WO2021142915A1 - 多目标流水车间调度方法、装置、计算机设备及存储介质 - Google Patents

多目标流水车间调度方法、装置、计算机设备及存储介质 Download PDF

Info

Publication number
WO2021142915A1
WO2021142915A1 PCT/CN2020/079877 CN2020079877W WO2021142915A1 WO 2021142915 A1 WO2021142915 A1 WO 2021142915A1 CN 2020079877 W CN2020079877 W CN 2020079877W WO 2021142915 A1 WO2021142915 A1 WO 2021142915A1
Authority
WO
WIPO (PCT)
Prior art keywords
individual
target
shop scheduling
objective
population
Prior art date
Application number
PCT/CN2020/079877
Other languages
English (en)
French (fr)
Inventor
郑峻浩
刘松柏
林秋镇
陈剑勇
Original Assignee
深圳大学
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 深圳大学 filed Critical 深圳大学
Publication of WO2021142915A1 publication Critical patent/WO2021142915A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Energy or water supply

Definitions

  • This application relates to the technical field of workshop production scheduling control, and in particular to a multi-object flow shop scheduling method, device, computer equipment and storage medium.
  • the sales department needs products to be produced on time to complete orders; the manufacturing department needs to reduce costs and increase machine utilization; the quality inspection department needs to have a high product qualification rate and higher quality than competitors, etc. That is, the scheduling system needs to optimize multiple goals at the same time.
  • the weights assigned to each target are combined into a single target solution.
  • the above methods are highly subjective, the solution scheme has certain pertinence, and the weights cannot be flexibly changed according to the conditions of the workshop.
  • the existing scheduling method based on multi-objective optimization has a huge search space, low efficiency in the solution process, and cannot maintain the diversity of feasible solutions, and it is difficult to obtain a globally optimal feasible solution set for scheduling decision-making.
  • the embodiments of the present application provide a multi-object flow shop scheduling method, device, computer equipment, and storage medium, aiming to solve the problem of using the traditional linear weighting method in the multi-object optimization-based shop scheduling method in the prior art. Due to the huge search space, In the solution process, the efficiency is low, and the diversity of feasible solutions cannot be maintained, and it is difficult to obtain the global optimal feasible solution set.
  • an embodiment of the present application provides a multi-object flow shop scheduling method, which includes:
  • the input data and constraint conditions corresponding to the shop scheduling request are acquired; wherein the input data corresponding to the shop scheduling request includes the number of workpieces, the number of processing procedures, and the number of machines;
  • an embodiment of the present application provides a multi-object flow shop scheduling device, which includes a unit for executing the multi-object flow shop scheduling method described in the first aspect.
  • an embodiment of the present application provides a computer device, which includes a memory, a processor, and a computer program stored on the memory and running on the processor, and the processor executes the computer
  • the program realizes the multi-object flow shop scheduling method described in the first aspect above.
  • an embodiment of the present application also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor executes the above-mentioned first On the one hand, the multi-object flow shop scheduling method.
  • FIG. 1 is a schematic diagram of an application scenario of a multi-object flow shop scheduling method provided by an embodiment of the application
  • FIG. 2 is a schematic flowchart of a multi-object flow shop scheduling method provided by an embodiment of the application
  • FIG. 3 is a schematic diagram of a sub-process of a multi-object flow shop scheduling method provided by an embodiment of the application;
  • FIG. 4 is a schematic diagram of another sub-process of the multi-object flow shop scheduling method provided by an embodiment of the application.
  • FIG. 5 is a schematic diagram of another sub-process of the multi-object flow shop scheduling method provided by an embodiment of the application.
  • Fig. 6 is a schematic block diagram of a multi-object flow shop scheduling device provided by an embodiment of the application.
  • FIG. 7 is a schematic block diagram of a computer device provided by an embodiment of the application.
  • Figure 1 is a schematic diagram of the application scenario of the multi-object flow shop scheduling method provided by an embodiment of the application
  • Figure 2 is a schematic flow chart of the multi-object flow shop scheduling method provided by an embodiment of the application.
  • the flow shop scheduling method is applied to the server, and the method is executed by the application software installed in the server.
  • the method includes steps S110 to S140.
  • the client can be understood as a user terminal.
  • the user terminal can be a smart phone, tablet computer, notebook computer, desktop computer, personal digital assistant, wearable device and other electronic devices with communication functions.
  • the user terminal sends workshop dispatch Request to the server.
  • the second is the server.
  • the server receives the workshop scheduling request sent by the client, and according to the input data and constraint conditions corresponding to the workshop scheduling request, and calls the pre-stored multi-objective workshop scheduling optimization model to solve the super-multi-objective evolutionary solution.
  • the optimal solution set After obtaining the optimal solution set in the server, it is sent to the client.
  • the server detects whether the workshop scheduling request sent by the client is received.
  • the subsequent step S120 is executed.
  • step S110 is executed again after waiting for the preset delay time.
  • the input data and constraint conditions corresponding to the shop scheduling request are obtained; wherein, the input data corresponding to the shop scheduling request includes the number of workpieces, the number of processing procedures, and the machine. number.
  • the server receives the workshop scheduling request sent by the client, it obtains the input data and constraint conditions corresponding to the workshop scheduling request. Since the multi-objective shop scheduling optimization model has been pre-stored in the server, it can be solved subsequently according to the input data and constraint conditions, so as to obtain the optimal solution set.
  • the multi-objective shop scheduling optimization model stored in the server is a multi-objective optimization model.
  • the optimization goals are as far as possible to achieve a satisfactory flow shop scheduling scheme.
  • the multi-objective shop scheduling optimization model includes five optimization objective functions, which are respectively denoted as:
  • n is the number of workpieces
  • C i represents the completion time of the i-th workpiece
  • n M represents the number of processing machines
  • p ijk is the processing time required for the j-th process of the k-th machine to process the i-th workpiece
  • x ijk is the j-th process used to judge the i-th workpiece
  • D i is the delivery time of the i-th workpiece
  • It is the value of processing cost per unit time of the k-th machine.
  • the multi-objective shop scheduling optimization model called after the received shop scheduling request can be expressed as m is the number of targets optimized for shop scheduling, the value of m is 5, and ⁇ represents the set of shop scheduling plans.
  • the 5 workshop scheduling optimization objectives are made (5 workshop scheduling optimization objectives are minf 1 (x), minf 2 (x), minf 3) (x), minf 4 (x), minf 5 (x)) reach the minimum at the same time.
  • a candidate solution x of the multi-objective shop scheduling optimization model it refers to satisfying the above 5 optimization objective functions (that is, satisfying minf 1 (x), minf 2 (x), minf 3 (x), minf 4 (x) ,
  • a scheduling plan of minf 5 (x)), X represents a set containing multiple candidate solutions, and the optimal solution set X composed of multiple candidate solutions is optimal .
  • this model is a model of high dimensional optimization, the objective function in conjunction with the five constraint conditions i.e. the input data and solving for X to give the optimal solution set to the optimal solution set of multi-objective optimization model shop scheduling , Which can satisfy the proposed optimization goals and constraints to the greatest extent.
  • the constraints corresponding to the workshop scheduling request are: all machines are available at the initial moment; the same machine can only process one workpiece at the same time; one workpiece can only be processed on one machine at the same time ; There is a sequence between the processing procedures of the workpiece; during the processing, the processing priority of each workpiece is the same.
  • the step S130 includes:
  • an initial multi-target population is randomly generated under the restriction of constraint conditions.
  • the initial multi-target population is the first-generation multi-target population.
  • the current iteration algebra does not reach the maximum iteration algebra, first use the initial multi-target population as the initial population to simulate binary crossover and polynomial mutation, that is, randomly select two individuals from the initial multi-target population to perform binary crossover in turn, Until N cross-processed new individuals are generated, the N cross-processed new individuals are subjected to polynomial mutation, and the new individuals after polynomial mutation form a subpopulation.
  • N new crossover processed individuals are obtained.
  • the process of randomly selecting two individuals for binary crossover is also an iterative process. Until the number of new individuals reaches the population size N, the above-mentioned multiple binary crossover processing process is stopped.
  • binary crossover and polynomial mutation are both conventional processing procedures, and will not be repeated here.
  • the initial multi-target population and the sub-population are combined to obtain a mixed population, and the total number of individuals included in the mixed population is twice the population size N.
  • non-dominated sorting can be performed on the individuals in the mixed population, thereby obtaining non-dominated solution sets and multi-layer solution sets.
  • the non-dominated solution set corresponding to the mixed population can be obtained through a non-dominated solution (also called Pareto solution) acquisition method.
  • a non-dominated solution also called Pareto solution
  • the definition of Pareto solution is to assume that for any two solutions S1 and S2, S1 is better than or the same as S2 for all targets, and there is at least one target, and the corresponding target value of S1 on this target is better than S2.
  • the corresponding target value on the target is called S1 dominates S2.
  • S1 is called the non-dominated solution (undominated solution), also called the Pareto solution (ie Pareto solution).
  • the obtained non-dominated solution set is denoted as Q 1 .
  • the multi-layer solution set is obtained.
  • the multi-layer solution set includes multiple solution set subsets and is respectively denoted as Q 2 to Q L , where Q 1 to The union of Q L is the mixed population, and the intersection of any two sets from Q 1 to Q L is an empty set, Q 1 ⁇ Q 2 ⁇ Q 3 ⁇ « ⁇ Q L ; where " ⁇ " indicates a dominance relationship , Q i ⁇ Q j means that Q j is dominated by the solution in Q i , and the relationship is transitive.
  • Q 1 ⁇ Q 2 means that for f 1 (x) to f 5 (x), each of Q 2 Each solution is dominated by at least one solution in Q 1 , and the relationship is transitive, that is, each solution in Q 3 is dominated by at least one solution in Q 1 or Q 2 , and the others can be deduced by analogy.
  • the position of the adaptive reference point is determined according to the position of each non-dominated individual in the non-dominated solution set in the normalized target space.
  • the individual vector corresponding to each individual in the mixed population can be determined, and the vector angle between each volume vector can be calculated, and then the mixed vector can be combined according to the vector angle by the clustering algorithm.
  • step S1305 includes:
  • the mixed population in order to obtain the position of the adaptive reference point, may be divided into multiple individual strata according to the Pareto dominance relationship, that is, respectively denoted as Q 1 to Q L , Individuals on the first layer form a non-dominated solution set Q 1 , and then select multiple individual stratifications from Q 1 to Q L in sequence, until the total number of individuals in the multiple individual stratification exceeds the population size N, to Make up a target set.
  • the mixed population may be divided into multiple individual strata according to the Pareto dominance relationship, that is, respectively denoted as Q 1 to Q L .
  • Individuals on the first layer form a non-dominated solution set Q 1 , and then select multiple individual stratifications from Q 1 to Q L in sequence, until the total number of individuals in the multiple individual stratification exceeds the population size N, to Make up a target set.
  • the individuals in the target set normalize the individuals in the target set. Specifically, first obtain the minimum individual and the maximum individual in the target set; wherein the minimum individual is input to the multi-objective shop scheduling optimization model and the target value obtained is the target corresponding to each individual in the target set.
  • the minimum target value among the values, and the target value obtained by inputting the maximum individual to the multi-target shop scheduling optimization model is the maximum target value among the target values corresponding to each individual in the target set.
  • f i (x) represents the value of the individual x in the target set on the ith shop scheduling optimization target
  • f i '( x) represents the normalized value of the individual x in the target set on the ith shop scheduling optimization objective
  • f i min represents the value of the minimum individual in the target set on the ith shop scheduling optimization objective
  • f i max represents The value of the maximum individual of the target set on the i-th shop scheduling optimization target.
  • each non-dominated individual in the normalized non-dominated solution set is obtained and denoted as B 1 to B M respectively .
  • the specific function for obtaining the hyperplane distance D i corresponding to the non-dominated individual B i to the target hyperplane is That is, the vertical distance from the non-dominated individual B i to the target hyperplane, and the value of m is 5.
  • calculating the hypersurface fitness fit p first calculate the average Minkowski distance based on all individuals in the normalized non-dominated solution set, and then calculate the difference between the average Minkowski distance and 1.0 as a measure of the normalization The fitness of the individuals in the non-dominated solution set distributed on the corresponding hypersurface. The closer fit p is to 0, the higher the fitness of the individual distribution to the corresponding p value, that is, the more similar it is to the hypersurface.
  • the distribution of the optimal solution set of the current multi-objective shop scheduling optimization problem can be roughly predicted, and the algorithm strategy can be adjusted according to the distribution. , Enhance algorithm performance.
  • the initial position of the adaptive reference point is first determined according to the value of mpd, Then according to the value of fit p , it is determined that the adaptive reference point moves a certain distance in the diagonal direction based on the initial position, and finally the adaptive reference point r* (wherein, Represents the coordinate of the i-th dimension of the adaptive reference point).
  • m is the target number, which is equal to 5.
  • R is the current hypersurface
  • the radius when ru is the center of the circle.
  • the value of m is 5
  • ⁇ is the preset step size parameter used to determine the step size of the adaptive parameter movement.
  • the selection of the position of the adaptive reference point used in this application depends on the relative positional relationship between the initial position of the adaptive reference point and the distribution of the optimal solution and the hypersurface with p values of 0.5, 1.0, 2.0.
  • the adaptive reference point method is adopted, so that the angle-based clustering method can also have excellent performance in solving the problem of convex PF shape.
  • the multi-objective shop scheduling problem to be solved is transformed into a multi-objective optimization algorithm based on angle clustering that is good at solving
  • the optimal solution distribution is a concave hypersurface, that is, the PF (Pareto-Optimal Front) shape is concave.
  • step S1306 includes:
  • the formula Calculate the angle between the vector formed by the reference point and the cluster center, and use it to measure the similarity between clusters, where c i and c j are the centers of the i-th and j-th clusters, respectively, r * Is an adaptive reference point.
  • the formula calculates the cluster centers, where C k represents the k-th cluster in the clustering result,
  • the angle-based clustering method is adopted to ensure that the solutions selected in each generation have good diversity.
  • the purpose is to select individuals with better performance to form the current multi-target population, and use the current multi-target population as the initial multi-target population.
  • step S1307 includes:
  • the angle is composed of the smallest of the vector angles between each individual and the first target individual, the second target individual, the third target individual, and the fourth target individual.
  • the third individual included angle set corresponding to the clustering result after the third screening the individual corresponding to the largest vector included angle in the third individual included angle set is obtained as the fifth target individual, and the cluster corresponding to the fifth target individual Clusters are removed from the clustering result after the third screening to obtain the clustering result after the fourth screening;
  • the environment selection method when used to obtain multiple target individuals in the clustering result to form the current multi-target population, in order to ensure the diversity of the population, on the basis of clustering, first select the In the clustering result, the two individuals with the largest vector angle (denoted as the first target individual and the second target individual) are added to the current multi-target population and their corresponding clusters are deleted from the clustering results Get the clustering results after the first screening.
  • the current multi-target population only includes the first target individual and the second target individual
  • find out The individual with the largest angle from the current multi-target population (recorded as the third target individual) is added to the current multi-target population and its corresponding cluster is deleted from the cluster results after the first screening to obtain the second screening
  • repeat the above process of deleting and selecting the third target individual until the number of individuals in the current multi-target population reaches 5, which are respectively recorded as the first target individual, the second target individual, the third target individual, the fourth target individual, The fifth target individual.
  • the specific encoding method of each optimal solution in the optimal solution set (that is, the method that is finally displayed to the user for viewing) is a user-defined encoding format and saved in the server. This time, the specific encoding is not limited. Way.
  • the current multi-target population is obtained in step S1307, it indicates the end of this round of iterative process.
  • the current multi-target population is used as the initial multi-target population, so that when the next round of iterative process starts, the previous round of iteration ends.
  • the current multi-target population at time is taken as the new initial multi-target population at the beginning of the next-generation iteration process.
  • the client after obtaining the optimal solution set in the server, it can be sent to the client. Therefore, the client can determine the scheduling mode of the flow shop according to the optimal solution set.
  • This method achieves rapid solution under the premise of huge search space in the process of super-multi-objective evolutionary solution, and maintains the diversity of feasible solutions.
  • the embodiment of the present application also provides a multi-target flow shop scheduling device, which is used to execute any embodiment of the foregoing multi-target flow shop scheduling method.
  • FIG. 6, is a schematic block diagram of a multi-object flow shop scheduling device provided by an embodiment of the present application.
  • the multi-object flow shop scheduling device 100 may be configured in a server.
  • the multi-object flow shop scheduling apparatus 100 includes a shop scheduling request detection unit 110, an input data condition acquisition unit 120, an optimal solution set solving unit 130, and an optimal solution set sending unit 140.
  • the workshop scheduling request detection unit 110 is used to determine whether the workshop scheduling request sent by the client is received.
  • the input data condition obtaining unit 120 is configured to obtain input data and constraint conditions corresponding to the shop scheduling request if the shop scheduling request sent by the client is received; wherein, the input data corresponding to the shop scheduling request includes the number of workpieces , The number of processing steps, and the number of machines.
  • the optimal solution set solving unit 130 is configured to call a pre-stored multi-objective shop scheduling optimization model, use the input data as the input of the multi-objective shop scheduling optimization model, and compare all the parameters according to the constraints and the input data.
  • the multi-objective shop scheduling optimization model is used to solve super-multi-objective evolution and obtain the optimal solution set.
  • the optimal solution set sending unit 140 is configured to send the optimal solution set to the client.
  • the optimal solution set solving unit 130 includes:
  • the initial multi-target population generating unit is used to randomly generate an initial multi-target population according to the constraint conditions; wherein the initial multi-target population includes a plurality of individuals, and each individual corresponds to a workshop of the multi-target shop scheduling optimization model Scheduling the output solution, and the total number of multiple individuals in the initial multi-target population is recorded as the population size N;
  • the first judging unit of the current iteration algebra is used to obtain the current iteration algebra, and judge whether the current iteration algebra reaches the preset maximum iteration algebra;
  • the individual crossover mutation unit is used to perform simulated binary crossover and polynomial mutation on the initial multi-target population if the current iterative algebra does not reach the maximum iterative algebra, to obtain the total number of individuals that are the same as the initial multi-target population Subpopulation
  • a population merging unit for merging the initial multi-target population and the sub-population to obtain a mixed population
  • the adaptive reference point acquisition unit is used to acquire the non-dominated solution set and multi-layer solution set in the mixed population, and the adaptive reference point corresponding to the non-dominated solution set; wherein, the non-dominated solution set record Is Q 1 , the multi-layer solution set includes multiple solution set subsets and are respectively denoted as Q 2 to Q L , where the union of Q 1 to Q L is the mixed population, and any two of Q 1 to Q L The intersection of the two sets is the empty set, Q 1 ⁇ Q 2 ⁇ Q 3 ⁇ « ⁇ Q L ;
  • the angle clustering unit is used to obtain the individual vector corresponding to the target space point corresponding to each individual in the mixed population and the adaptive reference point, and to calculate the individual vector corresponding to each individual in the mixed population Perform clustering according to the vector angle similarity and the population size N, and obtain a clustering result including N clusters;
  • the initial multi-target population update unit is used to obtain multiple target individuals from the clustering result through an environmental selection method to form a current multi-target population, and use the current multi-target population as the initial multi-target population;
  • the current iteration algebra self-increment unit is used to increase the current iteration algebra by one as the current iteration algebra, and return the current iteration algebra to the first judgment unit to perform the step of judging whether the current iteration algebra reaches the preset maximum iteration algebra;
  • the optimal solution set output unit is configured to output the current multi-target population as the optimal solution set if the current iteration algebra reaches the maximum iteration algebra.
  • the adaptive reference point acquiring unit includes:
  • the target set acquisition unit is used to sequentially merge multiple solution set subsets in the non-dominated solution set and the multi-layer solution set to obtain multiple sets until the total number of individuals exceeds the population size N to form a target gather;
  • the endpoint individual acquiring unit is used to acquire the minimum individual and the maximum individual in the target set; wherein the minimum individual is input to the multi-objective shop scheduling optimization model and the target value obtained is each individual in the target set
  • the minimum target value among the corresponding target values, and the target value obtained by inputting the maximum individual to the multi-target shop scheduling optimization model is the maximum target value among the target values corresponding to each individual in the target set;
  • the normalization processing unit is configured to perform normalization processing on each individual in the target set according to the minimum individual and the maximum individual to obtain a normalized target set; wherein, the normalized target The set of normalized individuals in the set corresponding to the non-dominated solution set is recorded as the normalized non-dominated solution set;
  • the hypersurface fitness acquisition unit is used to acquire the hypersurface fitness corresponding to each normalized non-dominated individual in the normalized non-dominated solution set and the target hypersurface set; wherein the hypersurface fitness is represented by fit p ,
  • the adaptive reference point positioning unit is configured to call a preset adaptive reference point acquisition strategy, and acquire an adaptive reference point corresponding to the non-dominated solution set according to the adaptive reference point acquisition strategy and the hypersurface fitness ;in,
  • the value of m is 5.
  • the angle clustering unit includes:
  • An initial clustering result acquiring unit configured to acquire each normalized target individual in the normalized target set, and divide each normalized target individual into an initial cluster cluster to form an initial clustering result
  • An initial cluster center acquiring unit configured to acquire the initial cluster center corresponding to each initial cluster cluster in the initial clustering result
  • a current individual vector set acquiring unit configured to acquire current individual vectors corresponding to each initial cluster center corresponding to the initial clustering result and the adaptive reference point to form a current individual vector set
  • the cluster selection unit is used to merge the two current individual vectors whose vector included angle is the current minimum included angle value in the current individual vector set into one cluster cluster, and the two current individual vectors corresponding to the cluster cluster Remove from the current individual vector set to update to obtain the current individual vector set, and return to execute the merging of the two current individual vectors whose vector included angle is the current minimum included angle value in the current individual vector set into one cluster Clusters until the number of clusters corresponding to the current individual vector set is equal to the population size N, and a clustering result including N clusters is obtained.
  • the initial multi-target population update unit includes:
  • the first individual screening unit is used to obtain the two target individuals with the largest vector included angle from the clustering result, and move the cluster clusters corresponding to the two target individuals with the largest vector included angle from the clustering result Divide to obtain the clustering result after the first screening; wherein, the two target individuals with the largest vector included angles in the clustering result are respectively recorded as the first target individual and the second target individual;
  • the second individual screening unit is used to obtain the vector angles corresponding to the first target individual and the second target individual of each individual in the clustering result after the first screening, so that each individual The smallest of the vector included angles corresponding to the first target individual and the second target individual forms a first individual included angle set corresponding to the clustering result after the first screening, and the first individual included angle set is obtained.
  • the individual corresponding to the largest vector included angle in a set of body included angles is taken as the third target individual, and the cluster cluster corresponding to the third target individual is removed from the clustering results after the first screening to obtain the second Clustering results after the second screening;
  • the third individual screening unit is used to obtain the vector corresponding to the first target individual, the second target individual, and the third target individual for each individual in the clustering result after the second screening
  • the included angle which is composed of the smallest of the vector included angles between each individual and the first target individual, the second target individual, and the third target individual, and the cluster after the second screening
  • the second individual included angle set corresponding to the second individual included angle set is obtained, the individual corresponding to the largest vector included angle in the second individual included angle set is obtained as the fourth target individual, and the cluster corresponding to the fourth target individual is selected from the second time Remove from the clustering result after screening to obtain the clustering result after the third screening;
  • the fourth individual screening unit is used to obtain each individual in the clustering result after the third screening is related to the first target individual, the second target individual, and the third target individual, and
  • the angle of the vector corresponding to the fourth target individual is the vector corresponding to the first target individual, the second target individual, the third target individual, and the fourth target individual for each individual.
  • the smallest of the included angles forms a third individual included angle set corresponding to the clustering result after the third screening, and the individual corresponding to the largest vector included angle in the third individual included angle set is obtained as the fifth target individual, Removing the cluster cluster corresponding to the fifth target individual from the clustering result after the third screening to obtain the clustering result after the fourth screening;
  • the fifth individual screening unit is used to call the preset individual convergence function to obtain the individual with the smallest convergence in each cluster in the clustering result after the fourth screening, and to compare it with the first target individual and the second target individual.
  • the target individual, the third target individual, the fourth target individual, and the fifth target individual constitute the current multi-target population.
  • the individual cross mutation unit is also used for:
  • the device realizes rapid solution under the premise of huge search space in the process of super-multi-objective evolutionary solution, and maintains the diversity of feasible solutions.
  • the above-mentioned multi-object flow shop scheduling device can be implemented in the form of a computer program, and the computer program can be run on a computer device as shown in FIG. 7.
  • FIG. 7 is a schematic block diagram of a computer device according to an embodiment of the present application.
  • the computer device 500 is a server, and the server may be an independent server or a server cluster composed of multiple servers.
  • the computer device 500 includes a processor 502, a memory, and a network interface 505 connected through a system bus 501, where the memory may include a non-volatile storage medium 503 and an internal memory 504.
  • the non-volatile storage medium 503 can store an operating system 5031 and a computer program 5032.
  • the processor 502 can execute the multi-object flow shop scheduling method.
  • the processor 502 is used to provide calculation and control capabilities, and support the operation of the entire computer device 500.
  • the internal memory 504 provides an environment for the operation of the computer program 5032 in the non-volatile storage medium 503.
  • the processor 502 can execute the multi-object flow shop scheduling method.
  • the network interface 505 is used for network communication, such as providing data information transmission.
  • the structure shown in FIG. 7 is only a block diagram of part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device 500 to which the solution of the present application is applied.
  • the specific computer device 500 may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.
  • the processor 502 is configured to run a computer program 5032 stored in a memory to implement the multi-object flow shop scheduling method disclosed in the embodiment of the present application.
  • the embodiment of the computer device shown in FIG. 7 does not constitute a limitation on the specific configuration of the computer device.
  • the computer device may include more or less components than those shown in the figure. Or some parts are combined, or different parts are arranged.
  • the computer device may only include a memory and a processor. In such an embodiment, the structure and function of the memory and the processor are consistent with the embodiment shown in FIG. 7 and will not be repeated here.
  • the processor 502 may be a central processing unit (Central Processing Unit, CPU), and the processor 502 may also be other general-purpose processors, digital signal processors (Digital Signal Processors, DSPs), and special purpose processors.
  • Integrated circuit Application Specific Integrated Circuit, ASIC
  • off-the-shelf programmable gate array Field-Programmable Gate Array, FPGA
  • the general-purpose processor may be a microprocessor or the processor may also be any conventional processor.
  • a computer-readable storage medium may be a non-volatile computer-readable storage medium.
  • the computer-readable storage medium stores a computer program, where the computer program is executed by a processor to implement the multi-object flow shop scheduling method disclosed in the embodiments of the present application.
  • the storage medium is a physical, non-transitory storage medium, such as a U disk, a mobile hard disk, a read-only memory (Read-Only Memory, ROM), a magnetic disk, or an optical disk, etc., which can store program codes. .

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Economics (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Strategic Management (AREA)
  • General Business, Economics & Management (AREA)
  • Tourism & Hospitality (AREA)
  • Marketing (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Development Economics (AREA)
  • Game Theory and Decision Science (AREA)
  • Health & Medical Sciences (AREA)
  • Quality & Reliability (AREA)
  • Operations Research (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Educational Administration (AREA)
  • Algebra (AREA)
  • Mathematical Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • Pure & Applied Mathematics (AREA)
  • Databases & Information Systems (AREA)
  • Computational Mathematics (AREA)
  • Mathematical Optimization (AREA)
  • Software Systems (AREA)
  • Public Health (AREA)
  • Water Supply & Treatment (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

一种多目标流水车间调度方法、装置(100)、计算机设备(500)及存储介质。调度方法包括：判断是否接收到客户端发送的车间调度请求（S110）；若接收到客户端发送的车间调度请求，获取与车间调度请求对应的输入数据和约束条件（S120）；其中，与车间调度请求对应的输入数据包括工件数、加工工序数和机器数；调用预先存储的多目标车间调度优化模型，以输入数据为多目标车间调度优化模型的输入，根据约束条件和输入数据对多目标车间调度优化模型进行超多目标的进化求解，得到最优解集（S130）；将最优解集发送至客户端（S140）。

Description

多目标流水车间调度方法、装置、计算机设备及存储介质
本申请要求于2020年1月15日提交中国专利局、申请号为202010041521.7、申请名称为“多目标流水车间调度方法、装置、计算机设备及存储介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及车间生产调度控制技术领域，尤其涉及一种多目标流水车间调度方法、装置、计算机设备及存储介质。
背景技术
在企业实际生产的环境下，企业中的不同部门对调度方法产生的决策提出了不同的要求。例如，销售部门需要产品按时生产以完成订单；制造部门需要降低成本，提高机器利用率；质检部门需要产品合格率高并且质量高于竞争对手等，即调度系统需要同时优化多个目标。
目前，调度系统中采用传统的线性加权法时，是将各个目标赋予权值合并为单目标求解。上述方法主观性较强，求解方案有一定针对性，权值无法根据车间状况灵活改变。而现有的基于多目标优化的调度方法由于搜索空间巨大，在求解过程中效率低下，并且无法保持可行解的多样性，难以得到全局最优的可行解集用于调度决策。
发明内容
本申请实施例提供了一种多目标流水车间调度方法、装置、计算机设备及存储介质，旨在解决现有技术中基于多目标优化的车间调度方法采用传统的线性加权法，由于搜索空间巨大，在求解过程中效率低下，并且无法保持可行解的多样性，难以得到全局最优的可行解集的问题。
第一方面，本申请实施例提供了一种多目标流水车间调度方法，其包括：
判断是否接收到客户端发送的车间调度请求；
若接收到客户端发送的车间调度请求，获取与所述车间调度请求对应的输入数据和约束条件；其中，与所述车间调度请求对应的输入数据包括工件数、加工工序数、和机器数；
调用预先存储的多目标车间调度优化模型，以所述输入数据为所述多目标车间调度优化模型的输入，根据所述约束条件和所述输入数据对所述多目标车间调度优化模型进行超多目标的进化求解，得到最优解集；以及
将所述最优解集发送至客户端。
第二方面，本申请实施例提供了一种多目标流水车间调度装置，其包括用于执行上述第一方面所述的多目标流水车间调度方法的单元。
第三方面，本申请实施例又提供了一种计算机设备，其包括存储器、处理 器及存储在所述存储器上并可在所述处理器上运行的计算机程序，所述处理器执行所述计算机程序时实现上述第一方面所述的多目标流水车间调度方法。
第四方面，本申请实施例还提供了一种计算机可读存储介质，其中所述计算机可读存储介质存储有计算机程序，所述计算机程序当被处理器执行时使所述处理器执行上述第一方面所述的多目标流水车间调度方法。
附图说明
为了更清楚地说明本申请实施例技术方案，下面将对实施例描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为本申请实施例提供的多目标流水车间调度方法的应用场景示意图；
图2为本申请实施例提供的多目标流水车间调度方法的流程示意图；
图3为本申请实施例提供的多目标流水车间调度方法的子流程示意图；
图4为本申请实施例提供的多目标流水车间调度方法的另一子流程示意图；
图5为本申请实施例提供的多目标流水车间调度方法的另一子流程示意图；
图6为本申请实施例提供的多目标流水车间调度装置的示意性框图；
图7为本申请实施例提供的计算机设备的示意性框图。
具体实施方式
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
应当理解，当在本说明书和所附权利要求书中使用时，术语“包括”和“包含”指示所描述特征、整体、步骤、操作、元素和/或组件的存在，但并不排除一个或多个其它特征、整体、步骤、操作、元素、组件和/或其集合的存在或添加。
还应当理解，在此本申请说明书中所使用的术语仅仅是出于描述特定实施例的目的而并不意在限制本申请。如在本申请说明书和所附权利要求书中所使用的那样，除非上下文清楚地指明其它情况，否则单数形式的“一”、“一个”及“该”意在包括复数形式。
还应当进一步理解，在本申请说明书和所附权利要求书中使用的术语“和/或”是指相关联列出的项中的一个或多个的任何组合以及所有可能组合，并且包括这些组合。
请参阅图1和图2，图1为本申请实施例提供的多目标流水车间调度方法的应用场景示意图；图2为本申请实施例提供的多目标流水车间调度方法的流程示意图，该多目标流水车间调度方法应用于服务器中，该方法通过安装于服务器中的应用软件进行执行。
如图2所示，该方法包括步骤S110～S140。
S110、判断是否接收到客户端发送的车间调度请求。
为了更清楚的理解本申请的技术方案，下面对所涉及到的终端进行介绍。本申请是在服务器的角度描述技术方案。
第一是客户端，客户端可以理解为用户终端，用户终端可以是智能手机、平板电脑、笔记本电脑、台式电脑、个人数字助理和穿戴式设备等具有通信功能的电子设备，用户终端发送车间调度请求至服务器。第二是服务器，服务器接收客户端发送的车间调度请求，根据与所述车间调度请求对应的输入数据和约束条件，及调用预先存储的多目标车间调度优化模型进行超多目标的进化求解，得到最优解集。在服务器中得到所述最优解集后，将其发送至客户端。
在本实施例中，通过服务器检测是否接收到客户端发送的车间调度请求，当服务器接收到客户端发送的车间调度请求时则执行后续的步骤S120，当服务器未接收到客户端发送的车间调度请求时则等待预设的延迟时间后再次执行步骤S110。
S120、若接收到客户端发送的车间调度请求，获取与所述车间调度请求对应的输入数据和约束条件；其中，与所述车间调度请求对应的输入数据包括工件数、加工工序数、和机器数。
在本实施例中，若服务器接收到客户端发送的车间调度请求，获取与所述车间调度请求对应的输入数据和约束条件。由于服务器中已经预先存储了多目标车间调度优化模型，后续根据所述输入数据和约束条件即可进行求解，从而得到最优解集。
S130、调用预先存储的多目标车间调度优化模型，以所述输入数据为所述多目标车间调度优化模型的输入，根据所述约束条件和所述输入数据对所述多目标车间调度优化模型进行超多目标的进化求解，得到最优解集。
在本实施例中，服务器中存储的多目标车间调度优化模型，是一种多目标优化模型，通过对多目标车间调度优化模型进行求解，使得优化目标都尽可能达到满足的流水车间调度方案。
在一实施例中，所述多目标车间调度优化模型包括5个优化目标函数，分别记为：
最大完工时间优化目标函数f 1(x)、机器最大负荷优化目标函数f 2(x)、机器总负荷优化目标函数f 3(x)、总拖期优化目标函数f 4(x)、生产成本优化目标函数f 5(x)；
f 1(x)＝max{C i|i＝1,...,n}
Figure PCTCN2020079877-appb-000001
Figure PCTCN2020079877-appb-000002
Figure PCTCN2020079877-appb-000003
Figure PCTCN2020079877-appb-000004
其中，n为工件数量，C i表示第i个工件的完工时间，
Figure PCTCN2020079877-appb-000005
表示工件i的工序数量，n M表示加工机器数量，p ijk为第k个机器加工第i个工件的第j道工序所 需的加工时间，x ijk为用于判断第i个工件的第j道工序是否在第k个机器上进行加工的状态变量，D i为第i个工件的交货时间，
Figure PCTCN2020079877-appb-000006
为第i个工件的原材料成本值，
Figure PCTCN2020079877-appb-000007
为第k个机器的单位时间加工费用值。
即收到的车间调度请求后对应调用的多目标车间调度优化模型可以表示为
Figure PCTCN2020079877-appb-000008
m为车间调度优化的目标数，m的数值为5，Ω表示车间调度方案的集合。通过计算F(x)的最优解，即求F(x)最小值，使5个车间调度优化目标(5个车间调度优化目标分别为minf 1(x)、minf 2(x)、minf 3(x)、minf 4(x)、minf 5(x))同时达到最小。
对于所述多目标车间调度优化模型的一个候选解x，其指满足以上5个优化目标函数(即满足minf 1(x)、minf 2(x)、minf 3(x)、minf 4(x)、
minf 5(x))的一条调度方案，X表示包含多个候选解的集合，多个候选解组成的最优解集X 最优。以所述多目标车间调度优化模型进行最优解集的获取时，该模型是一种高维优化模型，结合上述5个目标函数即输入数据和约束条件求解得到最优解集X 最优时，能够最大地满足所提出的优化目标和约束条件。
在本实施例中，与所述车间调度请求对应的约束条件为：所有机器在最初时刻都可用；同一机器在同一时刻最多只可以加工一个工件；一个工件同一时刻只可以在一台机器上加工；工件的加工工序间存在先后顺序；加工过程中，每个工件加工优先级相同。
在一实施例中，如图3所示，所述步骤S130包括：
S1301、根据所述约束条件随机生成初始多目标种群；其中，所述初始多目标种群中包括多个个体，每一个体对应所述多目标车间调度优化模型的一个车间调度输出解，所述初始多目标种群中包括多个个体的总个数记为种群大小N；
S1302、获取当前迭代代数，判断所述当前迭代代数是否达到预设的最大迭代代数；
S1303、若所述当前迭代代数未达到所述最大迭代代数，对所述初始多目标种群进行模拟二进制交叉和多项式变异，得到与所述初始多目标种群有相同个体总个数的子种群；
S1304、将所述初始多目标种群与所述子种群进行合并，得到混合种群；
S1305、获取所述混合种群中的非支配解集及多层解集，及与所述非支配解集对应的自适应参考点；其中，所述非支配解集记为Q 1，所述多层解集中包括多个解集子集且分别记为Q 2至Q L，其中Q 1至Q L的并集为所述混合种群，Q 1至Q L中任意两个集合的交集为空集，Q 1≥Q 2≥Q 3≥……≥Q L；
S1306、获取所述混合种群中每一个体对应的目标空间点与所述自适应参考点之间分别对应的个体向量，将所述混合种群中每一个体对应的个体向量根据向量夹角相似度及所述种群大小N进行聚类，得到包括N个聚类簇的聚类结果；
S1307、通过环境选择法在所述聚类结果获取多个目标个体，以组成当前多目标种群，将所述当前多目标种群作为初始多目标种群；
S1308、将所述当前迭代代数加一以作为当前迭代代数，返回执行判断所述当前迭代代数是否达到预设的最大迭代代数的步骤；
S1309、若所述当前迭代代数达到所述最大迭代代数，将所述当前多目标种群输出作为最优解集。
在本实施例中，在约束条件的限制下随机生成一个初始多目标种群，该初始多目标种群为第一代多目标种群，此时先判断当前迭代代数是否达到预设的 最大迭代代数，以确定是否继续迭代执行后续步骤以获取最优解集。其中，当前迭代代数的初始值设置为1。若当前迭代代数达到了所述最大迭代代数，将所述当前多目标种群输出作为路径最优解集。
若当前迭代代数未达到所述最大迭代代数，先以初始多目标种群为初始种群进行模拟二进制交叉和多项式变异，也即在所述初始多目标种群中任意挑选两个个体以依次进行二进制交叉，直到生成N个交叉处理后新个体，对N个交叉处理后新个体进行多项式变异，由多项式变异后的新个体组成子种群。
在本实施例中，根据所述初始多目标种群中任意挑选两个个体进行二进制交叉处理后，得到N个交叉处理后新个体。这里多次任意挑选两个个体进行二进制交叉的过程也是一种迭代过程，直到新个体数达到种群大小N，才停止上述多次二进制交叉的处理过程。另外，二进制交叉和多项式变异均为常规处理过程，此处不再赘述。
之后将所述初始多目标种群与所述子种群进行合并，得到混合种群后，所述混合种群中所包括个体的总个数为所述种群大小N的2倍。
此时，可对所述混合种群中各个体进行非支配排序，从而得到非支配解集和多层解集。具体对所述混合种群中各个体进行非支配排序时，可通过非支配解(也可以称为帕累托解)的获取方式，来得到与所述混合种群对应的非支配解集。其中，帕累托解的定义为假设任何二解S1及S2对所有目标而言，S1均优于或同于S2，并且存在至少一个目标，S1在该目标上对应的目标值优于S2该目标上对应的目标值，则称S1支配S2，若S1的解没有被其他解所支配，则S1称为非支配解(不受支配解)，也称Pareto解(即帕累托解)。具体的，对所述混合种群中求解非支配解时，得到的非支配解集记为Q 1。所述混合种群中去掉非支配解集对应的个体之后，得到的为多层解集，所述多层解集中包括多个解集子集且分别记为Q 2至Q L，其中Q 1至Q L的并集为所述混合种群，Q 1至Q L中任意两个集合的交集为空集，Q 1≥Q 2≥Q 3≥……≥Q L；其中，“≥”表示支配关系，Q i≥Q j表示存在Q i中的解支配Q j，该关系是具有传递性，Q 1≥Q 2表示对于f 1(x)至f 5(x)而言，Q 2中的每个解都至少被Q 1中的一个解所支配，该关系具有传递性，即Q 3中的每个解至少被Q 1或Q 2中的一个解支配，其他的也依次类推。
当获取了所述非支配解集、及多层解集后，根据所述非支配解集中各非支配个体在归一化目标空间的位置确定自适应参考点的位置。自适应参考点的位置确定后，即可确定所述混合种群中每一个体对应的个体向量并求出各个体向量之间的向量夹角，进而通过聚类算法根据向量夹角将所述混合种群划分为N类，得到聚类结果C＝{C 1,C 2,…,C N}。
在一实施例中，如图4所示，步骤S1305包括：
S13051、在所述非支配解集、及多层解集中多个解集子集依序合并从而获取多个集合直至个体的总个数超出所述种群大小N，以组成目标集合；
S13052、获取所述目标集合中的最小值个体和最大值个体；其中，所述最小值个体输入至所述多目标车间调度优化模型得到的目标值为目标集合中每个个体对应的目标值中最小目标值，所述最大值个体输入至所述多目标车间调度优化模型得到的目标值为目标集合中每个个体对应的目标值中最大目标值；
S13053、根据所述最小值个体、所述最大值个体将所述目标集合中每一个个体进行归一化处理，得到归一化目标集合；其中，所述归一化目标集合中与所述非支配解集对应的归一化个体集合记为归一化非支配解集；
S13054、获取归一化非支配解集中每一非支配个体，分别记为B 1至B M；其中，M的取值与所述归一化非支配解集中归一化非支配个体的总个数相同；
S13055、获取所述归一化非支配解集中每一归一化非支配个体到目标超平面对应的超平面距离D i、及超平面距离D 1至D M对应的超平面距离平均值mpd；其中，i的取值范围为[1,M]，目标超平面为f 1(x)+f 2(x)+f 3(x)+f 4(x)+f 5(x)＝1；
S13056、获取所述归一化非支配解集中每一归一化非支配个体与目标超曲面集合对应的超曲面适应度；其中，超曲面适应度用fit p表示，所述目标超曲面集合为
Figure PCTCN2020079877-appb-000009
p＝{0.5,1.0,2.0}，
Figure PCTCN2020079877-appb-000010
S13057、调用预设的自适应参考点获取策略，根据所述自适应参考点获取策略和所述超曲面适应度获取与所述非支配解集对应的自适应参考点；其中，
Figure PCTCN2020079877-appb-000011
m的取值为5，
Figure PCTCN2020079877-appb-000012
表示自适应参考点第i维的坐标，β为预设的用于确定自适应参数移动步长的步长参数。
在本实施例中，为了获取所述自适应参考点的位置，可以先根据帕累托支配关系将所述混合种群划分为多个个体分层，也即分别记为Q 1至Q L，第一层的个体组成非支配解集Q 1，之后依序从Q 1至Q L选中多个个体分层，直至这多个个体分层中的个体的总个数超过所述种群大小N，以组成目标集合。
然后，对所述目标集合中的个体进行归一化处理。具体是，先获取所述目标集合中的最小值个体和最大值个体；其中，所述最小值个体输入至所述多目标车间调度优化模型得到的目标值为目标集合中每个个体对应的目标值中最小目标值，所述最大值个体输入至所述多目标车间调度优化模型得到的目标值为目标集合中每个个体对应的目标值中最大目标值。获取了所述最小值个体和所述最大值个体后，采用公式
Figure PCTCN2020079877-appb-000013
对目标集合中每个个体进行归一化操作，以得到归一化目标集合，其中，f i(x)表示目标集合中个体x在第i个车间调度优化目标上的值，f i'(x)表示目标集合中个体x在第i个车间调度优化目标上归一化之后的值，f imin表示目标集合中最小值个体在第i个车间调度优化目标上的值，f imax表示目标集合最大值个体在第i个车间调度优化目标上的值。
之后，获取归一化非支配解集中每一非支配个体，分别记为B 1至B M。计算B 1至B M到目标超平面对应的超平面距离D i、及超平面距离D 1至D M对应的超平面距离平均值mpd；
其中，i的取值范围为[1,M]；
目标超平面为f 1(x)+f 2(x)+f 3(x)+f 4(x)+f 5(x)＝1；
获取非支配个体B i到目标超平面对应的超平面距离D i的具体函数为
Figure PCTCN2020079877-appb-000014
即非支配个体B i到目标超平面的垂直距离，m的取值为5。
在获取所述归一化非支配解集中每一归一化非支配个体与目标超曲面集合对应的超曲面适应度的过程中，所述归一化非支配解集中个体分布与目标超曲面集合之间的超曲面适应度为
Figure PCTCN2020079877-appb-000015
其中p＝{0.5,1.0,2.0}，分别代表目标空间中的凸形超曲面、超平面和凹形超曲面。计算超曲面适应度fit p时，首先根据所述归一化非支配解集中所有个体计算平均闵可夫斯基距离，然后通过计算平均闵可夫斯基距离与1.0的差值来作为衡量所述归一化非支配解集中个体分布于对应超曲面的适应度。fit p越趋近于0，则认为个体分布与对应p值的适应度越高，即与该超曲面越相似。
通过计算出所述归一化非支配解集的分布与不同形状的超曲面的适应度，可以粗略预测出当前多目标车间调度优化问题最优解集的分布，进而可以根据该分布调整算法策略，增强算法性能。
最后，根据所述自适应参考点获取策略和所述超曲面适应度获取与所述非支配解集对应的自适应参考点的过程中，首先根据mpd的值确定自适应参考点的初始位置，然后根据fit p的取值确定自适应参考点在初始位置的基础上沿对角线方向移动一定距离，最终得到自适应参考点r*(其中，
Figure PCTCN2020079877-appb-000016
表示自适应参考点第i维的坐标)。
当mpd≥0时，代表所述归一化非支配解集中的大部分个体位于超平面f 1(x)+f 2(x)+...+f m(x)＝1的上方，此时可认为当前多目标车间调度优化问题的PF形状为平面或凹形，因此自适应参考点的初始位置为r u＝(0,0,…,0)。在此基础上，自适应参考点的位置由如下公式求出：
Figure PCTCN2020079877-appb-000017
在上式1中，
Figure PCTCN2020079877-appb-000018
表示自适应参考点第i维的坐标，α为预设参数(用于确定自适应参数移动的步长)，由于此时参考点初始位置为(0,0,…,0)，因此在式1中省略。
当mpd＜0时，代表所述归一化非支配解集中大部分个体位于超平面f 1(x)+f 2(x)+...+f m(x)＝1的下方，此时可认为当前多目标车间调度优化问题的PF为凸形。此时，应将最优解分布转化为凹形超曲面，使得本申请中基于角度的聚类方法得到更好的效果，因此参考点初始位置r u由下式求出：
Figure PCTCN2020079877-appb-000019
其中，m为目标数即等于5，
Figure PCTCN2020079877-appb-000020
为自适应参考点初始位置第i维的取值，R为当超曲面
Figure PCTCN2020079877-appb-000021
以r u为圆心时的半径。通过上式求得的初始位置如下：
Figure PCTCN2020079877-appb-000022
在此基础上，自适应参考点的位置由如下公式求出：
Figure PCTCN2020079877-appb-000023
其中，m的取值为5，
Figure PCTCN2020079877-appb-000024
表示自适应参考点第i维的坐标，β为预设的用于确定自适应参数移动步长的步长参数。本申请采用的自适应参考点位置的选取取决于自适应参考点的初始位置和最优解的分布与p取值为0.5、1.0、2.0的超曲面相对位置关系。采用自适应参考点的方法，使基于角度的聚类方法在解决PF形状为凸的问题时同样能具有优秀的表现。
之后，本申请中通过粗略地估计多目标车间调度问题的最优解的分布采用自适应参考点，使得待解决的多目标车间调度问题被转化为基于角度聚类的多目标优化算法擅长解决的最优解分布为凹形超曲面，即PF(Pareto-Optimal Front)形状为凹的问题。
在一实施例中，如图5所示，步骤S1306包括：
S13061、获取所述归一化目标集合中的每一归一化目标个体，将每一归一化目标个体划分为一个初始聚类簇，以组成初始聚类结果；
S13062、获取所述初始聚类结果中每一初始聚类簇对应的初始聚类中心；
S13063、获取所述初始聚类结果对应的各初始聚类中心与所述自适应参考点之间对应的当前个体向量，以组成当前个体向量集合；
S13064、将所述当前个体向量集合中向量夹角为当前最小夹角值的两个当前个体向量合并为一个聚类簇，将该聚类簇对应的两个当前个体向量从所述当前个体向量集合中移除以更新得到当前个体向量集合，返回执行所述将所述当前个体向量集合中向量夹角为当前最小夹角值的两个当前个体向量合并为一个聚类簇，直至当前个体向量集合对应的聚类数与种群大小N相等，得到包括N个聚类簇的聚类结果。
在本实施例中，采用公式
Figure PCTCN2020079877-appb-000025
计算参考点到聚类中心形成的向量间的夹角，并用以衡量聚类间的相似度，其中c i和c j分别为第i个聚类簇和第j个聚类簇的中心，r*为自适应参考点。计算出的聚类中心夹 角越小，对应的两个聚类的相似度越高。将相似度最高的两个聚类合并，重新计算新聚类的聚类中心和聚类间的相似度。重复此步骤直到聚类结果中的聚类数为N。其中，获取所述初始聚类结果中每一初始聚类簇对应的初始聚类中心时，采用公式
Figure PCTCN2020079877-appb-000026
计算聚类中心，其中C k表示聚类结果中第k个聚类簇，|C k|表示第k个聚类簇中个体的数量，c k,i表示第k个聚类中心的第i维坐标，i∈{1,2,...,m}。采用基于角度的聚类方式，保证每一代选出的解具有良好的多样性。
之后通过环境选择法在所述聚类结果获取多个目标个体时，是为了选择性能较好的个体以组成当前多目标种群，将所述当前多目标种群作为初始多目标种群。
在一实施例中，步骤S1307包括：
在所述聚类结果中获取向量夹角最大的两个目标个体，将向量夹角最大的两个目标个体对应的聚类簇从所述聚类结果中移除，以得到第一次筛选后聚类结果；其中，所述聚类结果中获取向量夹角最大的两个目标个体分别记为第一目标个体和第二目标个体；
获取所述第一次筛选后聚类结果中每一个体分别与所述第一目标个体和与所述第二目标个体对应的向量夹角，以每一个体分别与所述第一目标个体和与所述第二目标个体对应的向量夹角中最小者组成与所述第一次筛选后聚类结果对应的第一个体夹角集合，获取所述第一个体夹角集合中最大的向量夹角对应的个体作为第三目标个体，将第三目标个体对应的聚类簇从所述第一次筛选后聚类结果中移除，以得到第二次筛选后聚类结果；
获取所述第二次筛选后聚类结果中每一个体分别与所述第一目标个体、与第二目标个体、和与所述第三目标个体对应的向量夹角，以每一个体分别与所述第一目标个体、与所述第二目标个体和与所述第三目标个体对应的向量夹角中最小者组成与所述第二次筛选后聚类结果对应的第二个体夹角集合，获取所述第二个体夹角集合中最大的向量夹角对应的个体作为第四目标个体，将第四目标个体对应的聚类簇从所述第二次筛选后聚类结果中移除，以得到第三次筛选后聚类结果；
获取所述第三次筛选后聚类结果中每一个体分别与所述第一目标个体、与第二目标个体、与所述第三目标个体、和与所述第四目标个体对应的向量夹角，以每一个体分别与所述第一目标个体、与所述第二目标个体、与所述第三目标个体和与所述第四目标个体对应的向量夹角中最小者组成与所述第三次筛选后聚类结果对应的第三个体夹角集合，获取所述第三个体夹角集合中最大的向量夹角对应的个体作为第五目标个体，将第五目标个体对应的聚类簇从所述第三次筛选后聚类结果中移除，以得到第四次筛选后聚类结果；
调用预设的个体收敛度函数获取所述第四次筛选后聚类结果中每一聚类簇中收敛度最小的个体，以与第一目标个体、第二目标个体、第三目标个体、第四目标个体、第五目标个体组成当前多目标种群。
在本实施例中，在使用环境选择法获取所述聚类结果中的多个目标个体以组成当前多目标种群时，为了保证种群的多样性，在聚类的基础上，首先选出所述聚类结果中向量夹角最大的两个个体(分别记为第一目标个体和第二目标个体)，将其加入当前多目标种群并将其对应的聚类簇从所述聚类结果中删除得到第一次筛选后聚类结果。
然后通过计算第一次筛选后聚类结果内的每个个体与当前多目标种群(此时当前多目标种群中仅包括第一目标个体和第二目标个体)中个体的向量夹角，找 出距离当前多目标种群角度最大的一个个体(记为第三目标个体)，将其加入当前多目标种群并将其对应的聚类从第一次筛选后聚类结果中删除以得到第二次筛选后聚类结果，重复上述删选第三目标个体的过程，直到当前多目标种群中个体数量达到5，分别记为第一目标个体、第二目标个体、第三目标个体、第四目标个体、第五目标个体。
为了保证种群的收敛性，当已从聚类结果中挑选出5个目标个体后，从第四次筛选后聚类结果中剩余的每个聚类中选择一个收敛性最好的个体加入到当前多目标种群中，个体收敛性根据公式
Figure PCTCN2020079877-appb-000027
来衡量，其中m为优化目标数，f i'(x)表示个体x的第i个归一化后的目标值，收敛性越好，conf(x)值越小。在选择目标个体中所使用的环境选择法在聚类的基础上，选择个体的过程中在强化了多样性的同时又兼顾了个体的收敛性，能较好地解决基于多目标优化的调度方法多样性不足的问题。
具体实施时，所述最优解集中每一最优解的具体编码方式(也即最终展示给用户查看的方式)，是用户自定义编码格式并保存在服务器中，此次并不限定具体编码方式。
在步骤S1307中获取了当前多目标种群后，表示此轮迭代过程的结束，此时将所述当前多目标种群作为初始多目标种群，是为了在下一轮迭代过程开始时，以上一轮迭代结束时的所述当前多目标种群作为下一代迭代过程开始时新的初始多目标种群。
S140、将所述最优解集发送至客户端。
在本实施例中，当在服务器中完成了最优解集的获取之后，即可发送至客户端。从而客户端可根据所述最优解集确定流水车间的调度方式。
该方法实现了在超多目标的进化求解的过程中在搜索空间巨大的前提下快速求解，且保持了可行解的多样性。
本申请实施例还提供一种多目标流水车间调度装置，该多目标流水车间调度装置用于执行前述多目标流水车间调度方法的任一实施例。具体地，请参阅图6，图6是本申请实施例提供的多目标流水车间调度装置的示意性框图。该多目标流水车间调度装置100可以被配置于服务器中。
如图6所示，多目标流水车间调度装置100包括车间调度请求检测单元110、输入数据条件获取单元120、最优解集求解单元130、及最优解集发送单元140。
其中，车间调度请求检测单元110，用于判断是否接收到客户端发送的车间调度请求。输入数据条件获取单元120，用于若接收到客户端发送的车间调度请求，获取与所述车间调度请求对应的输入数据和约束条件；其中，与所述车间调度请求对应的输入数据包括工件数、加工工序数、和机器数。最优解集求解单元130，用于调用预先存储的多目标车间调度优化模型，以所述输入数据为所述多目标车间调度优化模型的输入，根据所述约束条件和所述输入数据对所述多目标车间调度优化模型进行超多目标的进化求解，得到最优解集。最优解集发送单元140，用于将所述最优解集发送至客户端。
在一实施例中，所述最优解集求解单元130包括：
初始多目标种群生成单元，用于根据所述约束条件随机生成初始多目标种 群；其中，所述初始多目标种群中包括多个个体，每一个体对应所述多目标车间调度优化模型的一个车间调度输出解，所述初始多目标种群中包括多个个体的总个数记为种群大小N；
当前迭代代数第一判断单元，用于获取当前迭代代数，判断所述当前迭代代数是否达到预设的最大迭代代数；
个体交叉变异单元，用于若所述当前迭代代数未达到所述最大迭代代数，对所述初始多目标种群进行模拟二进制交叉和多项式变异，得到与所述初始多目标种群有相同个体总个数的子种群；
种群合并单元，用于将所述初始多目标种群与所述子种群进行合并，得到混合种群；
自适应参考点获取单元，用于获取所述混合种群中的非支配解集及多层解集，及与所述非支配解集对应的自适应参考点；其中，所述非支配解集记为Q 1，所述多层解集中包括多个解集子集且分别记为Q 2至Q L，其中Q 1至Q L的并集为所述混合种群，Q 1至Q L中任意两个集合的交集为空集，Q 1≥Q 2≥Q 3≥……≥Q L；
夹角聚类单元，用于获取所述混合种群中每一个体对应的目标空间点与所述自适应参考点之间分别对应的个体向量，将所述混合种群中每一个体对应的个体向量根据向量夹角相似度及所述种群大小N进行聚类，得到包括N个聚类簇的聚类结果；
初始多目标种群更新单元，用于通过环境选择法在所述聚类结果获取多个目标个体，以组成当前多目标种群，将所述当前多目标种群作为初始多目标种群；
当前迭代代数自增单元，用于将所述当前迭代代数加一以作为当前迭代代数，返回当前迭代代数第一判断单元以执行判断所述当前迭代代数是否达到预设的最大迭代代数的步骤；
最优解集输出单元，用于若所述当前迭代代数达到所述最大迭代代数，将所述当前多目标种群输出作为最优解集。
在一实施例中，所述自适应参考点获取单元，包括：
目标集合获取单元，用于在所述非支配解集、及多层解集中多个解集子集依序合并从而获取多个集合直至个体的总个数超出所述种群大小N，以组成目标集合；
端点个体获取单元，用于获取所述目标集合中的最小值个体和最大值个体；其中，所述最小值个体输入至所述多目标车间调度优化模型得到的目标值为目标集合中每个个体对应的目标值中最小目标值，所述最大值个体输入至所述多目标车间调度优化模型得到的目标值为目标集合中每个个体对应的目标值中最大目标值；
归一化处理单元，用于根据所述最小值个体、所述最大值个体将所述目标集合中每一个个体进行归一化处理，得到归一化目标集合；其中，所述归一化目标集合中与所述非支配解集对应的归一化个体集合记为归一化非支配解集；
获取归一化非支配解集中每一非支配个体，分别记为B 1至B M；其中，M 的取值与所述归一化非支配解集中归一化非支配个体的总个数相同；
超平面距离获取单元，用于获取所述归一化非支配解集中每一归一化非支配个体到目标超平面对应的超平面距离D i、及超平面距离D 1至D M对应的超平面距离平均值mpd；其中，i的取值范围为[1,M]，目标超平面为f 1(x)+f 2(x)+f 3(x)+f 4(x)+f 5(x)＝1；
超曲面适应度获取单元，用于获取所述归一化非支配解集中每一归一化非支配个体与目标超曲面集合对应的超曲面适应度；其中，超曲面适应度用fit p表示，所述目标超曲面集合为
Figure PCTCN2020079877-appb-000028
p＝{0.5,1.0,2.0}，
Figure PCTCN2020079877-appb-000029
自适应参考点定位单元，用于调用预设的自适应参考点获取策略，根据所述自适应参考点获取策略和所述超曲面适应度获取与所述非支配解集对应的自适应参考点；其中，
Figure PCTCN2020079877-appb-000030
m的取值为5，
Figure PCTCN2020079877-appb-000031
表示自适应参考点第i维的坐标，β为预设的用于确定自适应参数移动步长的步长参数。
在一实施例中，所述夹角聚类单元包括：
初始聚类结果获取单元，用于获取所述归一化目标集合中的每一归一化目标个体，将每一归一化目标个体划分为一个初始聚类簇，以组成初始聚类结果；
初始聚类中心获取单元，用于获取所述初始聚类结果中每一初始聚类簇对应的初始聚类中心；
当前个体向量集合获取单元，用于获取所述初始聚类结果对应的各初始聚类中心与所述自适应参考点之间对应的当前个体向量，以组成当前个体向量集合；
聚类簇筛选单元，用于将所述当前个体向量集合中向量夹角为当前最小夹角值的两个当前个体向量合并为一个聚类簇，将该聚类簇对应的两个当前个体向量从所述当前个体向量集合中移除以更新得到当前个体向量集合，返回执行所述将所述当前个体向量集合中向量夹角为当前最小夹角值的两个当前个体向量合并为一个聚类簇，直至当前个体向量集合对应的聚类数与种群大小N相等，得到包括N个聚类簇的聚类结果。
在一实施例中，所述初始多目标种群更新单元包括：
第一次个体筛选单元，用于在所述聚类结果中获取向量夹角最大的两个目标个体，将向量夹角最大的两个目标个体对应的聚类簇从所述聚类结果中移除，以得到第一次筛选后聚类结果；其中，所述聚类结果中获取向量夹角最大的两 个目标个体分别记为第一目标个体和第二目标个体；
第二次个体筛选单元，用于获取所述第一次筛选后聚类结果中每一个体分别与所述第一目标个体和与所述第二目标个体对应的向量夹角，以每一个体分别与所述第一目标个体和与所述第二目标个体对应的向量夹角中最小者组成与所述第一次筛选后聚类结果对应的第一个体夹角集合，获取所述第一个体夹角集合中最大的向量夹角对应的个体作为第三目标个体，将第三目标个体对应的聚类簇从所述第一次筛选后聚类结果中移除，以得到第二次筛选后聚类结果；
第三次个体筛选单元，用于获取所述第二次筛选后聚类结果中每一个体分别与所述第一目标个体、与第二目标个体、和与所述第三目标个体对应的向量夹角，以每一个体分别与所述第一目标个体、与所述第二目标个体和与所述第三目标个体对应的向量夹角中最小者组成与所述第二次筛选后聚类结果对应的第二个体夹角集合，获取所述第二个体夹角集合中最大的向量夹角对应的个体作为第四目标个体，将第四目标个体对应的聚类簇从所述第二次筛选后聚类结果中移除，以得到第三次筛选后聚类结果；
第四次个体筛选单元，用于获取所述第三次筛选后聚类结果中每一个体分别与所述第一目标个体、与第二目标个体、与所述第三目标个体、和与所述第四目标个体对应的向量夹角，以每一个体分别与所述第一目标个体、与所述第二目标个体、与所述第三目标个体和与所述第四目标个体对应的向量夹角中最小者组成与所述第三次筛选后聚类结果对应的第三个体夹角集合，获取所述第三个体夹角集合中最大的向量夹角对应的个体作为第五目标个体，将第五目标个体对应的聚类簇从所述第三次筛选后聚类结果中移除，以得到第四次筛选后聚类结果；
第五次个体筛选单元，用于调用预设的个体收敛度函数获取所述第四次筛选后聚类结果中每一聚类簇中收敛度最小的个体，以与第一目标个体、第二目标个体、第三目标个体、第四目标个体、第五目标个体组成当前多目标种群。
在一实施例中，所述个体交叉变异单元还用于：
在所述初始多目标种群中任意挑选两个个体以依次进行二进制交叉，直到生成N个交叉处理后新个体，对N个交叉处理后新个体进行多项式变异，由多项式变异后的新个体组成子种群。
该装置实现了在超多目标的进化求解的过程中在搜索空间巨大的前提下快速求解，且保持了可行解的多样性。
上述多目标流水车间调度装置可以实现为计算机程序的形式，该计算机程序可以在如图7所示的计算机设备上运行。
请参阅图7，图7是本申请实施例提供的计算机设备的示意性框图。该计算机设备500是服务器，服务器可以是独立的服务器，也可以是多个服务器组成的服务器集群。
参阅图7，该计算机设备500包括通过系统总线501连接的处理器502、存储器和网络接口505，其中，存储器可以包括非易失性存储介质503和内存储器504。
该非易失性存储介质503可存储操作系统5031和计算机程序5032。该计算 机程序5032被执行时，可使得处理器502执行多目标流水车间调度方法。
该处理器502用于提供计算和控制能力，支撑整个计算机设备500的运行。
该内存储器504为非易失性存储介质503中的计算机程序5032的运行提供环境，该计算机程序5032被处理器502执行时，可使得处理器502执行多目标流水车间调度方法。
该网络接口505用于进行网络通信，如提供数据信息的传输等。本领域技术人员可以理解，图7中示出的结构，仅仅是与本申请方案相关的部分结构的框图，并不构成对本申请方案所应用于其上的计算机设备500的限定，具体的计算机设备500可以包括比图中所示更多或更少的部件，或者组合某些部件，或者具有不同的部件布置。
其中，所述处理器502用于运行存储在存储器中的计算机程序5032，以实现本申请实施例公开的多目标流水车间调度方法。
本领域技术人员可以理解，图7中示出的计算机设备的实施例并不构成对计算机设备具体构成的限定，在其他实施例中，计算机设备可以包括比图示更多或更少的部件，或者组合某些部件，或者不同的部件布置。例如，在一些实施例中，计算机设备可以仅包括存储器及处理器，在这样的实施例中，存储器及处理器的结构及功能与图7所示实施例一致，在此不再赘述。
应当理解，在本申请实施例中，处理器502可以是中央处理单元(Central ProcessingUnit，CPU)，该处理器502还可以是其他通用处理器、数字信号处理器(Digital Signal Processor，DSP)、专用集成电路(Application Specific Integrated Circuit，ASIC)、现成可编程门阵列(Field-Programmable GateArray，FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件等。其中，通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等。
在本申请的另一实施例中提供计算机可读存储介质。该计算机可读存储介质可以为非易失性的计算机可读存储介质。该计算机可读存储介质存储有计算机程序，其中计算机程序被处理器执行时实现本申请实施例公开的多目标流水车间调度方法。
所述存储介质为实体的、非瞬时性的存储介质，例如可以是U盘、移动硬盘、只读存储器(Read-OnlyMemory，ROM)、磁碟或者光盘等各种可以存储程序代码的实体存储介质。
所属领域的技术人员可以清楚地了解到，为了描述的方便和简洁，上述描述的设备、装置和单元的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。
以上所述，仅为本申请的具体实施方式，但本申请的保护范围并不局限于此，任何熟悉本技术领域的技术人员在本申请揭露的技术范围内，可轻易想到各种等效的修改或替换，这些修改或替换都应涵盖在本申请的保护范围之内。因此，本申请的保护范围应以权利要求的保护范围为准。

Claims (10)

  1. 一种多目标流水车间调度方法，包括：
    判断是否接收到客户端发送的车间调度请求；
    若接收到客户端发送的车间调度请求，获取与所述车间调度请求对应的输入数据和约束条件；其中，与所述车间调度请求对应的输入数据包括工件数、加工工序数、和机器数；
    调用预先存储的多目标车间调度优化模型，以所述输入数据为所述多目标车间调度优化模型的输入，根据所述约束条件和所述输入数据对所述多目标车间调度优化模型进行超多目标的进化求解，得到最优解集；以及
    将所述最优解集发送至客户端。
  2. 根据权利要求1所述的多目标流水车间调度方法，其中，所述多目标车间调度优化模型包括5个优化目标函数，分别记为最大完工时间优化目标函数f 1(x)、机器最大负荷优化目标函数f 2(x)、机器总负荷优化目标函数f 3(x)、总拖期优化目标函数f 4(x)、生产成本优化目标函数f 5(x)；
    f 1(x)＝max{C i|i＝1,...,n}
    Figure PCTCN2020079877-appb-100001
    Figure PCTCN2020079877-appb-100002
    Figure PCTCN2020079877-appb-100003
    Figure PCTCN2020079877-appb-100004
    其中，n为工件数量，C i表示第i个工件的完工时间，
    Figure PCTCN2020079877-appb-100005
    表示工件i的工序数量，n M表示加工机器数量，p ijk为第k个机器加工第i个工件的第j道工序所需的加工时间，x ijk为用于判断第i个工件的第j道工序是否在第k个机器上进行加工的状态变量，D i为第i个工件的交货时间，
    Figure PCTCN2020079877-appb-100006
    为第i个工件的原材料成本值，
    Figure PCTCN2020079877-appb-100007
    为第k个机器的单位时间加工费用值。
  3. 根据权利要求2所述的多目标流水车间调度方法，其中，所述以所述输入数据为所述多目标车间调度优化模型的输入，根据所述约束条件和所述输入数据对所述多目标车间调度优化模型进行超多目标的进化求解，得到最优解集，包括：
    根据所述约束条件随机生成初始多目标种群；其中，所述初始多目标种群中包括多个个体，每一个体对应所述多目标车间调度优化模型的一个车间调度输出解，所述初始多目标种群中包括多个个体的总个数记为种群大小N；
    获取当前迭代代数，判断所述当前迭代代数是否达到预设的最大迭代代数；
    若所述当前迭代代数未达到所述最大迭代代数，对所述初始多目标种群进行模拟二进制交叉和多项式变异，得到与所述初始多目标种群有相同个体总个数的子种群；
    将所述初始多目标种群与所述子种群进行合并，得到混合种群；
    获取所述混合种群中的非支配解集及多层解集，及与所述非支配解集对应的自适应参考点；其中，所述非支配解集记为Q 1，所述多层解集中包括多个解集子集且分别记为Q 2至Q L，其中Q 1至Q L的并集为所述混合种群，Q 1至Q L中任意两个集合的交集为空集，Q 1≥Q 2≥Q 3≥……≥Q L；
    获取所述混合种群中每一个体对应的目标空间点与所述自适应参考点之间分别对应的个体向量，将所述混合种群中每一个体对应的个体向量根据向量夹角相似度及所述种群大小N进行聚类，得到包括N个聚类簇的聚类结果；
    通过环境选择法在所述聚类结果获取多个目标个体，以组成当前多目标种群，将所述当前多目标种群作为初始多目标种群；
    将所述当前迭代代数加一以作为当前迭代代数，返回执行判断所述当前迭代代数是否达到预设的最大迭代代数的步骤；
    若所述当前迭代代数达到所述最大迭代代数，将所述当前多目标种群输出作为最优解集。
  4. 根据权利要求3所述的多目标流水车间调度方法，其中，所述获取所述混合种群中的非支配解集及多层解集，及与所述非支配解集对应的自适应参考点，包括：
    在所述非支配解集、及多层解集中多个解集子集依序合并从而获取多个集合直至个体的总个数超出所述种群大小N，以组成目标集合；
    获取所述目标集合中的最小值个体和最大值个体；其中，所述最小值个体输入至所述多目标车间调度优化模型得到的目标值为目标集合中每个个体对应的目标值中最小目标值，所述最大值个体输入至所述多目标车间调度优化模型得到的目标值为目标集合中每个个体对应的目标值中最大目标值；
    根据所述最小值个体、所述最大值个体将所述目标集合中每一个个体进行归一化处理，得到归一化目标集合；其中，所述归一化目标集合中与所述非支配解集对应的归一化个体集合记为归一化非支配解集；
    获取归一化非支配解集中每一非支配个体，分别记为B 1至B M；其中，M的取值与所述归一化非支配解集中归一化非支配个体的总个数相同；
    获取所述归一化非支配解集中每一归一化非支配个体到目标超平面对应的超平面距离D i、及超平面距离D 1至D M对应的超平面距离平均值mpd；其中，i的取值范围为[1,M]，目标超平面为f 1(x)+f 2(x)+f 3(x)+f 4(x)+f 5(x)＝1；
    获取所述归一化非支配解集中每一归一化非支配个体与目标超曲面集合对应的超曲面适应度；其中，超曲面适应度用fit p表示，所述目标超曲面集合为f 1 p+f 2 p+f 3 p+f 4 p+f 5 p＝1，p＝{0.5,1.0,2.0}，
    Figure PCTCN2020079877-appb-100008
    调用预设的自适应参考点获取策略，根据所述自适应参考点获取策略和所述超曲面适应度获取与所述非支配解集对应的自适应参考点；其中，
    Figure PCTCN2020079877-appb-100009
    m的取值为5，
    Figure PCTCN2020079877-appb-100010
    表示自适应参考点第i维的坐标，β为预设的用于确定自适应参数移动步长的步长参数。
  5. 根据权利要求4所述的多目标流水车间调度方法，其中，所述获取所述混合种群中每一个体对应的目标空间点与所述自适应参考点之间分别对应的个体向量，将所述混合种群中每一个体对应的个体向量根据向量夹角相似度及所述种群大小N进行聚类，得到包括N个聚类簇的聚类结果，包括：
    获取所述归一化目标集合中的每一归一化目标个体，将每一归一化目标个体划分为一个初始聚类簇，以组成初始聚类结果；
    获取所述初始聚类结果中每一初始聚类簇对应的初始聚类中心；
    获取所述初始聚类结果对应的各初始聚类中心与所述自适应参考点之间对应的当前个体向量，以组成当前个体向量集合；
    将所述当前个体向量集合中向量夹角为当前最小夹角值的两个当前个体向量合并为一个聚类簇，将该聚类簇对应的两个当前个体向量从所述当前个体向量集合中移除以更新得到当前个体向量集合，返回执行所述将所述当前个体向量集合中向量夹角为当前最小夹角值的两个当前个体向量合并为一个聚类簇，直至当前个体向量集合对应的聚类数与种群大小N相等，得到包括N个聚类簇的聚类结果。
  6. 根据权利要求3所述的多目标流水车间调度方法，其中，所述通过环境选择法在所述聚类结果获取多个目标个体，以组成当前多目标种群，包括：
    在所述聚类结果中获取向量夹角最大的两个目标个体，将向量夹角最大的两个目标个体对应的聚类簇从所述聚类结果中移除，以得到第一次筛选后聚类结果；其中，所述聚类结果中获取向量夹角最大的两个目标个体分别记为第一目标个体和第二目标个体；
    获取所述第一次筛选后聚类结果中每一个体分别与所述第一目标个体和与所述第二目标个体对应的向量夹角，以每一个体分别与所述第一目标个体和与所述第二目标个体对应的向量夹角中最小者组成与所述第一次筛选后聚类结果对应的第一个体夹角集合，获取所述第一个体夹角集合中最大的向量夹角对应的个体作为第三目标个体，将第三目标个体对应的聚类簇从所述第一次筛选后聚类结果中移除，以得到第二次筛选后聚类结果；
    获取所述第二次筛选后聚类结果中每一个体分别与所述第一目标个体、与第二目标个体、和与所述第三目标个体对应的向量夹角，以每一个体分别与所述第一目标个体、与所述第二目标个体和与所述第三目标个体对应的向量夹角 中最小者组成与所述第二次筛选后聚类结果对应的第二个体夹角集合，获取所述第二个体夹角集合中最大的向量夹角对应的个体作为第四目标个体，将第四目标个体对应的聚类簇从所述第二次筛选后聚类结果中移除，以得到第三次筛选后聚类结果；
    获取所述第三次筛选后聚类结果中每一个体分别与所述第一目标个体、与第二目标个体、与所述第三目标个体、和与所述第四目标个体对应的向量夹角，以每一个体分别与所述第一目标个体、与所述第二目标个体、与所述第三目标个体和与所述第四目标个体对应的向量夹角中最小者组成与所述第三次筛选后聚类结果对应的第三个体夹角集合，获取所述第三个体夹角集合中最大的向量夹角对应的个体作为第五目标个体，将第五目标个体对应的聚类簇从所述第三次筛选后聚类结果中移除，以得到第四次筛选后聚类结果；
    调用预设的个体收敛度函数获取所述第四次筛选后聚类结果中每一聚类簇中收敛度最小的个体，以与第一目标个体、第二目标个体、第三目标个体、第四目标个体、第五目标个体组成当前多目标种群。
  7. 根据权利要求3所述的多目标流水车间调度方法，其中，所述对所述初始多目标种群进行模拟二进制交叉和多项式变异，得到与所述初始多目标种群有相同个体总个数的子种群，包括：
    在所述初始多目标种群中任意挑选两个个体以依次进行二进制交叉，直到生成N个交叉处理后新个体，对N个交叉处理后新个体进行多项式变异，由多项式变异后的新个体组成子种群。
  8. 一种多目标流水车间调度装置，其中，包括：
    车间调度请求检测单元，用于判断是否接收到客户端发送的车间调度请求；
    输入数据条件获取单元，用于若接收到客户端发送的车间调度请求，获取与所述车间调度请求对应的输入数据和约束条件；其中，与所述车间调度请求对应的输入数据包括工件数、加工工序数、和机器数；
    最优解集求解单元，用于调用预先存储的多目标车间调度优化模型，以所述输入数据为所述多目标车间调度优化模型的输入，根据所述约束条件和所述输入数据对所述多目标车间调度优化模型进行超多目标的进化求解，得到最优解集；
    最优解集发送单元，用于将所述最优解集发送至客户端。
  9. 一种计算机设备，包括存储器、处理器及存储在所述存储器上并可在所述处理器上运行的计算机程序，其中，所述处理器执行所述计算机程序时实现以下步骤：
    判断是否接收到客户端发送的车间调度请求；
    若接收到客户端发送的车间调度请求，获取与所述车间调度请求对应的输入数据和约束条件；其中，与所述车间调度请求对应的输入数据包括工件数、加工工序数、和机器数；
    调用预先存储的多目标车间调度优化模型，以所述输入数据为所述多目标车间调度优化模型的输入，根据所述约束条件和所述输入数据对所述多目标车间调度优化模型进行超多目标的进化求解，得到最优解集；以及
    将所述最优解集发送至客户端。
  10. 一种计算机可读存储介质，其中，所述计算机可读存储介质存储有计算机程序，所述计算机程序当被处理器执行时使所述处理器执行以下操作：
    判断是否接收到客户端发送的车间调度请求；
    若接收到客户端发送的车间调度请求，获取与所述车间调度请求对应的输入数据和约束条件；其中，与所述车间调度请求对应的输入数据包括工件数、加工工序数、和机器数；
    调用预先存储的多目标车间调度优化模型，以所述输入数据为所述多目标车间调度优化模型的输入，根据所述约束条件和所述输入数据对所述多目标车间调度优化模型进行超多目标的进化求解，得到最优解集；以及
    将所述最优解集发送至客户端。
PCT/CN2020/079877 2020-01-15 2020-03-18 多目标流水车间调度方法、装置、计算机设备及存储介质 WO2021142915A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010041521.7 2020-01-15
CN202010041521.7A CN111259312B (zh) 2020-01-15 2020-01-15 多目标流水车间调度方法、装置、计算机设备及存储介质

Publications (1)

Publication Number Publication Date
WO2021142915A1 true WO2021142915A1 (zh) 2021-07-22

Family

ID=70954057

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/079877 WO2021142915A1 (zh) 2020-01-15 2020-03-18 多目标流水车间调度方法、装置、计算机设备及存储介质

Country Status (2)

Country Link
CN (1) CN111259312B (zh)
WO (1) WO2021142915A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114707294A (zh) * 2022-01-28 2022-07-05 湘南学院 有限运输能力约束的作业车间多目标调度方法

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112859785B (zh) * 2021-01-19 2021-12-17 嘉兴学院 基于多目标优化算法的纸盆车间生产调度方法及调度系统
CN113807608B (zh) * 2021-10-09 2022-07-01 哈尔滨学院 基于物流优化的冷链供应管理系统、存储介质及设备

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101295379A (zh) * 2008-06-19 2008-10-29 上海交通大学 面向客户需求的车间作业调度系统
CN104880949A (zh) * 2015-05-27 2015-09-02 江南大学 基于改进鸡群算法获得工件加工最优调度的方法
CN104914835A (zh) * 2015-05-22 2015-09-16 齐鲁工业大学 一种柔性作业车间调度多目标的方法
JP2019168763A (ja) * 2018-03-22 2019-10-03 三菱電機株式会社 運転開始条件変換装置および運転開始条件変換方法

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108320057B (zh) * 2018-02-07 2021-06-18 江南大学 一种基于有限制稳定配对策略的柔性作业车间调度方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101295379A (zh) * 2008-06-19 2008-10-29 上海交通大学 面向客户需求的车间作业调度系统
CN104914835A (zh) * 2015-05-22 2015-09-16 齐鲁工业大学 一种柔性作业车间调度多目标的方法
CN104880949A (zh) * 2015-05-27 2015-09-02 江南大学 基于改进鸡群算法获得工件加工最优调度的方法
JP2019168763A (ja) * 2018-03-22 2019-10-03 三菱電機株式会社 運転開始条件変換装置および運転開始条件変換方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
WANG XIAOJUAN: "Research on Methods for Multi-objective Flexible Job-shop Scheduling Problem", CHINESE DOCTORAL DISSERTATIONS FULL-TEXT DATABASE, UNIVERSITY OF CHINESE ACADEMY OF SCIENCES, CN, 15 October 2011 (2011-10-15), CN, XP055828477, ISSN: 1674-022X *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114707294A (zh) * 2022-01-28 2022-07-05 湘南学院 有限运输能力约束的作业车间多目标调度方法
CN114707294B (zh) * 2022-01-28 2023-02-07 湘南学院 有限运输能力约束的作业车间多目标调度方法

Also Published As

Publication number Publication date
CN111259312A (zh) 2020-06-09
CN111259312B (zh) 2021-08-17

Similar Documents

Publication Publication Date Title
WO2021142917A1 (zh) 多车场车辆路径规划方法、装置、计算机设备及存储介质
WO2021142915A1 (zh) 多目标流水车间调度方法、装置、计算机设备及存储介质
US11574202B1 (en) Data mining technique with distributed novelty search
CN110869943A (zh) Gpu增强的图形模型构建和评分引擎
JP2010514035A (ja) データセットの変更を最適化するシステム及び方法
WO2021042844A1 (zh) 大规模数据聚类方法、装置、计算机设备及计算机可读存储介质
WO2020134099A1 (zh) 物品识别方法、设备和系统
TW202217662A (zh) 視覺定位方法及相關模型的訓練方法、電子設備和電腦可讀儲存介質
CN115238062A (zh) 技术产权匹配方法和系统
US9749277B1 (en) Systems and methods for estimating sender similarity based on user labels
US20240012966A1 (en) Method and system for providing a three-dimensional computer aided-design (cad) model in a cad environment
CN112966054A (zh) 基于企业图谱节点间关系的族群划分方法和计算机设备
CN111813902A (zh) 智能应答方法、系统及计算设备
CN112215655A (zh) 一种客户画像的标签管理方法及系统
CN113657525B (zh) 基于KMeans的跨特征联邦聚类方法及相关设备
WO2021036190A1 (zh) 节点布局确定方法以及装置
CN115186890A (zh) 预警方法、装置、计算机设备和存储介质
CN106709572A (zh) 一种数据处理方法及设备
CN113537308A (zh) 基于本地化差分隐私的两阶段k-means聚类处理系统及方法
CN112637880A (zh) 问题小区处理方法以及装置
US20230054360A1 (en) Machine learning consolidation
US20230244700A1 (en) System and method for identifying approximate k-nearest neighbors in web scale clustering
CN109002476A (zh) 高维数据的聚类方法、电子设备及系统
Shekhar et al. Geolocation Clustering for Radio Resource Management in 6G Networks
Othman et al. Evaluation of Geo-SPEBH algorithm based on Bandwidth for Big Data retrieval in Cloud Computing

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20913798

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 15/11/2022)

122 Ep: pct application non-entry in european phase

Ref document number: 20913798

Country of ref document: EP

Kind code of ref document: A1