provides a balanced distribution method for a mixed flow assembly line under uncertain requirements, which comprises the following steps:
acquiring a process set to be distributed, a priority value of each process to be distributed and a candidate robot set;
under the condition that an assembly line distribution constraint condition is met, randomly generating a target number of distribution coding groups based on the priority values of all the procedures to be distributed, wherein each procedure to be distributed in the distribution coding groups is executed by one candidate robot;
determining a crossover operator, and performing a crossover processing process on the target number of allocation code groups based on the crossover operator to obtain a target number of reference code groups, wherein the assembly cost of the reference code groups is less than that of the allocation code groups;
determining a mutation operator, and executing a mutation processing process on the target number of distributed coding groups based on the mutation operator to obtain target number of alternative coding groups, wherein the assembly cost of the alternative coding groups is less than that of the distributed coding groups;
obtaining an updated allocation code group based on the allocation code group, the reference code group and the alternative code group, iteratively executing the cross processing process and the mutation processing process, changing the probability of the cross processing process and the probability of the mutation processing process in each iteration process until the iteration duration exceeds a duration threshold, and outputting the updated allocation code group obtained by the last iteration.
According to the mixed flow assembly line balance distribution method under the uncertain requirement provided by the invention, the determining cross operator comprises the following steps:
acquiring a cross processing Q value table, wherein the initial value of the cross processing Q value table is 0;
after each iteration process, based on the formula:
updating the cross-process Q-value table, wherein newQs，aRepresenting the updated Q value corresponding to the current state s and the current action a; qs，aRepresenting the current Q value corresponding to the current state s and the current action a;
representing the maximum Q value in the next state, estimated from the current Q value, Rs，aA reward value representing the taking of action a in the current state s; α and γ represent importance coefficients;
determining the crossover operator from a double-point crossover, a single-point crossover, a sequential crossover, a localized crossover, and a cyclic crossover based on the crossover processing Q-value table.
According to the mixed flow assembly line balance distribution method under the uncertain requirements, the method for determining the mutation operator comprises the following steps:
obtaining a variation processing Q value table, wherein the initial value of the variation processing Q value table is 0;
after each iteration process, based on the formula:
updating the mutation processing Q value table, wherein newqs，aRepresenting the updated Q value corresponding to the current state s and the current action a; q. q.ss，aRepresenting the current Q value corresponding to the current state s and the current action a;
representing the maximum Q value in the next state, estimated from the current Q value, rs，aA reward value representing the taking of action a in the current state s; α and γ represent importance coefficients;
determining the mutation operator from the swapping, the forward insertion, and the backward insertion based on the mutation processing Q-value table.
According to the mixed flow assembly line balanced distribution method under the uncertain requirements provided by the invention, the method for changing the probability of the cross processing process and the probability of the variation processing process in each iteration process comprises the following steps:
based on the formula:
P＝1-CR；
determining a probability of a crossover process or a mutation process, wherein T represents a current running time and n.n.p represents the duration threshold.
According to the mixed-flow assembly line balanced allocation method under uncertain requirements provided by the invention, under the condition of meeting the assembly line allocation constraint condition, the allocation coding groups with target number are randomly generated based on the priority value of each process to be allocated, and the method comprises the following steps:
step X, selecting the procedure to be distributed with the maximum priority value from the procedure set to be distributed as a target procedure, and deleting the target procedure from the procedure to be distributed;
step Y, selecting the robot with the earliest assembling time of the target working procedure from the candidate robot set as a target robot, and deleting the target robot from the candidate robot set;
and traversing the step X and the step Y until the process set to be distributed is an empty set, and determining the distribution coding group based on the target process and the corresponding target robot.
According to the mixed flow assembly line balance distribution method under uncertain requirements provided by the invention, the robot with the earliest target process time is selected from the candidate robot set to be used as a target robot, and the method comprises the following steps:
and if at least two robots with the earliest target process time are present in the candidate robot set, selecting the robot with the smallest idle time between the target process and the previous process from the at least two robots with the earliest target process time as the target robot.
According to the mixed flow assembly line balance distribution method under uncertain requirements provided by the invention, the robot with the earliest target process time is selected from the candidate robot set to be used as a target robot, and the method further comprises the following steps:
and if the number of the robots with the smallest idle time between the target process and the previous process is at least two selected from the robots with the earliest time for assembling the target process, selecting the robot with the smallest label as the target robot.
The invention also provides a balance distribution device for the mixed flow assembly line under the uncertain requirements, which comprises:
the acquisition module is used for acquiring a process set to be distributed, priority values of all processes to be distributed and a candidate robot set;
the generating module is used for randomly generating a target number of distribution coding groups based on the priority values of all the procedures to be distributed under the condition of meeting the distribution constraint conditions of the assembly line, wherein each procedure to be distributed in the distribution coding groups is executed by one candidate robot;
the cross processing module is used for determining cross operators and executing a cross processing process on the distribution coding groups of the target number based on the cross operators to obtain reference coding groups of the target number, and the assembly cost of the reference coding groups is less than that of the distribution coding groups;
a mutation processing module, configured to determine a mutation operator, and perform a mutation processing process on the target number of allocated coding groups based on the mutation operator to obtain a target number of candidate coding groups, where an assembly cost of the candidate coding groups is less than an assembly cost of the allocated coding groups;
and the output module is used for obtaining an updated distribution code group based on the distribution code group, the reference code group and the alternative code group, iteratively executing the cross processing process and the variation processing process, changing the probability of the cross processing process and the probability of the variation processing process in each iteration process until the iteration duration exceeds a duration threshold, and outputting the updated distribution code group obtained by the last iteration.
The invention also provides an electronic device, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor executes the program to realize the steps of the mixed flow assembly line balanced distribution method under the uncertain requirements.
The present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program that, when executed by a processor, performs the steps of the mixed flow assembly line balanced allocation method under uncertain demand as any of the above.
According to the mixed flow assembly line balanced distribution method and device under the uncertain requirements, the distributed coding groups are iteratively updated by designing the crossover operator and the mutation operator, the most balanced distribution scheme of the working procedures and the robot is determined under the uncertain requirements, the assembly cost can be reduced, the assembly requirements are met, the application scene is expanded, and the assembly working efficiency is improved.
Drawings
In order to more clearly illustrate the technical solutions of the present invention or the prior art, the drawings needed for the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 is a schematic flow chart of a mixed flow assembly line balanced distribution method under uncertain demands provided by the invention;
FIG. 2 is a schematic structural diagram of a mixed flow assembly line balanced distribution device under uncertain requirements provided by the invention;
fig. 3 is a schematic structural diagram of an electronic device provided in the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The uncertain demand mixed flow assembly line balance distribution method and device of the invention are described below with reference to fig. 1 to 3.
The uncertain demand mixed flow assembly line balance distribution method and device are mainly used for demand order distribution of automobile and electronic manufacturing and assembly workshops, the manufacturing workshops can be provided with a plurality of stations and a plurality of robots, the stations and the robots can correspond to each other, and the uncertain demand mixed flow assembly line balance distribution method and device mainly aim to achieve efficient distribution of demand orders under the condition that demands are uncertain so as to reduce production cost and punishment cost, wherein the punishment cost is generated by overdue delivery of products.
The invention takes cost as guidance and focuses on researching the problem of multi-robot mixed flow assembly balance under uncertain requirements. Managers seek to reduce the number of robots and workstations to reduce production costs, but do not want to incur penalty costs due to overdue completion. Production cost and penalty cost are often contradictory goals, with a minimum production cost potentially resulting in a higher penalty cost and vice versa. The method adopts a Pareto mode to simultaneously reduce the production cost and the punishment cost, constructs a multi-objective mixed integer optimization scheme, and designs a multi-objective evolutionary algorithm based on reinforcement learning to solve the optimal result of the scheme, namely the optimal allocation mode of the demand order.
As shown in FIG. 1, the present invention provides an uncertain demand mixed flow assembly line balance distribution method, which comprises the following steps 110 to 150.
Wherein, step 110: and acquiring a process set to be distributed, a priority value of each process to be distributed and a candidate robot set.
It will be understood that the set of processes to be distributed is the set of demand orders to be distributed, the set of candidate robots is the set of robots that can be used to process the processes to be distributed, a new workstation wc +1 can be created, and the workstation contains umaxA personal robot. A candidate set of alternative procedures may be constructed. Whether a process to be allocated can be stored in the set of processes to be allocated requires ensuring that the process has not been allocated before and that processes before the process have been allocated.
The priority value of the procedure to be allocated refers to the priority of the procedure to be allocated. If a robot can assemble the selected procedure to be allocated under all demand plans, the robot may be deposited in a set of candidate robots. If none of the robots meet the above requirements, it indicates that the current workstation is fully loaded.
And 120, under the condition that the assembly line distribution constraint condition is met, randomly generating a target number of distribution code groups based on the priority values of all the processes to be distributed, wherein each process to be distributed in the distribution code groups is executed by one candidate robot.
It can be understood that, under the condition that the assembly line allocation constraint condition is satisfied, a target number of allocation code groups can be randomly generated according to the priority value sequence of the to-be-allocated processes, that is, the to-be-allocated processes and the candidate robots are allocated in a one-to-one correspondence manner, so as to obtain the pairing codes of the to-be-allocated processes and the candidate robots, where the assembly line allocation constraint condition may be a candidate robot in which one to-be-allocated process can be allocated with only one station, or a to-be-allocated process can be allocated only after the previous process is allocated, where a specific assembly line allocation constraint condition is not defined, and a person skilled in the art can select an appropriate assembly line allocation constraint condition according to actual situations.
N-dimensional vector pi ═ pi can be designed1，π2，...，πnRepresents a complete solution. Each element piiIndicates the priority value of step i. Each priority value is generated randomly and repeatedly, and ranges from 1 to n. To effectively reduce the idle time between sequences and remaining, the processes to be allocated are allocated to specific robots and work stations in a process-robot oriented manner, provided that all constraints are met.
In some embodiments, the process to be allocated and the allocation scheme of the candidate robot matching trip need to satisfy the limitation of production cost and penalty cost, the production cost and the penalty cost can be represented by an objective function and a constraint function condition thereof, the objective function can be composed of a formula one and a formula two, and the constraint condition can be composed of a formula three to a formula twenty three.
The first and second equations of the objective function may be expressed as:
the formula I is as follows: minf1＝∑j∈J∑k∈KYjk·COO+∑j∈JZj·COW；
The formula II is as follows: minf2＝Σj∈J∑k∈K∑e∈EOjke·CPO+∑j∈J∑e∈ERje·CPW；
Wherein, formula one represents the minimized robot production cost, and the production cost comprises: and the lease cost and the station opening cost are calculated, and the formula II represents the punishment cost of the robot and the station exceeding the beat.
The formula III is as follows:
and the formula III represents a candidate robot which is distributed to a station by each process to be distributed.
The formula four is as follows:
the formula four represents a priority relationship constraint, that is, a process to be allocated can only be allocated after the previous process is allocated.
The formula five is as follows:
formula six:
wherein formula five and formula six represent robust tempo constraints, i.e. the completion time for each workstation is not allowed to exceed the maximum value of tempo.
The formula seven:
wherein, the formula seven represents the start time of the restraint control process according to the priority relationship.
The formula eight:
the formula is nine:
wherein, the formula eight and the formula nine represent the starting time of the process without priority relation under the same robot.
Formula ten:
wherein equation ten represents the lower bound of the completion time.
Formula eleven:
equation twelve:
formula thirteen:
wherein formula eleven defines a binary variable YjkI.e. if there is a process assigned to a certain robot at a certain station, its corresponding YjkThe value is 1.
The formula fourteen:
equation fifteen:
the formula sixteen:
wherein the formula fourteen to sixteenth define a binary variable ZjI.e. if a process is assigned to a station, the corresponding ZjThe value is 1.
The formula seventeen:
eighteen formulas:
the formula is nineteen:
wherein the seventeenth to nineteenth formulas define a binary variable XGijkeI.e. only when the variable GieAnd XijkWhile 1, the corresponding XGijkeThe value can be 1.
The formula twenty:
the formula twenty one:
wherein the equations twenty to twenty-one control the variable O in a manner similar to the equations seventeen to nineteenjke。
The formula twenty-two:
the formula twenty-three:
wherein the formulas twenty-two to twenty-three adopt the similar way of the formulas seventeen to nineteen to define the binary variable Rje。
It should be noted that the physical meanings of the relevant parameters existing in formula one to twenty-three are described as table one: the meaning table of formula parameters.
Table one: formula parameter meaning table
The physical meanings of the relevant variables present in equations one through twenty-three are illustrated in table two: the formula variables have the meaning shown.
Table two: meaning of formula variables
It can be understood that, because different crossover operators will produce different sub-solutions and have a wide exploration capability, the crossover operator is used to perform crossover processing on the allocation code group determined in the above step to update the allocation code group, which is an iterative update process, after the allocation code group is updated, the obtained reference code group and the allocation code group have the same number, in other words, after each allocation code group is subjected to crossover processing by the crossover operator, a reference code group can be obtained, the assembly cost is equal to the sum of the production cost obtained by the above formula one and the penalty cost obtained by the above formula two, the obtained reference code group can be substituted into the above formula one to formula twenty three to obtain the assembly cost, and the assembly cost of the reference code group is less than the assembly cost of the allocation code group. In this way, a more distributed coding scheme can be obtained than a distributed coding group, i.e. the assembly cost of the updated reference coding group is less than the assembly cost of the distributed coding group after the crossover operator processing.
Here, 5 problem-specific crossover operators can be designed: double-point crossing, single-point crossing, sequential crossing, localized crossing, and cyclic crossing.
The specific details of the above 5 crossover operators are as follows:
double-point crossing: two cross points are randomly selected, elements between the two points on the distribution coding group are deleted, and then the elements on the cross objects are filled in blank positions of the distribution coding group in sequence and without repetition.
Single-point crossing: randomly selecting a cross point, deleting the elements on the distribution coding group after the cross point, and filling the elements on the cross object to the blank positions of the distribution coding group in sequence without repetition.
And (3) sequential crossing: this approach is the opposite idea of two-point crossing, i.e., elements between two points on the assigned code group are retained, and then the elements on the crossing object are sequentially and non-repeatedly filled into blank positions of the assigned code group.
Positioning and crossing: randomly selecting a plurality of positions on the distribution coding group, and reserving elements on the positions; the elements on the cross-object are then filled sequentially and non-repeatedly to other positions of the assigned encoding group.
And (3) circulating and crossing: unlike the 4 interleaving approaches described above, the interleaving uses a round robin pattern to determine the positions of elements to be reserved for allocation of code groups. In particular, the first position s on the code set is allocated1Randomly selecting; then cross over the object at s1The elements in position are denoted as pi1；π1The position on the assigned code set is denoted s2(ii) a On intersecting object at s2The elements in position are denoted as pi2(ii) a Repeating the above operation until the position siIs equal to the first position s1. Elements at selected positions on the assigned encoding groups are retained and elements on the interleaved objects are sequentially and non-repetitively padded at other positions on the assigned encoding groups.
In each iteration, each solution in the population is regarded as an allocation code group, and a cross object is selected for each allocation code group by adopting a similarity selection strategy.
In some embodiments, the determining the crossover operator in step 130 includes: acquiring a cross processing Q value table, wherein the initial value of the cross processing Q value table is 0; after each iteration process, based on the formula twenty-four:
updating an intersection processing Q value table, wherein newQ _ (s, a) represents an updated Q value corresponding to the current state s and the current action a; q _ (s, a) represents the current Q value corresponding to the current state s and the current action a; maxQ _ (s, a) represents the maximum Q value in the next state, which is estimated from the current Q value, and R _ (s, a) represents the reward value for taking action a in the current state s; α and γ represent importance coefficients; based on the cross-processing Q-value table, a cross operator is determined from the double-point cross, the single-point cross, the sequential cross, the localized cross, and the cyclic cross.
It can be understood that, in order to ensure that each interleaving can generate different reference coding groups, the present embodiment proposes a selection strategy based on Q learning to select the interleaving operation for each allocated coding group.
In order to find the best crossover or mutation operation suitable for assigning code groups, the present embodiment proposes a new Q learning selection strategy. The strategy dynamically evaluates the improvement degree of different operations in each iteration. And selecting the operation capable of improving the Pareto front edge solution probability to the maximum according to the history and the current evaluation information. The strategy mainly comprises four elements: status, action, Q-value table, and reward function.
Based on the problem studied, the present embodiment designs three states: a start state, an improved state, and an unmodified state. Before the first iteration, no data can be referred to as no crossover operation is performed, and the initial state is set to the starting state. In each iteration, if the new solution generated by the corresponding crossover operation updates the external archive set, the current state is changed to an improved state; otherwise, the current state is an unmodified state.
Actions refer to interleaving operations. Thus, the crossover stage includes 5 actions, namely a two-point crossover, a one-point crossover, a sequential crossover, a localized crossover, and a cyclic crossover.
The Q value table is represented as an S × a two-dimensional matrix in the present embodiment, and is used for storing the decision information. Where S and A represent the number of states and actions, respectively. Therefore, this embodiment adopts
The matrix represents a table of Q values for the crossover stage. The initial value of the Q-value table is set to 0.
After different evolution operations are executed in each iteration, the corresponding Q-value table needs to be updated by using the bellman formula, that is, the Q-value table is updated based on the above formula twenty-four.
For the reward function, the present embodiment defines a reward value based on the number of new solutions that can update the external archive set. If there are v new solutions that can update the external archive set, then the corresponding reward value is added v, otherwise the corresponding reward value is subtracted by the population size value PS.
In summary, before each iteration, the cross process with the largest Q value at the current state needs to be selected to generate a new solution. After all operations are executed, the Q value table needs to be updated according to the Bellman formula, and the state of the next iteration is determined.
And 140, determining mutation operators, and performing mutation processing on the target number of the allocated code groups based on the mutation operators to obtain the target number of the alternative code groups, wherein the assembly cost of the alternative code groups is less than that of the allocated code groups.
It can be understood that, since different mutation operators will generate different sub-solutions and have a certain exploration capability, the assignment code group determined in the above step is mutated by using the mutation operator to update the assignment code group, which is also an iterative update process, after the distribution code groups are updated, the obtained alternative code groups have the same number as the distribution code groups, that is, after each distribution code group is subjected to mutation processing by a mutation operator, one alternative code group can be obtained, the assembly cost is equal to the sum of the production cost obtained by the formula one and the punishment cost obtained by the formula two, the alternative coding set obtained here can be solved by being substituted into the above formula one to formula twenty three, and the assembly cost of the alternative coding set is less than that of the allocation coding set. This allows to obtain a more distributed coding scheme than the distributed coding groups, i.e. the assembly cost of the updated candidate coding groups is less than the assembly cost of the distributed coding groups after the mutation operator processing.
Here, 3 problem-specific mutation operators can be designed: swap, insert forward, and insert backward.
The specific details of the above 3 mutation operators are as follows:
exchanging: randomly selecting two different positions s on the assigned code set1And s2(ii) a The two elements at that position are interchanged;
forward insertion: randomly selecting two different positions s on the assigned code set1And s2(ii) a Position s2Move the element on to position s1And is in position s1And s2Elements between-1 are shifted backwards one position in turn.
Backward insertion: randomly selecting two different positions s on the assigned code set1And s2(ii) a Position s1Move the element on to position s2And is in position s1+1 and s2The elements in between move forward one by one in turnLocation.
In some embodiments, the determining the mutation operator in step 140 includes: acquiring a variation processing Q value table, wherein the initial value of the variation processing Q value table is 0; after each iteration process, based on the formula twenty-five:
updating the mutation processing Q value table, wherein newqs，aRepresenting the updated Q value corresponding to the current state s and the current action a; q. q.ss，aRepresenting the current Q value corresponding to the current state s and the current action a;
representing the maximum Q value in the next state, estimated from the current Q value, rs，aA reward value representing the taking of action a in the current state s; α and γ represent importance coefficients;
determining the mutation operator from the swapping, the forward insertion, and the backward insertion based on the mutation processing Q-value table.
It is to be appreciated that, in order to ensure that each mutation can generate different sub-solution, the present embodiment further employs a selection strategy based on Q learning to select mutation operations for each assigned coding group.
In order to find the best mutation operation suitable for allocating coding groups, the application proposes a new Q learning selection strategy. The strategy dynamically evaluates the improvement degree of different operations in each iteration. And selecting the operation capable of improving the Pareto front edge solution probability to the maximum according to the history and the current evaluation information. The strategy mainly comprises four elements: status, action, Q-value table, and reward function.
Based on the problem studied, the present application has devised three states: a start state, an improved state, and an unmodified state. Before the first iteration, no data can be referred to because no mutation operation is performed, and the initial state is set as the starting state by the application. In each iteration, if the external archive set is updated by a new solution generated by the corresponding mutation operation, the current state is changed to an improved state; otherwise, the current state is an unmodified state.
Actions refer to mutation operations, and mutation stages comprise 3 actions, namely swapping, forward insertion, and backward insertion.
The Q value table is represented as an S × a two-dimensional matrix in the present embodiment, and is used for storing the decision information. Where S and A represent the number of states and actions, respectively. Therefore, the present application adopts
The matrix represents the Q value table of the variation stage. The initial value of the Q-value table is set to 0.
After different evolution operations are executed in each iteration, the corresponding Q value table needs to be updated by adopting the following Bellman formula. Namely, the data update is performed based on the above-mentioned formula twenty-five.
For the reward function, the present embodiment defines a reward value based on the number of new solutions that can update the external archive set. If there are v new solutions that can update the external archive set, then the corresponding reward value is incremented by v; otherwise, the corresponding reward value is subtracted by the population size value PS.
In summary, before each iteration, the dynamic variation process with the largest Q value at the current state needs to be selected to generate a new solution. After all operations are executed, the Q value table needs to be updated according to the Bellman formula, and the state of the next iteration is determined.
And 150, obtaining an updated distribution code group based on the distribution code group, the reference code group and the alternative code group, iteratively executing the cross processing process and the variation processing process, changing the probability of the cross processing process and the probability of the variation processing process in each iteration process until the iteration duration exceeds a duration threshold, and outputting the updated distribution code group obtained by the last iteration.
It can be understood that, in the conventional evolutionary algorithm, the probability of intersection and mutation is set to a fixed value and does not change with the change of the number of iterations. In fact, the effect of the crossover and mutation operations may fluctuate somewhat with subsequent increases in the number of iterations. Specifically, the improvement effect is reduced along with the time lapse of the cross operation; whereas mutation has the opposite effect. Therefore, the present embodiment proposes a time-based probability adaptive strategy to balance the dynamic probabilities of crossover and mutation operations.
The method specifically comprises the steps of changing the probability of the cross processing process and the probability of the variation processing process in each iteration process until the iteration duration exceeds a duration threshold, and outputting an updated distribution coding group obtained by the last iteration.
The obtained distribution coding group is an optimal distribution scheme, so that the assembly line robust balance scheme can be quickly and efficiently obtained through a multi-target mathematical optimization model under uncertain requirements and a multi-target evolutionary algorithm based on reinforcement learning.
According to the mixed-flow assembly line balanced distribution method under the uncertain requirements, the distributed coding group is iteratively updated by designing the crossover operator and the mutation operator, and the most balanced distribution scheme of the working procedure and the robot is determined under the uncertain requirements, so that the assembly cost can be reduced, the assembly requirements can be met, the application scene can be expanded, and the assembly work efficiency can be improved.
In some embodiments, altering the probability of the intersection process and the probability of the mutation process during each iteration comprises: based on the formula twenty-six:
P＝1-CR；
CR＝1.0-1/(n·n·p)·T；
a probability of a crossover process or a mutation process is determined, where T represents a current runtime and n.n.p represents a duration threshold.
It is understood that, since the crossover operation becomes good in the initial stage, the present embodiment sets the probability of the crossover processing procedure or the probability of the mutation processing procedure to 1.0 at the time of the first iteration. Because the proposed algorithm adopts time n · n · p as a termination condition, in the subsequent iteration process, the probability of the cross processing process or the probability of the mutation processing process linearly decreases with time, in the twenty-sixth formula, when the running time reaches n · n · p, the probability of the cross processing process or the probability of the mutation processing process also becomes 0, and the probability of the cross processing process or the probability of the mutation processing process is dynamically updated through 1-CR.
In some embodiments, in the case that the assembly line allocation constraint condition is satisfied, randomly generating a target number of allocation code groups based on the priority values of the respective processes to be allocated includes: step X, selecting the procedure to be distributed with the maximum priority value from the procedure set to be distributed as a target procedure, and deleting the target procedure from the procedure to be distributed; step Y, selecting the robot with the earliest assembly target process time from the candidate robot set as a target robot, and deleting the target robot from the candidate robot set; and traversing the step X and the step Y until the process set to be distributed is an empty set, and determining a distribution coding group based on the target process and the corresponding target robot.
It will be appreciated that a candidate set of alternative procedures may be constructed. Whether a process to be allocated can be stored in the set of processes to be allocated requires ensuring that the process has not been allocated before and that processes before the process have been allocated.
The priority value of the procedure to be allocated refers to the priority of the procedure to be allocated. If a robot can assemble the selected procedure to be allocated under all demand plans, the robot may be deposited in a set of candidate robots. If none of the robots meet the above requirements, it indicates that the current workstation is fully loaded.
To effectively reduce the idle time remaining between sequences, when a robot is selected from the set of candidate robots, the following requirements may be followed: and (5) selecting a robot which can assemble the procedures at the earliest to traverse the step X and the step Y until the set of the procedures to be distributed is an empty set, and determining a plurality of distribution coding groups based on each pair of matched target procedures and corresponding target robots.
In some embodiments, selecting a robot with the earliest assembly target process time from the set of candidate robots as the target robot includes: if at least two robots with the earliest assembly target process time appear in the candidate robot set, the robot with the smallest idle time between the target process and the previous process is selected from the robots with the earliest assembly target process time in the at least two robots with the earliest assembly target process time to be used as the target robot.
It is to be understood that if it appears that more than one robot meets the earliest assembly target process time based on the rule in the above embodiment, that is, when the robot with the earliest assembly target process time is selected from the candidate robot set, the robot with the smallest idle time between the target process and the previous process is selected as the target robot from among the at least two robots with the earliest assembly target process time.
In some embodiments, selecting, as the target robot, the robot with the earliest assembly target process time from the candidate robot set further includes: if the number of robots with the smallest idle time between the target process and the previous process is at least two selected from the robots with the earliest time of at least two assembly target processes, the robot with the smallest label is selected as the target robot.
It is understood that if it occurs that more than one robot meets the minimum idle time when selecting a robot having the minimum idle time between a target process and a previous process from among at least two robots whose assembly target process time is the earliest, based on the rule in the above-described embodiment, the robot having the smallest label is selected as the target robot.
It is worth mentioning that in the actual assembly enterprises, the order requirements are not fixed and the emergency order often occurs. And there is also a great deal of variability between different orders. In the prior art, an optimization model and an intelligent optimization algorithm are designed on the basis of fixing orders, and production fluctuation caused by order difference and difference of processing time between different robots and different products are ignored.
Compared with the prior art, the invention has the remarkable advantages that:
on the aspect of representing uncertain order requirements, a beat relaxation processing technology is adopted, so that the processing load of most stations is controlled to be beat, and the processing load of a few stations is controlled to be between the beat and the maximum relaxation beat under different order requirements. And ensures robust performance of the computational process by introducing penalty cost targets beyond the beat.
And in the aspect of processing uncertain order demands, a decoding strategy based on idle time reduction is provided, and the processes are reasonably distributed to a specific robot at a specific station in sequence according to an optimal weight. When the selectable process set and the selectable robot set are constructed, relaxed beat constraint is integrated, and feasibility of a scheduling scheme is improved.
In the third aspect, 5 kinds of cross operations based on specific problems and 3 kinds of mutation operations based on specific problems are designed in the aspect of optimizing a robust balance scheme; effectively coordinating a plurality of operation algorithms together through a selection strategy based on Q learning; a similarity selection strategy is proposed to replace a random selection strategy, and a cross object is selected for each cross parent; a time-based probability adaptive strategy is designed to dynamically update crossover and mutation probabilities.
Next, 269 reference test cases, which are not a limitation on the protection scope, are combined to further analyze the efficiency of the mixed-flow assembly line balanced distribution method under uncertain requirements provided by the present invention.
In the first aspect, case description and evaluation index setting are as follows:
in order to analyze the performance of the mixed-flow assembly line balanced distribution method under uncertain requirements, 269 reference test cases are divided into 22 case groups according to the size of the process quantity, and the case groups are respectively as follows: p7, P8, P9, P11, P21, P25, P28, P29, P30, P32, P35, P45, P53, P58, P70, P75, P83, P89, P94, P111, P148 and P297. Each case group contains 6, 1, 5, 9, 6, 7, 9, 6, 7, 10, 5, 16, 24, 16, 23, 13, 17, 35 and 26 cases, respectively.
The priority relationships and assembly times for cases may be downloaded from an assembly line balancing professional repository. All algorithms are programmed in the C + + language and run on a computer. 2 multi-target evaluation indexes are adopted to evaluate the convergence and the distribution of the mixed flow assembly line balance distribution method under the uncertain requirements.
Of these, 2 multi-objective evaluation indexes are respectively a hyper-volume ratio (HVR) and a convergence index (C (P, Q)). The closer the hyper-volume ratio (HVR) is to 1, the closer the acquired leading edge solution is to the true leading edge solution. The closer the convergence index (C (P, Q)) is to 1, the stronger the ability of the leading edge solution set Q to be dominated by the leading edge solution set P.
In a second aspect, the present invention compares results with a variety of optimization techniques
The invention compares RL-MOEA with 6 multi-objective optimization technologies to verify the superiority of the mixed flow assembly line balance allocation method under uncertain requirements. The comparison technique mainly comprises: EJAYA, ICA, MABC, MOPSO, MOSA, and SPEA 2.
The final comparative test results are shown in table 3 for 22 case groups: average HVR values for all algorithms, table 4: average C (RL-MOEA, …) values and Table 5: average C (…, RL-MOEA) values.
Table 3: average HVR value for all algorithms
From table 3, it can be found that the average HVR values of the EJAYA, ICA, MABC, MOPSO, MOSA, SPEA2 and RL-MOEA algorithms are: 0.88, 0.84, 0.89, 0.8, 0.83, 0.88 and 0.9. The average HVR value of the mixed flow assembly line balanced distribution method under the uncertain requirements is closest to 1, which shows that the mixed flow assembly line balanced distribution method under the uncertain requirements is superior to a compared multi-target algorithm in comprehensive evaluation of convergence and distribution.
TABLE 4 average C (RL-MOEA, …) values
| Case group | EJAYA | ICA | MABC | MOPSO | MOSA | SEPA2 |
| P7 | 0 | 0 | 0 | 0 | 0 | 0 |
| P8 | 0 | 0 | 0 | 0 | 0 | 0 |
| P9 | 0 | 0 | 0 | 0 | 0 | 0 |
| P11 | 0 | 0 | 0 | 0 | 0 | 0 |
| P21 | 0.24 | 0.35 | 0.2 | 0.34 | 0.26 | 0.25 |
| P25 | 0.01 | 0.16 | 0 | 0.13 | 0.01 | 0.02 |
| P28 | 0.02 | 0.17 | 0.1 | 0.21 | 0.14 | 0.01 |
| P29 | 0.48 | 0.78 | 0.4 | 0.82 | 0.71 | 0.47 |
| P30 | 0.31 | 0.57 | 0.37 | 0.59 | 0.45 | 0.34 |
| P32 | 0.39 | 0.85 | 0.31 | 0.77 | 0.58 | 0.29 |
| P35 | 0.67 | 0.82 | 0.58 | 0.73 | 0.73 | 0.6 |
| P45 | 0.16 | 0.22 | 0.13 | 0.25 | 0.23 | 0.14 |
| P53 | 0.09 | 0.31 | 0.06 | 0.33 | 0.21 | 0.13 |
| P58 | 0.9 | 0.99 | 0.83 | 0.98 | 0.99 | 0.84 |
| P70 | 0.73 | 0.81 | 0.61 | 0.81 | 0.78 | 0.57 |
| P75 | 0.89 | 0.99 | 0.63 | 1 | 1 | 0.87 |
| P83 | 0.87 | 0.99 | 0.9 | 0.99 | 1 | 0.84 |
| P89 | 0.91 | 1 | 0.86 | 1 | 1 | 0.86 |
| P94 | 0.72 | 0.91 | 0.63 | 0.93 | 0.91 | 0.64 |
| P111 | 0.8 | 0.92 | 0.74 | 0.9 | 0.91 | 0.71 |
| P148 | 0.85 | 0.92 | 0.63 | 0.95 | 0.94 | 0.82 |
| P297 | 0.94 | 0.98 | 0.68 | 1 | 0.99 | 0.85 |
| Mean value | 0.66 | 0.77 | 0.55 | 0.77 | 0.75 | 0.61 |
By comparing the data of tables 4 to 5, it can be found that there is a high probability that the leading edge solution set obtained by the comparison algorithm is dominated by the leading edge solution set obtained by the proposed algorithm. In contrast, the leading edge solution set acquired by the mixed-flow assembly line balanced allocation method under the uncertain requirements is difficult to be dominated by the leading edge solution set acquired by the comparison algorithm. Therefore, the mixed-flow assembly line balance distribution method under the uncertain requirements has good convergence in solving the balance problem of the mixed-flow assembly line of multiple robots under the uncertain order requirements.
TABLE 5 average C (…, RL-MOEA) values
Referring to fig. 2, the uncertain demand mixed flow assembly line balance distribution device provided by the present invention is described below, and the uncertain demand mixed flow assembly line balance distribution device described below and the uncertain demand mixed flow assembly line balance distribution method described above may be referred to with each other.
The invention also provides a balance distribution device for the mixed flow assembly line under the uncertain requirements, which comprises: an acquisition module 210, a generation module 220, a crossover processing module 230, a mutation processing module 240, and an output module 250.
The obtaining module 210 is configured to obtain a set of processes to be allocated, a priority value of each process to be allocated, and a candidate robot set.
And a generating module 220, configured to randomly generate a target number of allocation code groups based on the priority values of the to-be-allocated processes when the assembly line allocation constraint condition is satisfied, where each to-be-allocated process in the allocation code groups is executed by one candidate robot.
And the intersection processing module 230 is configured to determine an intersection operator, and perform an intersection processing process on the target number of allocated coding groups based on the intersection operator to obtain a target number of reference coding groups, where an assembly cost of the reference coding groups is less than an assembly cost of the allocated coding groups.
And a mutation processing module 240, configured to determine a mutation operator, and perform a mutation processing process on the target number of allocated coding groups based on the mutation operator to obtain a target number of candidate coding groups, where an assembly cost of the candidate coding groups is less than an assembly cost of the allocated coding groups.
And an output module 250, configured to obtain an updated assigned code group based on the assigned code group, the reference code group, and the candidate code group, iteratively execute the crossover processing procedure and the mutation processing procedure, change the probability of the crossover processing procedure and the probability of the mutation processing procedure in each iteration procedure until the iteration duration exceeds the duration threshold, and output the updated assigned code group obtained by the last iteration.
According to the mixed-flow assembly line balanced distribution device under the uncertain requirements, the distributed coding groups are iteratively updated by designing the crossover operator and the mutation operator, the most balanced distribution scheme of the working procedures and the robot is determined under the uncertain requirements, the assembly cost can be reduced, the assembly requirements can be met, the application scene can be expanded, and the assembly working efficiency can be improved.
As shown in fig. 3, the present invention also provides an electronic device.
The electronic device may include: a processor 310, a communication interface 320, a memory 330 and a bus 340, wherein the processor 310, the communication interface 320 and the memory 330 are communicated with each other through the bus 340.
Those skilled in the art will appreciate that the electronic device configurations shown in the figures do not constitute a limitation of electronic devices, and may include more or fewer components than shown in FIG. 3, or some components may be combined, or a different arrangement of components.
Wherein the processor 310 may call logic instructions in the memory 330 to perform a mixed-flow assembly line balanced allocation method under uncertain demand, the method comprising: acquiring a process set to be distributed, a priority value of each process to be distributed and a candidate robot set; under the condition of meeting the assembly line distribution constraint condition, randomly generating distribution coding groups with target quantity based on the priority values of all the procedures to be distributed, wherein each procedure to be distributed in the distribution coding groups is executed by one candidate robot; determining a crossover operator, and performing a crossover processing process on the distribution code groups of the target number based on the crossover operator to obtain reference code groups of the target number, wherein the assembly cost of the reference code groups is less than that of the distribution code groups; determining a mutation operator, and executing a mutation processing process on the target number of the distributed coding groups based on the mutation operator to obtain the target number of the alternative coding groups, wherein the assembly cost of the alternative coding groups is less than that of the distributed coding groups; and obtaining an updated distribution coding group based on the distribution coding group, the reference coding group and the alternative coding group, iteratively executing the cross processing process and the variation processing process, changing the probability of the cross processing process and the probability of the variation processing process in each iteration process until the iteration duration exceeds a duration threshold, and outputting the updated distribution coding group obtained by the last iteration.
The above operations can be implemented in the foregoing embodiments, and are not described in detail herein.
In addition, the logic instructions in the memory 330 may be implemented in the form of software functional units and stored in a computer readable storage medium when the software functional units are sold or used as independent products. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention.
The storage medium may include: read Only Memory (ROM), Random Access Memory (RAM), magnetic or optical disks, and the like.
In another aspect, the present invention also provides a computer program product, the computer program product including a computer program, the computer program being stored on a non-transitory computer readable storage medium, wherein when the computer program is executed by a processor, the computer is capable of executing the mixed flow assembly line balance distribution method under uncertain demand provided by the above methods, the method comprising: acquiring a process set to be distributed, a priority value of each process to be distributed and a candidate robot set; under the condition of meeting the assembly line distribution constraint condition, randomly generating distribution coding groups with target quantity based on the priority values of all the procedures to be distributed, wherein each procedure to be distributed in the distribution coding groups is executed by one candidate robot; determining a crossover operator, and performing a crossover processing process on the distribution code groups of the target number based on the crossover operator to obtain reference code groups of the target number, wherein the assembly cost of the reference code groups is less than that of the distribution code groups; determining a mutation operator, and executing a mutation processing process on the target number of the distributed coding groups based on the mutation operator to obtain the target number of the alternative coding groups, wherein the assembly cost of the alternative coding groups is less than that of the distributed coding groups; and obtaining an updated distribution coding group based on the distribution coding group, the reference coding group and the alternative coding group, iteratively executing the cross processing process and the variation processing process, changing the probability of the cross processing process and the probability of the variation processing process in each iteration process until the iteration duration exceeds a duration threshold, and outputting the updated distribution coding group obtained by the last iteration.
The above operations can be implemented in the foregoing embodiments, and are not described in detail herein.
In yet another aspect, the present invention also provides a non-transitory computer-readable storage medium having stored thereon a computer program, which when executed by a processor, implements the method for uncertain demand downmixed assembly line balanced distribution provided by the above methods, the method comprising: acquiring a process set to be distributed, a priority value of each process to be distributed and a candidate robot set; under the condition of meeting the assembly line distribution constraint condition, randomly generating distribution coding groups with target quantity based on the priority values of all the procedures to be distributed, wherein each procedure to be distributed in the distribution coding groups is executed by one candidate robot; determining a crossover operator, and performing a crossover processing process on the distribution code groups of the target number based on the crossover operator to obtain reference code groups of the target number, wherein the assembly cost of the reference code groups is less than that of the distribution code groups; determining a mutation operator, and executing a mutation processing process on the target number of the distributed coding groups based on the mutation operator to obtain the target number of the alternative coding groups, wherein the assembly cost of the alternative coding groups is less than that of the distributed coding groups; and obtaining an updated distribution coding group based on the distribution coding group, the reference coding group and the alternative coding group, iteratively executing the cross processing process and the variation processing process, changing the probability of the cross processing process and the probability of the variation processing process in each iteration process until the iteration duration exceeds a duration threshold, and outputting the updated distribution coding group obtained by the last iteration.
The above operations can be implemented in the foregoing embodiments, and are not described in detail herein.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. Can be understood and implemented by those skilled in the art without inventive effort.
The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software unit executed by a processor, or in a combination of the two. The software cells may reside in Random Access Memory (RAM), memory, Read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.
Finally, it should be noted that: the above descriptions in the embodiments describe the technical solutions of the present invention in detail, and the principles and embodiments of the present invention are explained herein by using specific examples, and the above descriptions in the embodiments are only used to help understanding the method and the core ideas of the present invention; further, since numerous modifications and changes will readily occur to those skilled in the art, it is not desired to limit the invention to the exact construction and operation illustrated and described, and accordingly, all suitable modifications and equivalents may be resorted to, falling within the scope of the invention.
Claims (10)
1. A balanced distribution method for a mixed flow assembly line under uncertain demands is characterized by comprising the following steps:
acquiring a process set to be distributed, a priority value of each process to be distributed and a candidate robot set;
under the condition that an assembly line distribution constraint condition is met, randomly generating a target number of distribution coding groups based on the priority values of all the procedures to be distributed, wherein each procedure to be distributed in the distribution coding groups is executed by one candidate robot;
determining a crossover operator, and performing a crossover processing process on the target number of allocation code groups based on the crossover operator to obtain a target number of reference code groups, wherein the assembly cost of the reference code groups is less than that of the allocation code groups;
determining a mutation operator, and executing a mutation processing process on the target number of distributed coding groups based on the mutation operator to obtain target number of alternative coding groups, wherein the assembly cost of the alternative coding groups is less than that of the distributed coding groups;
obtaining an updated allocation code group based on the allocation code group, the reference code group and the alternative code group, iteratively executing the cross processing process and the mutation processing process, changing the probability of the cross processing process and the probability of the mutation processing process in each iteration process until the iteration duration exceeds a duration threshold, and outputting the updated allocation code group obtained by the last iteration.
2. The uncertain demand mixed flow assembly line balanced distribution method according to claim 1, wherein the determining a crossover operator comprises:
acquiring a cross processing Q value table, wherein the initial value of the cross processing Q value table is 0;
after each iteration process, based on the formula:
updating the cross-process Q-value table, wherein newQs，aRepresenting the updated Q value corresponding to the current state s and the current action a; qs，aRepresenting the current Q value corresponding to the current state s and the current action a;
representing the maximum Q value in the next state, estimated from the current Q value, Rs，aA reward value representing the taking of action a in the current state s; α and γ represent importance coefficients;
determining the crossover operator from a double-point crossover, a single-point crossover, a sequential crossover, a localized crossover, and a cyclic crossover based on the crossover processing Q-value table.
3. The uncertain demand mixed flow assembly line balanced distribution method according to claim 1, wherein the determining mutation operator comprises:
obtaining a variation processing Q value table, wherein the initial value of the variation processing Q value table is 0;
after each iteration process, based on the formula:
updating the mutation processing Q value table, wherein newqs，aRepresenting the updated Q value corresponding to the current state s and the current action a; q. q.ss，aRepresenting the current Q value corresponding to the current state s and the current action a;
representing the maximum Q value in the next state, estimated from the current Q value, rs，aA reward value representing the taking of action a in the current state s; α and γ represent importance coefficients;
determining the mutation operator from the swapping, the forward insertion, and the backward insertion based on the mutation processing Q-value table.
4. The uncertain demand mixed flow assembly line balanced distribution method according to any of claims 1 to 3, wherein the changing the probability of the cross processing process and the probability of the mutation processing process in each iteration process comprises:
based on the formula:
P＝1-CR；
determining a probability of a crossover process or a mutation process, wherein T represents a current running time and n.n.p represents the duration threshold.
5. The mixed flow assembly line balanced distribution method under uncertain demand according to any of claims 1 to 3, characterized in that, in the case of satisfying the assembly line distribution constraint condition, randomly generating a target number of distribution coding groups based on the priority value of each process to be distributed comprises:
step X, selecting the procedure to be distributed with the maximum priority value from the procedure set to be distributed as a target procedure, and deleting the target procedure from the procedure to be distributed;
step Y, selecting the robot with the earliest assembling time of the target working procedure from the candidate robot set as a target robot, and deleting the target robot from the candidate robot set;
and traversing the step X and the step Y until the process set to be distributed is an empty set, and determining the distribution coding group based on the target process and the corresponding target robot.
6. The uncertain demand mixed flow assembly line balance distribution method according to claim 5, wherein the selecting the robot with the earliest assembling time of the target working procedure from the candidate robot set as the target robot comprises:
and if at least two robots with the earliest target process time are present in the candidate robot set, selecting the robot with the smallest idle time between the target process and the previous process from the at least two robots with the earliest target process time as the target robot.
7. The uncertain demand mixed flow assembly line balance distribution method according to claim 6, wherein the selecting the robot with the earliest assembling time of the target working procedure from the candidate robot set as the target robot further comprises:
and if the number of the robots with the smallest idle time between the target process and the previous process is at least two selected from the robots with the earliest time for assembling the target process, selecting the robot with the smallest label as the target robot.
8. An uncertain demand down mixed flow assembly line balanced distribution device, comprising:
the acquisition module is used for acquiring a process set to be distributed, priority values of all processes to be distributed and a candidate robot set;
the generating module is used for randomly generating a target number of distribution coding groups based on the priority values of all the procedures to be distributed under the condition of meeting the distribution constraint conditions of the assembly line, wherein each procedure to be distributed in the distribution coding groups is executed by one candidate robot;
the cross processing module is used for determining cross operators and executing a cross processing process on the distribution coding groups of the target number based on the cross operators to obtain reference coding groups of the target number, and the assembly cost of the reference coding groups is less than that of the distribution coding groups;
a mutation processing module, configured to determine a mutation operator, and perform a mutation processing process on the target number of allocated coding groups based on the mutation operator to obtain a target number of candidate coding groups, where an assembly cost of the candidate coding groups is less than an assembly cost of the allocated coding groups;
and the output module is used for obtaining an updated distribution code group based on the distribution code group, the reference code group and the alternative code group, iteratively executing the cross processing process and the variation processing process, changing the probability of the cross processing process and the probability of the variation processing process in each iteration process until the iteration duration exceeds a duration threshold, and outputting the updated distribution code group obtained by the last iteration.
9. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor when executing the program implements the steps of the uncertain demand downmixed assembly line balanced distribution method of any of claims 1 to 7.
10. A non-transitory computer readable storage medium having stored thereon a computer program, wherein the computer program, when executed by a processor, implements the steps of the uncertain demand downmix assembly line balanced distribution method as recited in any of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210060798.3A CN114415615B (en) | 2022-01-19 | 2022-01-19 | Mixed flow assembly line balanced distribution method and device under uncertain demand |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210060798.3A CN114415615B (en) | 2022-01-19 | 2022-01-19 | Mixed flow assembly line balanced distribution method and device under uncertain demand |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114415615A true CN114415615A (en) | 2022-04-29 |
| CN114415615B CN114415615B (en) | 2024-05-28 |
Family
ID=81274436
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210060798.3A Active CN114415615B (en) | 2022-01-19 | 2022-01-19 | Mixed flow assembly line balanced distribution method and device under uncertain demand |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114415615B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115609593A (en) * | 2022-12-02 | 2023-01-17 | 天津大学 | Multi-type robot man-machine cooperative assembly scheduling method and device |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102323952A (en) * | 2011-09-09 | 2012-01-18 | 河海大学常州校区 | Reconfigurable assembly line sequencing method based on improved genetic algorithm |
| CN102354311A (en) * | 2011-09-09 | 2012-02-15 | 河海大学常州校区 | Balance method for reconfigurable assembly line customized on large scale |
| CN106022474A (en) * | 2016-06-01 | 2016-10-12 | 河海大学常州校区 | Reconstructible assembly line balancing optimization method |
| CN107316107A (en) * | 2017-06-15 | 2017-11-03 | 南京理工大学 | A kind of tricot machine assembly line balancing method towards multiple-objection optimization |
| GB201717125D0 (en) * | 2016-11-28 | 2017-11-29 | National Univ Of Defense Technology | Differential evolution method oriented to agile satellite multi-target task planning |
| CN111353604A (en) * | 2018-12-24 | 2020-06-30 | 南京理工大学 | Flexible assembly multi-objective dynamic optimization method |
-
2022
- 2022-01-19 CN CN202210060798.3A patent/CN114415615B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102323952A (en) * | 2011-09-09 | 2012-01-18 | 河海大学常州校区 | Reconfigurable assembly line sequencing method based on improved genetic algorithm |
| CN102354311A (en) * | 2011-09-09 | 2012-02-15 | 河海大学常州校区 | Balance method for reconfigurable assembly line customized on large scale |
| CN106022474A (en) * | 2016-06-01 | 2016-10-12 | 河海大学常州校区 | Reconstructible assembly line balancing optimization method |
| GB201717125D0 (en) * | 2016-11-28 | 2017-11-29 | National Univ Of Defense Technology | Differential evolution method oriented to agile satellite multi-target task planning |
| CN107316107A (en) * | 2017-06-15 | 2017-11-03 | 南京理工大学 | A kind of tricot machine assembly line balancing method towards multiple-objection optimization |
| CN111353604A (en) * | 2018-12-24 | 2020-06-30 | 南京理工大学 | Flexible assembly multi-objective dynamic optimization method |
Non-Patent Citations (1)
| Title |
|---|
| 赵联鹏，唐秋等: "预防维护下装配线平衡的多目标重启 变邻域搜索算法", 中国机械工程, vol. 32, no. 21, 30 November 2021 (2021-11-30), pages 2606 - 2616 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115609593A (en) * | 2022-12-02 | 2023-01-17 | 天津大学 | Multi-type robot man-machine cooperative assembly scheduling method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114415615B (en) | 2024-05-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106647262B (en) | Differential evolution method for agile satellite multi-target task planning | |
| CN104268722B (en) | Dynamic flexible job-shop scheduling method based on multi-objective Evolutionary Algorithm | |
| US7668788B2 (en) | Resource assignment optimization using direct encoding and genetic algorithms | |
| CN109472362B (en) | AGV dynamic scheduling method and device based on variable task window | |
| CN109271320B (en) | Higher-level multi-target test case priority ordering method | |
| CN108460463B (en) | High-end equipment assembly line production scheduling method based on improved genetic algorithm | |
| CN111832725B (en) | Multi-robot multi-task allocation method and device based on improved genetic algorithm | |
| CN113128839B (en) | High-end equipment distributed manufacturing and multi-mode transportation oriented cooperative scheduling method | |
| CN115793657B (en) | Distribution robot path planning method based on temporal logic control strategy | |
| CN114415615B (en) | Mixed flow assembly line balanced distribution method and device under uncertain demand | |
| CN114881504B (en) | Electronic precision part full-automatic flexible production line scheduling method based on order selection | |
| CN111369000A (en) | High-dimensional multi-target evolution method based on decomposition | |
| CN117075545A (en) | Multi-target flexible job shop scheduling method, electronic equipment and medium | |
| CN115617690A (en) | Parallel test task scheduling method based on improved adaptive genetic algorithm | |
| CN116880424A (en) | Multi-robot scheduling method and device based on multi-objective optimization | |
| CN109214695B (en) | High-end equipment research, development and manufacturing cooperative scheduling method and system based on improved EDA | |
| CN109086936B (en) | Production system resource allocation method, device and equipment for intelligent workshop | |
| CN114021934A (en) | Method for solving workshop energy-saving scheduling problem based on improved SPEA2 | |
| CN113485278B (en) | Flexible job shop scheduling multi-target distribution estimation method for optimizing two production indexes | |
| CN115933568A (en) | Multi-target distributed hybrid flow shop scheduling method | |
| CN114969148A (en) | System access amount prediction method, medium and equipment based on deep learning | |
| CN114912647A (en) | Apparatus, method and machine-readable storage medium for decision making | |
| CN116795054B (en) | Intermediate product scheduling method in discrete manufacturing mode | |
| Drugan | Stochastic pareto local search for many objective quadratic assignment problem instances | |
| CN114545884B (en) | Equivalent parallel machine dynamic intelligent scheduling method based on enhanced topological neural evolution |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |