Manufacturing resource scheduling method for multi-task cooperation in industrial interconnection environment

Abstract

A manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment comprises the following steps: 1) establishing a manufacturing resource scheduling model in combination with a nonlinear process planning theory; 2) solving the manufacturing resource scheduling model by adopting an improved genetic variable neighborhood algorithm, and outputting an optimal solution of a static manufacturing resource scheduling scheme; 3) executing the optimal solution of the static manufacturing resource scheduling scheme, judging whether the scheduling of all industrial manufacturing tasks is completed or not, if so, outputting the optimal manufacturing resource scheduling scheme, and if not, entering the step 4); and 4) adjusting the manufacturing resource scheduling model based on the event-driven and periodic rescheduling dynamic scheduling strategy, and returning the adjusted manufacturing resource scheduling model to the step 2). According to the method, the generation of the multi-task cooperation-oriented machine tool equipment resource dynamic scheduling scheme in the industrial interconnection environment is realized, and corresponding technical support is provided for networked integration and optimized operation of machine tool equipment resources in the industrial interconnection environment.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a manufacturing resource scheduling method oriented to multi-task collaboration in an industrial interconnection environment. Background Art

[0002] The networked integration and coordinated optimization of manufacturing resources, especially the large-scale and wide-ranging machine tool equipment resources, is the core and key to promoting the application of the industrial Internet service model to the bottom layer of the vast discrete manufacturing workshop. Compared with the scheduling of computing power, production materials and other resources, the problem of optimizing the scheduling of machine tool equipment resources is more difficult to solve. It is mainly reflected in the following aspects: the development of diversified manufacturing business under the industrial Internet model is completed through remote multi-point collaboration based on the network, but machine tool equipment resources have the characteristics of wide-area dispersion and high autonomy, which makes it difficult to achieve intensive collaborative management. Moreover, they are easily affected by random disturbances such as emergency orders from resource providers, abnormal production quality, and system failures, resulting in the risk of service suspension or even withdrawal of machine tool equipment resources that provide industrial Internet processing services, making it difficult to ensure the efficient and reliable operation of industrial Internet services. Combined with the characteristics of random and high-frequency production disturbances in discrete manufacturing environments, the dynamic scheduling and optimal configuration of machine tool equipment resources are realized, and the networked collaborative operation of multi-tasks and multi-resources under the industrial Internet model is supported. It has become a technical problem that needs to be solved urgently on the road of transformation and upgrading of traditional manufacturing enterprises. Summary of the invention

[0003] The purpose of the present invention is to provide a manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment, comprising the following steps:

[0004] 1) Combined with nonlinear process planning theory, a manufacturing resource scheduling model is established.

[0005] 2) The improved genetic variable neighborhood algorithm is used to solve the manufacturing resource scheduling model and output the optimal solution for the static manufacturing resource scheduling scheme.

[0006] 3) Execute the optimal solution of the static manufacturing resource scheduling plan to determine whether the scheduling of all industrial manufacturing tasks has been completed. If so, output the optimal manufacturing resource scheduling plan. If not, proceed to step 4).

[0007] 4) Adjust the manufacturing resource scheduling model based on the dynamic scheduling strategy of event-driven and periodic rescheduling, and return the adjusted manufacturing resource scheduling model to step 2).

[0008] Furthermore, the objective function of the manufacturing resource scheduling model is as follows:

[0009]

[0010] stf1≤T max (2)

[0011]

[0012]

[0013] Where F is the objective function of the manufacturing resource scheduling model. b is the function number. b is the weight value. max The maximum delivery time allowed by the demander of the manufacturing task. is the maximum service cost. is the minimum service quality. i is the manufacturing task number, n is the total number of manufacturing tasks. k is the process route number, l is the total number of process routes. u is the machine tool equipment resource number, m is the total number of machine tool equipment resources. w is the process number, s is the total number of processes, and j is the subtask number. β u The load level of machine tool equipment resources. They respectively represent the start processing time of the wth and w+1th processes on the machine tool equipment resources. Indicates the processing time of the process on the machine tool equipment resources. ijkwu Indicates whether the process selects a machine tool from the machine tool resource set. ijk Indicates whether the subtask selects a process from the process route set.

[0014] Among them, the standardized formula f b 'As shown below:

[0015]

[0016] Among them, the service time function f1, the service cost function f2, the service quality function f3, the service reliability function f4, the service energy consumption function f5, and the resource load balancing function f6 are as follows:

[0017]

[0018] Where h is the total number of subtasks. It indicates the transportation time when two adjacent processes are processed on machine tool equipment resources in different regions. It represents the processing cost per unit time of a process on machine tool equipment resources. It represents the transportation cost when two adjacent processes are processed on machine tool equipment resources in different regions. ij k wu Indicates the process in the machine tool equipment resource R u The quality pass rate during processing. i jk wu Indicates the process in the machine tool equipment resource R u On-time completion rate during processing.u Indicates machine tool equipment resources R u The failure rate. ijkwu , P′ ijkwu They respectively represent the cutting power and no-load power of the process when it is processed on the machine tool equipment resources. They represent auxiliary processing time and basic processing time respectively.

[0019] Among them, each manufacturing task T i The service cost cannot exceed the expected value given by the platform users, that is:

[0020]

[0021] In the formula, To create task T i service cost.

[0022] Each manufacturing task T i The service quality cannot be lower than the preset value given by the platform user, that is:

[0023]

[0024] In the formula, To create task T i service quality.

[0025] Load degree of machine tool equipment resources β u As shown below:

[0026]

[0027] In the formula, Equipping machine tools with resources u Available service hours on the Industrial Internet Service Platform.

[0028] Furthermore, the step of using the improved genetic variable neighborhood algorithm to solve the manufacturing resource scheduling model and outputting the optimal solution of the static manufacturing resource scheduling scheme includes:

[0029] 2.1) The manufacturing task number, each subtask number, process route number and machine tool equipment resource number are encoded and recorded in the gene to form the chromosome of each population individual, and the chromosome of the population individual is initialized.

[0030] 2.2) Decode and select the fitness of the initialized individuals in the population, and store the individuals with the highest fitness in the elite solution set.

[0031] 2.3) Determine whether the termination condition is met. If so, output the optimal solution of the static manufacturing resource scheduling plan. If not, proceed to step 2.4).

[0032] 2.4) Perform crossover and mutation processing on the remaining population individuals after fitness selection.

[0033] 2.5) Take each mutated offspring individual as the initial solution, construct the neighborhood structure, and obtain the optimized population individuals through the local variable neighborhood search algorithm.

[0034] 2.6) Mix the optimized population individuals with the individuals in the elite solution set and return to step 2.2).

[0035] Furthermore, the chromosomes of the individuals of each population are encoded in a uniform length.

[0036] The length unification process is as follows: Assume that the k-th process route of the j-th subtask of the i-th manufacturing task contains at most The coding length of the chromosome is If a layer code is missing, fill it with 0, where n is the total number of manufacturing tasks and h is the total number of subtasks.

[0037] Further, the steps of fitness selection are as follows:

[0038] Taking the objective function of the manufacturing resource scheduling model as the fitness function, some individuals are selected from the population for fitness comparison, and the individuals with the highest fitness are inserted into the mating pool. The iteration is repeated until the mating pool is filled, and the individuals with the top 10% fitness in the mating pool are retained in the elite solution set.

[0039] The number of the mating pools is a% of the population, where a is a positive integer.

[0040] Furthermore, the termination conditions include that the standard deviation of the fitness of the top 10% of the individuals in the elite solution set is less than 0.01 and the number of iterations is greater than or equal to a preset value.

[0041] Furthermore, the steps of performing crossover processing on the remaining population individuals after fitness selection are as follows:

[0042] 2.4.1) Define the crossover probability p c , for any two parent individuals Parent1 and Parent2, the probability p∈(0,1] is randomly generated. If p≤p c , then go to step 2.4.2), if p>p c , no cross processing is performed and jump to step 2.4.5).

[0043] The crossover probability p c As shown below:

[0044]

[0045] In the formula, is the maximum crossover probability, is the minimum crossover probability, f max is the optimal fitness in the current population. is the average fitness of the two parent individuals. avg is the average fitness of individuals in the population.

[0046] 2.4.2) Randomly select two crossover positions cross1 and cross2 from the two parent individuals Parent1 and Parent2.

[0047] 2.4.3) The gene fragments between the two crossover positions cross1 and cross2 of the parent individual Parent1 are reversely crossed and copied to the offspring individual Children1.

[0048] The gene fragments between the two crossover positions cross1 and cross2 of the parent individual Parent2 are reversely crossed and then copied to the offspring individual Children2.

[0049] 2.4.4) Delete the genes of the parent individual Parent1 after the crossover position cross2 that are repeated in the offspring individual Children2, and copy the remaining genes in the parent individual Parent1 to the offspring individual Children2 after the crossover position cross2 in sequence.

[0050] Delete the genes of the parent individual Parent2 that are repeated in the offspring individual Children1 after the crossover position cross2, and copy the remaining genes in the parent individual Parent2 to the offspring individual Children1 after the crossover position cross2 in sequence.

[0051] 2.4.5) Output the current individual.

[0052] Furthermore, the neighborhood structure includes neighborhood structure D1, neighborhood structure D2, neighborhood structure D3, and neighborhood structure D4.

[0053] The neighborhood structure D1 generates three different random numbers that are less than or equal to the chromosome length in the current solution, and arbitrarily exchanges the genes at the gene positions corresponding to these three numbers, and the resulting solution is the domain of the current solution.

[0054] The neighborhood structure D2 randomly selects several individuals from the elite solution set, randomly selects two gene positions in the gene sequence of the current solution, and the genes located at these two gene positions in the randomly selected individuals constitute the neighborhood of the current solution.

[0055] The neighborhood structure D3 randomly selects several individuals from the elite solution set, randomly selects two gene positions in the gene sequence of the current solution, and replaces the genes at the corresponding positions in the randomly selected individuals with the genes in the two gene positions of the current solution. The resulting solution is the domain of the current solution.

[0056] The neighborhood structure D4 randomly selects several individuals from the elite solution set, randomly selects two gene positions in the gene sequence of the current solution, sorts the genes of the current solution in the two gene positions in reverse order and replaces the genes in the corresponding positions in the randomly selected individuals, and the resulting solution is the domain of the current solution.

[0057] Furthermore, the steps of taking each mutated offspring individual as the initial solution, constructing a neighborhood structure, and obtaining the optimized population individual through a local variable neighborhood search algorithm are as follows:

[0058] 2.5.1) Let the initial solution be μ, and the constructed neighborhood structure is And initialize the domain number d = 1, where d max is the total number of fields.

[0059] 2.5.2) Determine whether d>d max Is it true? If so, output the solution μ. If not, go to step 2.5.3).

[0060] 2.5.3) In the solution of μ’s d-th neighborhood structure D d The solution μ' is randomly generated in .

[0061] 2.5.4) Take solution μ' as the initial solution for local search and use the local search algorithm to obtain the local optimal solution μ".

[0062] 2.5.5) If the local optimal solution μ” is better than the solution μ’, set μ=μ” and return to step 2.5.3), otherwise set d=d+1 and return to step 2.5.2).

[0063] Furthermore, the objective function of the manufacturing resource scheduling model after adjustment based on the event-driven and periodic rescheduling dynamic scheduling strategy is as follows:

[0064]

[0065] Where F' is the objective function of the manufacturing resource scheduling model after the dynamic scheduling strategy based on event-driven and periodic rescheduling is adjusted. b is the weight value. b ' is the standardized formula.

[0066] Among them, when b = 7, the standardized formula f b 'As shown below:

[0067]

[0068] Where f7 is the deviation function. w represents the production disturbance penalty coefficient, They are the start processing time and processing time of the process after adjustment by the dynamic scheduling strategy, and t is the time when the event occurs. Indicates the start time of processing on the machine tool equipment resources. Indicates the processing time of a process on machine tool equipment resources.

[0069] The technical effect of the present invention is unquestionable. The present invention realizes the generation of a dynamic scheduling solution for machine tool equipment resources for multi-task collaboration in an industrial Internet environment, and provides corresponding technical support for the networked integration and optimized operation of machine tool equipment resources in an industrial Internet environment.

[0070] The present invention establishes a multi-objective initial scheduling model including time, cost, energy consumption, load balancing, quality and reliability, designs an improved genetic variable neighborhood algorithm to solve the model, and proposes a dynamic scheduling strategy based on event-driven and periodic rescheduling, and adds a dynamic deviation indicator to the scheduling model. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 Describe the model diagram for the scheduling problem;

[0072] Figure 2 This is an example diagram of the chromosome encoding method of individuals in a population;

[0073] Figure 3 It is a flow chart of mutation operation;

[0074] Figure 4 It is a flow chart of the dynamic scheduling algorithm of machine tool equipment resources for multi-task collaboration;

[0075] Figure 5 Optimal fitness change curve;

[0076] Figure 6 Optimal scheduling Gantt chart;

[0077] Figure 7 Rescheduling fitness change curve;

[0078] Figure 8 Reschedule the Gantt chart. DETAILED DESCRIPTION

[0079] The present invention is further described below in conjunction with the embodiments, but it should not be understood that the above subject matter of the present invention is limited to the following embodiments. Without departing from the above technical ideas of the present invention, various substitutions and changes are made according to the common technical knowledge and customary means in the art, which should all be included in the protection scope of the present invention.

[0080] Embodiment 1:

[0081] See also Figures 1 to 8 , a manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment, comprising the following steps:

[0082] 1) Combined with nonlinear process planning theory, a manufacturing resource scheduling model is established.

[0083] 2) The improved genetic variable neighborhood algorithm is used to solve the manufacturing resource scheduling model and output the optimal solution for the static manufacturing resource scheduling scheme.

[0084] 3) Execute the optimal solution of the static manufacturing resource scheduling plan to determine whether the scheduling of all industrial manufacturing tasks has been completed. If so, output the optimal manufacturing resource scheduling plan. If not, proceed to step 4).

[0085] 4) Adjust the manufacturing resource scheduling model based on the dynamic scheduling strategy of event-driven and periodic rescheduling, and return the adjusted manufacturing resource scheduling model to step 2).

[0086] Embodiment 2:

[0087] A manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment. The main technical content is shown in Example 1. Further, the objective function of the manufacturing resource scheduling model is as follows:

[0088]

[0089] stf1≤T max (2)

[0090]

[0091] Where F is the objective function of the manufacturing resource scheduling model. b is the function number. b is the weight value. max The maximum delivery time allowed by the demander of the manufacturing task. is the maximum service cost. is the minimum service quality. i is the manufacturing task number, n is the total number of manufacturing tasks. k is the process route number, l is the total number of process routes. u is the machine tool equipment resource number, m is the total number of machine tool equipment resources. w is the process number, s is the total number of processes, and j is the subtask number. β u The load level of machine tool equipment resources. They respectively represent the start processing time of the wth and w+1th processes on the machine tool equipment resources. Indicates the processing time of the process on the machine tool equipment resources. ijkwu Indicates whether the process selects a machine tool from the machine tool resource set. ijk Indicates whether the subtask selects a process from the process route set.

[0092] Among them, the standardized formula f b 'As shown below:

[0093]

[0094] Among them, the service time function f1, the service cost function f2, the service quality function f3, the service reliability function f4, the service energy consumption function f5, and the resource load balancing function f6 are as follows:

[0095]

[0096] Where h is the total number of subtasks. It indicates the transportation time when two adjacent processes are processed on machine tool equipment resources in different regions. It represents the processing cost per unit time of a process on machine tool equipment resources. It represents the transportation cost when two adjacent processes are processed on machine tool equipment resources in different regions. ijkwu Indicates the process in the machine tool equipment resource R u The quality pass rate during processing. ijkwu Indicates the process in the machine tool equipment resource R u On-time completion rate during processing. u Indicates machine tool equipment resources R u The failure rate. ijkwu , P′ ijkwu They respectively represent the cutting power and no-load power of the process when it is processed on the machine tool equipment resources. They represent auxiliary processing time and basic processing time respectively.

[0097] Among them, each manufacturing task T i The service cost cannot exceed the expected value given by the platform users, that is:

[0098]

[0099] In the formula, To create task T i service cost.

[0100] Each manufacturing task T iThe service quality cannot be lower than the preset value given by the platform user, that is:

[0101]

[0102] In the formula, To create task T i service quality.

[0103] Load degree of machine tool equipment resources β u As shown below:

[0104]

[0105] In the formula, Equipping machine tools with resources u Available service hours on the Industrial Internet Service Platform.

[0106] Embodiment 3:

[0107] A manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment, the main technical content of which is shown in any one of Embodiments 1 to 2. Further, the step of using an improved genetic variable neighborhood algorithm to solve the manufacturing resource scheduling model and outputting the optimal solution of a static manufacturing resource scheduling scheme includes:

[0108] 2.1) The manufacturing task number, each subtask number, process route number and machine tool equipment resource number are encoded and recorded in the gene to form the chromosome of each population individual, and the chromosome of the population individual is initialized.

[0109] 2.2) Decode and select the fitness of the initialized individuals in the population, and store the individuals with the highest fitness in the elite solution set.

[0110] 2.3) Determine whether the termination condition is met. If so, output the optimal solution of the static manufacturing resource scheduling plan. If not, proceed to step 2.4).

[0111] 2.4) Perform crossover and mutation processing on the remaining population individuals after fitness selection.

[0112] 2.5) Take each mutated offspring individual as the initial solution, construct the neighborhood structure, and obtain the optimized population individuals through the local variable neighborhood search algorithm.

[0113] 2.6) Mix the optimized population individuals with the individuals in the elite solution set and return to step 2.2).

[0114] Embodiment 4:

[0115] A manufacturing resource scheduling method for multi-task collaboration in an industrial interconnected environment, the main technical content of which is shown in any one of Examples 1 to 3. Furthermore, the chromosome encoding of the individuals of each population is also subjected to length unification processing.

[0116] The length unification process is as follows: Assume that the k-th process route of the j-th subtask of the i-th manufacturing task contains at most The coding length of the chromosome is If a layer code is missing, fill it with 0, where n is the total number of manufacturing tasks and h is the total number of subtasks.

[0117] Embodiment 5:

[0118] A manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment, the main technical content of which is shown in any one of Embodiments 1 to 4. Further, the fitness selection step is as follows:

[0119] Taking the objective function of the manufacturing resource scheduling model as the fitness function, some individuals are selected from the population for fitness comparison, and the individuals with the highest fitness are inserted into the mating pool. The iteration is repeated until the mating pool is filled, and the individuals with the top 10% fitness in the mating pool are retained in the elite solution set.

[0120] The number of the mating pools is a% of the population, where a is a positive integer.

[0121] Embodiment 6:

[0122] A manufacturing resource scheduling method for multi-task collaboration in an industrial interconnected environment, the main technical content of which is shown in any one of Examples 1 to 5. Furthermore, the termination conditions include that the standard deviation of the individual fitness of the top 10% of the elite solution set is less than 0.01 and the number of iterations is greater than or equal to a preset value.

[0123] Embodiment 7:

[0124] A manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment, the main technical content of which is shown in any one of Embodiments 1 to 6. Further, the step of performing crossover processing on the remaining population individuals after fitness selection is as follows:

[0125] 2.4.1) Define the crossover probability p c , for any two parent individuals Parent1 and Parent2, the probability p∈(0,1] is randomly generated. If p≤p c , then go to step 2.4.2), if p>p c , no cross processing is performed and jump to step 2.4.5).

[0126] The crossover probability p c As shown below:

[0127]

[0128] In the formula, is the maximum crossover probability, is the minimum crossover probability, f max is the optimal fitness in the current population. is the average fitness of the two parent individuals. avg is the average fitness of individuals in the population.

[0129] 2.4.2) Randomly select two crossover positions cross1 and cross2 from the two parent individuals Parent1 and Parent2.

[0130] 2.4.3) The gene fragments between the two crossover positions cross1 and cross2 of the parent individual Parent1 are reversely crossed and copied to the offspring individual Children1.

[0131] The gene fragments between the two crossover positions cross1 and cross2 of the parent individual Parent2 are reversely crossed and then copied to the offspring individual Children2.

[0132] 2.4.4) Delete the genes of the parent individual Parent1 after the crossover position cross2 that are repeated in the offspring individual Children2, and copy the remaining genes in the parent individual Parent1 to the offspring individual Children2 after the crossover position cross2 in sequence.

[0133] Delete the genes of the parent individual Parent2 that are repeated in the offspring individual Children1 after the crossover position cross2, and copy the remaining genes in the parent individual Parent2 to the offspring individual Children1 after the crossover position cross2 in sequence.

[0134] 2.4.5) Output the current individual.

[0135] Embodiment 8:

[0136] A manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment, the main technical content of which is shown in any one of Examples 1 to 7. Furthermore, the neighborhood structure includes neighborhood structure D1, neighborhood structure D2, neighborhood structure D3, and neighborhood structure D4.

[0137] The neighborhood structure D1 generates three different random numbers that are less than or equal to the chromosome length in the current solution, and arbitrarily exchanges the genes at the gene positions corresponding to these three numbers, and the resulting solution is the domain of the current solution.

[0138] The neighborhood structure D2 randomly selects several individuals from the elite solution set, randomly selects two gene positions in the gene sequence of the current solution, and the genes located at these two gene positions in the randomly selected individuals constitute the neighborhood of the current solution.

[0139] The neighborhood structure D3 randomly selects several individuals from the elite solution set, randomly selects two gene positions in the gene sequence of the current solution, and replaces the genes at the corresponding positions in the randomly selected individuals with the genes in the two gene positions of the current solution. The resulting solution is the domain of the current solution.

[0140] The neighborhood structure D4 randomly selects several individuals from the elite solution set, randomly selects two gene positions in the gene sequence of the current solution, sorts the genes of the current solution in the two gene positions in reverse order and replaces the genes in the corresponding positions in the randomly selected individuals, and the resulting solution is the domain of the current solution.

[0141] Embodiment 9:

[0142] A manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment, the main technical content of which is shown in any one of embodiments 1 to 8. Further, the steps of taking each mutated offspring individual as an initial solution, constructing a neighborhood structure, and obtaining an optimized population individual through a local variable neighborhood search algorithm are as follows:

[0143] 2.5.1) Let the initial solution be μ, and the constructed neighborhood structure be {D1,D 2, …,D d ,…,Dd max}, and initialize the domain number d = 1, where d max is the total number of fields.

[0144] 2.5.2) Determine whether d>d max Is it true? If so, output the solution μ. If not, go to step 2.5.3).

[0145] 2.5.3) In the solution of μ’s d-th neighborhood structure D d The solution μ' is randomly generated in .

[0146] 2.5.4) Take solution μ' as the initial solution for local search and use the local search algorithm to obtain the local optimal solution μ".

[0147] 2.5.5) If the local optimal solution μ” is better than the solution μ’, set μ=μ” and return to step 2.5.3), otherwise set d=d+1 and return to step 2.5.2).

[0148] Embodiment 10:

[0149] A manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment, the main technical content of which is shown in any one of embodiments 1 to 9. Further, the objective function of the manufacturing resource scheduling model after the dynamic scheduling strategy based on event-driven and periodic rescheduling is adjusted is as follows:

[0150]

[0151] Where F' is the objective function of the manufacturing resource scheduling model after the dynamic scheduling strategy based on event-driven and periodic rescheduling is adjusted. b is the weight value. b ' is the standardized formula.

[0152] Among them, when b = 7, the standardized formula f b 'As shown below:

[0153]

[0154] Where f7 is the deviation function. w represents the production disturbance penalty coefficient, They are the start processing time and processing time of the process after adjustment by the dynamic scheduling strategy, and t is the time when the event occurs. Indicates the start time of processing on the machine tool equipment resources. Indicates the processing time of a process on machine tool equipment resources.

[0155] Embodiment 11:

[0156] See also Figures 1 to 8 , a manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment, comprising the following steps:

[0157] 1) Combined with nonlinear process planning theory, a manufacturing resource scheduling model is established.

[0158] 2) The improved genetic variable neighborhood algorithm is used to solve the manufacturing resource scheduling model and output the optimal solution for the static manufacturing resource scheduling scheme.

[0159] 3) Execute the optimal solution of the static manufacturing resource scheduling plan to determine whether the scheduling of all industrial manufacturing tasks has been completed. If so, output the optimal manufacturing resource scheduling plan. If not, proceed to step 4).

[0160] 4) Adjust the manufacturing resource scheduling model based on the dynamic scheduling strategy of event-driven and periodic rescheduling, and return the adjusted manufacturing resource scheduling model to step 2).

[0161] Embodiment 12:

[0162] A manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment. The main technical content is shown in Example 11. Further, the objective function of the manufacturing resource scheduling model is as follows:

[0163]

[0164] stf1≤T max (2)

[0165]

[0166] Where F is the objective function of the manufacturing resource scheduling model. b is the function number. b is the weight value. max The maximum delivery time allowed by the demander of the manufacturing task. is the maximum service cost. is the minimum service quality. i is the manufacturing task number, n is the total number of manufacturing tasks. k is the process route number, l is the total number of process routes. u is the machine tool equipment resource number, m is the total number of machine tool equipment resources. w is the process number, s is the total number of processes, and j is the subtask number. β u The load level of machine tool equipment resources. They respectively represent the start processing time of the wth and w+1th processes on the machine tool equipment resources. Indicates the processing time of the process on the machine tool equipment resources. ijkwu Indicates whether the process selects a machine tool from the machine tool resource set. ijk Indicates whether the subtask selects a process from the process route set.

[0167] Among them, the standardized formula f b 'As shown below:

[0168]

[0169] The above multi-objective function includes benefit-type indicators and cost-type indicators. It is necessary to use the linear proportional transformation method for standardization and then convert the multi-objective function into a single objective function for solution through the weighted method. The benefit-type indicators include service quality f3 and service reliability f4, and the cost-type indicators include service time, service cost, service energy consumption and resource load balance.

[0170] Among them, the service time function f1, the service cost function f2, the service quality function f3, the service reliability function f4, the service energy consumption function f5, and the resource load balancing function f6 are as follows:

[0171]

[0172] Where h is the total number of subtasks. It indicates the transportation time when two adjacent processes are processed on machine tool equipment resources in different regions. It represents the processing cost per unit time of a process on machine tool equipment resources. It represents the transportation cost when two adjacent processes are processed on machine tool equipment resources in different regions. ijkwu Indicates the process in the machine tool equipment resource R u The quality pass rate during processing. ijkwu Indicates the process in the machine tool equipment resource R u On-time completion rate during processing. u Indicates machine tool equipment resources R u The failure rate. ijkwu , P′ ijkwu They respectively represent the cutting power and no-load power of the process when it is processed on the machine tool equipment resources. They represent auxiliary processing time and basic processing time respectively.

[0173] Among them, each manufacturing task T i The service cost cannot exceed the expected value given by the platform users, that is:

[0174]

[0175] In the formula, To create task T i service cost.

[0176] Each manufacturing task T i The service quality cannot be lower than the preset value given by the platform user, that is:

[0177]

[0178] In the formula, To create task T i service quality.

[0179] Load degree of machine tool equipment resources β u As shown below:

[0180]

[0181] In the formula, Equipping machine tools with resources u Available service hours on the Industrial Internet Service Platform.

[0182] Embodiment 13:

[0183] A manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment, the main technical content of which is shown in any one of Embodiments 11 to 12. Further, the step of using an improved genetic variable neighborhood algorithm to solve the manufacturing resource scheduling model and outputting the optimal solution of the static manufacturing resource scheduling scheme includes:

[0184] 2.1) The manufacturing task number, each subtask number, process route number and machine tool equipment resource number are encoded and recorded in the gene to form the chromosome of each population individual, and the chromosome of the population individual is initialized.

[0185] The random initialization method and NEH algorithm are combined to initialize the population, where 70% of the initial population is generated by random method and 30% by NEH algorithm.

[0186] Random initialization: randomly generated The first level of chromosome encoding is performed for each manufacturing task number, and a random number is generated under the same manufacturing task number. The subtask numbers contained in the task are encoded into the corresponding gene positions on the second layer of the chromosome; at the same time, in the same column as the manufacturing task and subtask numbers, s process numbers of the selected process route and optional machine tool equipment resource numbers are randomly generated and included in the corresponding gene positions on the third and fourth layers of the chromosome respectively.

[0187] NEH algorithm initialization: 30% of the population is initialized based on the method proposed by Framinan et al. The principle is that manufacturing tasks with a small total processing time have priority processing rights.

[0188] 2.2) Decode and select the fitness of the initialized individuals in the population, and store the individuals with the highest fitness in the elite solution set.

[0189] Decoding is the process of converting chromosome codes into solutions. Scan the chromosomes from left to right. When the same code appears for the first and second layers for the first time, it indicates the first step of the subtask corresponding to the task. Figure 2 The decoding is performed by taking the encoding shown in the figure as an example, starting from the first column, taking the first and second layers of the column as the reference, and the decoded process route is P 2121 →P 2122 The corresponding machine tool equipment resources for its processing are R1→R4.

[0190] 2.3) Determine whether the termination condition is met. If so, output the optimal solution of the static manufacturing resource scheduling plan. If not, proceed to step 2.4).

[0191] 2.4) Perform crossover and mutation processing on the remaining population individuals after fitness selection.

[0192] 2.5) Take each mutated offspring individual as the initial solution, construct the neighborhood structure, and obtain the optimized population individuals through the local variable neighborhood search algorithm.

[0193] 2.6) Mix the optimized population individuals with the individuals in the elite solution set and return to step 2.2).

[0194] Embodiment 14:

[0195] A manufacturing resource scheduling method for multi-task collaboration in an industrial interconnected environment, the main technical content of which is shown in any one of Examples 11 to 13. Furthermore, the chromosome encoding of the individuals in each population is also subjected to length unification.

[0196] The length unification process is as follows: Assume that the k-th process route of the j-th subtask of the i-th manufacturing task contains at most The coding length of the chromosome is If a layer code is missing, fill it with 0, where n is the total number of manufacturing tasks and h is the total number of subtasks.

[0197] Embodiment 15:

[0198] A manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment, the main technical content of which is shown in any one of Embodiments 11 to 14. Further, the fitness selection step is as follows:

[0199] Taking the objective function of the manufacturing resource scheduling model as the fitness function, 5% of the individuals are selected from the population for fitness comparison, and the individual with the highest fitness is inserted into the mating pool. The iteration is repeated until the mating pool is filled, and the individuals with the top 10% fitness in the mating pool are retained in the elite solution set.

[0200] The number of the mating pools is 50% of the population.

[0201] Embodiment 16:

[0202] A manufacturing resource scheduling method for multi-task collaboration in an industrial interconnected environment. The main technical content is shown in any one of Examples 11 to 15. Furthermore, the termination conditions include that the standard deviation of the individual fitness of the top 10% of the elite solution set is less than 0.01 and the number of iterations is greater than or equal to a preset value.

[0203] Embodiment 17:

[0204] A manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment, the main technical content of which is shown in any one of Embodiments 11 to 16. Further, the step of performing crossover processing on the remaining population individuals after fitness selection is as follows:

[0205] 2.4.1) Define the crossover probability pc , for any two parent individuals Parent1 and Parent2, the probability p∈(0,1] is randomly generated. If p≤p c , then go to step 2.4.2), if p>p c , no cross processing is performed and jump to step 2.4.5).

[0206] In the crossover evolution process, in order to achieve a faster search speed, when the average fitness f of the two parent individuals is lower than the average fitness f of the population, avg When , we need to increase the crossover probability p c On the contrary, the crossover probability p is reduced c , that is, p c It needs to be automatically adjusted with the fitness value. To this end, the crossover probability p c As shown below:

[0207]

[0208] In the formula, is the maximum crossover probability, is the minimum crossover probability, f max is the optimal fitness in the current population. is the average fitness of the two parent individuals. avg is the average fitness of individuals in the population.

[0209] 2.4.2) Randomly select two crossover positions cross1 and cross2 from the two parent individuals Parent1 and Parent2.

[0210] 2.4.3) The gene fragments between the two crossover positions cross1 and cross2 of the parent individual Parent1 are reversely crossed and copied to the offspring individual Children1.

[0211] The gene fragments between the two crossover positions cross1 and cross2 of the parent individual Parent2 are reversely crossed and then copied to the offspring individual Children2.

[0212] 2.4.4) Delete the genes of the parent individual Parent1 after the crossover position cross2 that are repeated in the offspring individual Children2, and copy the remaining genes in the parent individual Parent1 to the offspring individual Children2 after the crossover position cross2 in sequence.

[0213] Delete the genes of the parent individual Parent2 that are repeated in the offspring individual Children1 after the crossover position cross2, and copy the remaining genes in the parent individual Parent2 to the offspring individual Children1 after the crossover position cross2 in sequence.

[0214] 2.4.5) Output the current individual.

[0215] Embodiment 18:

[0216] A manufacturing resource scheduling method for multi-task collaboration in an industrial Internet environment. The main technical content is shown in any one of Examples 11 to 17. Furthermore, the neighborhood structure includes neighborhood structure D1, neighborhood structure D2, neighborhood structure D3, and neighborhood structure D4.

[0217] The neighborhood structure D1 generates three different random numbers that are less than or equal to the chromosome length in the current solution, and arbitrarily exchanges the genes at the gene positions corresponding to these three numbers, and the resulting solution is the domain of the current solution.

[0218] The neighborhood structure D2 randomly selects several individuals from the elite solution set, randomly selects two gene positions in the gene sequence of the current solution, and the genes located at these two gene positions in the randomly selected individuals constitute the neighborhood of the current solution.

[0219] The neighborhood structure D3 randomly selects several individuals from the elite solution set, randomly selects two gene positions in the gene sequence of the current solution, and replaces the genes at the corresponding positions in the randomly selected individuals with the genes in the two gene positions of the current solution. The resulting solution is the domain of the current solution.

[0220] The neighborhood structure D4 randomly selects several individuals from the elite solution set, randomly selects two gene positions in the gene sequence of the current solution, sorts the genes of the current solution in the two gene positions in reverse order and replaces the genes in the corresponding positions in the randomly selected individuals, and the resulting solution is the domain of the current solution.

[0221] Embodiment 19:

[0222] A manufacturing resource scheduling method for multi-task collaboration in an industrial interconnected environment, the main technical content of which is shown in any one of Examples 11 to 18. Further, the steps of taking each mutated offspring individual as an initial solution, constructing a neighborhood structure, and obtaining an optimized population individual through a local variable neighborhood search algorithm are as follows:

[0223] 2.5.1) Let the initial solution be μ, and the constructed neighborhood structure is And initialize the domain number d = 1, where d max is the total number of fields.

[0224] 2.5.2) Determine whether d>dmax Is it true? If so, output the solution μ. If not, go to step 2.5.3).

[0225] 2.5.3) In the solution of μ’s d-th neighborhood structure D d The solution μ' is randomly generated in .

[0226] 2.5.4) Take solution μ' as the initial solution for local search and use the local search algorithm to obtain the local optimal solution μ".

[0227] 2.5.5) If the local optimal solution μ” is better than the solution μ’, set μ=μ” and return to step 2.5.3), otherwise set d=d+1 and return to step 2.5.2).

[0228] Embodiment 20:

[0229] A manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment, the main technical content of which is shown in any one of Embodiments 11 to 19. Further, the objective function of the manufacturing resource scheduling model after the dynamic scheduling strategy based on event-driven and periodic rescheduling is adjusted is as follows:

[0230]

[0231] Where F' is the objective function of the manufacturing resource scheduling model after the dynamic scheduling strategy based on event-driven and periodic rescheduling is adjusted. b is the weight value. b ' is the standardized formula.

[0232] Among them, when b = 7, the standardized formula f b 'As shown below:

[0233]

[0234] Where f7 is the deviation function. w represents the production disturbance penalty coefficient, They are the start processing time and processing time of the process after adjustment by the dynamic scheduling strategy, and t is the time when the event occurs. Indicates the start time of processing on the machine tool equipment resources. Indicates the processing time of a process on machine tool equipment resources.

[0235] Embodiment 21:

[0236] See also Figures 1 to 8 , a manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment, comprising the following steps:

[0237] 1. Problem description:

[0238] Table 1 Model parameter definitions

[0239]

[0240] The industrial Internet model supports the networked collaborative operation of multiple manufacturing tasks. According to the nonlinear process planning theory, the machine tool equipment resource scheduling problem can be described as follows: there are n manufacturing tasks {T1,…,T i ,…,T n}Cooperate on the industrial Internet service platform, each manufacturing task T i Can be decomposed into h subtasks {T i1 ,…,T ij ,…,T ih}, any subtask T ij There are l optional process routes {P ij1 ,…,P ijk ,…,P ijl}, each process route P ijk By s process {O ijk1 ,…,O ijkw ,…,O ijks} (determined according to the change of processing method), each process can be composed of m machine tool equipment resources {R1,…,R u ,…,R m} for processing, such as Figure 1 Other parameter definitions are shown in Table 1:

[0241] In addition, the following constraints must be met during the execution of industrial Internet tasks:

[0242] (1) The process route being executed cannot be changed after it is selected;

[0243] (2) There are timing constraints between the processes of the same subtask, but there are no timing constraints between the processes of different subtasks;

[0244] (3) The same machine tool equipment resource can only process one process at a time;

[0245] (4) All subtasks can be processed at time zero, and once the process is in progress, it is not allowed to be interrupted (unless the machine tool fails);

[0246] (5) If a machine tool equipment resource failure occurs during the processing of the process, the process will be reprocessed;

[0247] (6) The repair time after a machine tool equipment resource fails can be calculated;

[0248] (7) The transportation cost and time between machine tool equipment resources in different regions are proportional to the distance.

[0249] 2. Model construction: While incorporating traditional economic indicators such as service time, service cost, and service quality into the objective function, the performance indicators that affect the reliable and stable operation of industrial Internet services (including service reliability, service energy consumption, and load balancing of machine tools and equipment) are also considered uniformly. The initial scheduling model is as follows:

[0250] (1) Service time refers to the time it takes for a manufacturing service demand to be submitted through the Industrial Internet service platform and to obtain a response from a matching machine tool equipment resource to complete the corresponding processing task. The service time is determined by the longest subtask, which mainly includes task processing time and logistics transportation time. Its objective function is:

[0251]

[0252] in That is, the processing time of each process is the sum of the auxiliary processing time and the basic processing time.

[0253] (2) Service cost refers to the total cost of the selected machine tool equipment resources to complete the corresponding manufacturing task, mainly including processing cost and logistics transportation cost. Its objective function is:

[0254]

[0255] Among them, each manufacturing task T i The service cost cannot exceed the expected value given by the platform users, and the calculation formula is:

[0256]

[0257] (3) Service quality refers to the qualified rate of the processing procedures corresponding to the completion of the manufacturing tasks by the machine tool equipment resources. It can be statistically quantified based on the service history of the machine tool equipment resources. Its objective function is:

[0258]

[0259] Among them, each manufacturing task T i The service quality cannot be lower than the preset value given by the platform user, and the calculation formula is:

[0260]

[0261] (4) Service reliability is mainly characterized by the on-time completion rate of all processing procedures of the machine tool equipment resources to complete the manufacturing task and the failure rate of the machine tool equipment resources. Among them, the on-time completion rate can be obtained by statistical regression analysis based on the estimated completion time and actual completion time given by the machine tool equipment resources to complete similar manufacturing tasks, and the failure rate can be statistically quantified through the historical service records of the machine tool equipment. Its objective function is:

[0262]

[0263] (5) Service energy consumption mainly includes the processing energy consumption and no-load energy consumption generated by machine tool equipment resources in the process of executing manufacturing tasks. Its objective function can be expressed as:

[0264]

[0265] (6) Resource load balancing When performing unified resource scheduling for multiple manufacturing tasks on the industrial Internet service platform, it is necessary to comprehensively consider the load balancing of each candidate machine tool equipment resource to minimize the load standard deviation. The objective function is:

[0266]

[0267] The above multi-objective function includes benefit-type indicators and cost-type indicators. It is necessary to use the linear proportional transformation method for standardization and then convert the multi-objective function into a single objective function for solution through the weighted method. The benefit-type indicators include service quality f3 and service reliability f4, and the cost-type indicators include service time, service cost, service energy consumption and resource load balance. The standardized formula is:

[0268]

[0269] In summary, the final objective function is:

[0270]

[0271] stf1≤T max (13)

[0272]

[0273] in is the weight value of each objective function, which can be solved based on the extension analytic hierarchy process (EAHP); T max , They are the maximum delivery time, maximum service cost and minimum service quality allowed by the manufacturing task demander. In the actual processing process, the load of machine tool equipment resources generally cannot exceed 80%, so β u ≤80%.

[0274] 3. Model solving: The genetic algorithm is improved, and a dynamic scheduling method based on the improved genetic variable neighborhood algorithm is designed to solve the above model.

[0275] 3.1 Improved genetic variable neighborhood algorithm

[0276] (1) Encoding and decoding: According to the operation characteristics of industrial Internet services and the requirements of the constructed model, chromosomes need to be encoded in layers, that is, the manufacturing task number, subtask number, processing procedure and machine tool equipment resource number information are encoded and recorded on the gene to form the chromosomes of each population individual. The process information contained can be expressed as (i, j, w, u). Since different process routes have different numbers of processing procedures, in order to ensure the feasibility of operations such as crossover and mutation during the algorithm solution process, the chromosome lengths of the population individuals are unified. Suppose the kth process route of the jth subtask of the i-th manufacturing task contains at most The coding length of the chromosome is Then the various numbers of the manufacturing task coding layer and subtask coding layer should appear and If a layer of code is missing, it needs to be filled with 0. Figure 2 shown.

[0277] Decoding is the process of converting chromosome codes into solutions. Scan the chromosomes from left to right. When the same code appears for the first and second layers for the first time, it indicates the first step of the subtask corresponding to the task. Figure 2 The decoding is performed by taking the encoding shown in the figure as an example, starting from the first column, taking the first and second layers of the column as the reference, and the decoded process route is P 2121 →P 2122 The corresponding machine tool equipment resources for its processing are R1→R4.

[0278] (2) Population initialization: The quality of the population's initial solution has a great impact on the speed and effect of the algorithm's solution. The quality of the final solution obtained by randomly initializing the population is usually poor, which requires increasing the number of algorithm iterations or expanding the population size to obtain the optimal solution, increasing the amount of calculation. The random initialization method and the NEH algorithm are combined to initialize the population. 70% of the initial population is generated by random methods, and 30% is generated by the NEH algorithm.

[0279] Random initialization: randomly generated The first level of chromosome encoding is performed for each manufacturing task number, and a random number is generated under the same manufacturing task number. The subtask numbers contained in the task are encoded into the corresponding gene positions on the second layer of the chromosome; at the same time, in the same column as the manufacturing task and subtask numbers, s process numbers of the selected process route and optional machine tool equipment resource numbers are randomly generated and included in the corresponding gene positions on the third and fourth layers of the chromosome respectively.

[0280] NEH algorithm initialization: 30% of the population is initialized based on the method proposed by Framinan et al. The principle is that manufacturing tasks with a small total processing time have priority processing rights.

[0281] (3) Fitness function: The fitness function is used to evaluate chromosomes. This paper selects the objective function as the fitness function.

[0282] (4) Selection operation: A combination of tournament and elite strategies is used for selection operation. First, 5% of individuals are selected from the initial population for fitness comparison, and the individuals with the highest fitness are inserted into the mating pool (the number of individuals in the mating pool is 50% of the population). This cycle is repeated until the mating pool is filled, and the top 10% of individuals in the mating pool are retained in the elite solution set.

[0283] (5) Crossover operation: The crossover operator is the main operator to ensure the global search capability of the genetic algorithm. The specific operation steps are as follows:

[0284] 5.1) Randomly generate p∈(0,1], if p≤p c (crossover probability), then go to 5.2)~5.4), otherwise go to 5.5);

[0285] 5.2) Randomly select two crossover positions cross1 and cross2 from the two parents Parent1 and Parent2.

[0286] 5.3) The gene fragment between the two crossover positions of Parent1 is copied to Children1 in reverse order, and the same is true for the other one;

[0287] 5.4) Starting from the second crossover point cross2, delete the genes in Parent1 that are duplicated with Children2, and then start from the second crossover point cross2 to copy the remaining genes in Parent1 to Children2 one by one, and the same goes for the other;

[0288] 5.5)End.

[0289] In the crossover evolution process, in order to achieve a faster search speed, when the average fitness of the two parent individuals Lower than the population average fitness f avg When , we need to increase the crossover probability p c On the contrary, the crossover probability p is reduced c , that is, p c It needs to be automatically adjusted with the fitness value. To this end, this paper uses formula (18) to calculate the adaptive crossover probability:

[0290]

[0291] In the formula, is the maximum crossover probability, is the minimum crossover probability, f max is the optimal fitness in the current population.

[0292] (6) Mutation operation: The mutation operator is an auxiliary operator that ensures the local search capability of the genetic algorithm. The mutation operation steps are as follows: Figure 3 shown.

[0293] (7) Variable neighborhood structure design: Four types of neighborhood structures are used for local search, represented by D1, D2, D3, and D4 respectively.

[0294] 7.1) Neighborhood structure D1 refers to randomly generating three different random numbers that are less than or equal to the chromosome length in the current solution, and arbitrarily exchanging the genes at the gene positions corresponding to these three numbers. The new solution generated is the neighborhood of the current solution.

[0295] 7.2) Neighborhood structure D2 draws on the idea of ​​crossover operation in genetic algorithms, randomly selects x individuals from the elite solution, and randomly selects two gene positions in the gene sequence of the current solution. The genes located at the two positions in the x individuals constitute the neighborhood of the current solution.

[0296] 7.3) Neighborhood structure D3 draws on the idea of ​​crossover operation in genetic algorithms, randomly selects x individuals from the elite solution, randomly selects two gene positions in the gene sequence of the current solution, takes the middle part as a whole, and replaces the corresponding positions in the elite solution with the selected genes in the current solution. The resulting new solution is the neighborhood of the current solution.

[0297] 7.4) Neighborhood structure D4 draws on the idea of ​​mutation operation in genetic algorithms, randomly selects x individuals from the elite solution, randomly selects two gene positions in the gene sequence of the current solution, and inserts the genes between these two positions into the corresponding positions in the elite solution in reverse order. The resulting new solution is the neighborhood of the current solution.

[0298] (8) Algorithm solution steps: Step 1 Determine the parameters: population size NS, maximum number of iterations TS, crossover probability p c , mutation probability p m , neighborhood structure d max ; Step 2 chromosome encoding, generate the initial population through random initialization and NEH algorithm; Step 3 decode each individual, calculate the individual fitness function, select individuals using the tournament method, and retain the top 10% of individuals in the elite solution set; Step 4 if the termination condition is met (the fitness standard deviation of 10% elite individuals is less than 0.01 or the number of iterations ts ≥ TS), stop and output the optimal solution; otherwise, go to step 5; Step 5 crossover operation; Step 6 mutation operation; Step 7 use each mutated offspring as the initial solution μ of the local variable neighborhood search algorithm to construct the neighborhood structure Step 8: Set d = 1 and repeat the following steps until d = d max ; Step 8.1 In the d-th neighborhood structure D of μd Randomly generate μ'(μ'∈D d ); Step 8.2 Take μ' as the local search initial solution and use the local search algorithm to obtain the local optimal solution μ"; Step 8.3 If μ" is better than μ', set μ = μ", and continue to search in the neighborhood structure D d Inner search; otherwise d = d + 1. Step 9: Mix the optimized individual population with the initial elite solution and select according to the strategy; Step 10: Determine the termination condition, if it is satisfied, output the optimal solution; otherwise, go to step 2.

[0299] 3.2 Dynamic Scheduling Method

[0300] (1) Determination of dynamic scheduling strategy: The industrial Internet environment has a high degree of freedom and dynamic uncertainty. During the operation of cloud services, machine tool equipment resources are easily affected by random and high-frequency production disturbances such as emergency orders, equipment failures, and quality abnormalities, making it difficult for the initial resource scheduling scheme to support the reliable and stable operation of industrial Internet tasks and their services. To this end, a dynamic scheduling strategy based on event-driven and periodic rescheduling is proposed to ensure the steady-state execution of the industrial Internet resource scheduling scheme. In the absence of any production disturbance, a periodic rescheduling strategy is usually adopted. Its rescheduling period ΔT can be determined according to the amount of manufacturing tasks in the industrial Internet platform, the degree of machine tool equipment resource load, etc. Generally, ΔT selects the minimum processing time period of the manufacturing task to be executed, so as to better ensure the rationality of scheduling and reduce the platform computing cost. However, if relevant production disturbances occur in the industrial Internet environment, and the disturbance causes the maximum completion time of the manufacturing task to change (increase or decrease), the event-driven dynamic scheduling strategy is triggered.

[0301] (2) Dynamic scheduling objective function: In the dynamic scheduling process, in addition to considering the objective function established above, it is also necessary to consider the degree of deviation of the rescheduling from the initial scheduling plan to minimize the impact on other manufacturing task scheduling plans on the Industrial Internet service platform. The deviation degree is the deviation degree of the machine tool equipment resources relative to the selected processing technology, including the deviation degree of the start processing time and the deviation degree of the processing time. Its objective function can be expressed as

[0302]

[0303] In the formula, p w represents the production disturbance penalty coefficient, are the start processing time and processing time of the rescheduling plan, and t is the time when the event occurs. The index is standardized as follows: Therefore, the rescheduling objective function is

[0304]

[0305] (3) Dynamic scheduling algorithm: Dynamic scheduling involves the change of encoding in the improved genetic variable neighborhood algorithm and the adjustment of the fitness function. Before executing dynamic rescheduling, some manufacturing tasks’ related processes have been completed, or some machine tool equipment resources are unavailable, so their chromosome encoding should be adjusted accordingly. The rescheduling fitness function should consider the deviation index, that is, use the rescheduling objective function (20) as the corresponding fitness function. In summary, the dynamic scheduling algorithm process of machine tool equipment resources for multi-task collaboration is as follows: Figure 4 shown.

[0306] Embodiment 22:

[0307] A manufacturing resource scheduling method for multi-task collaboration in an industrial Internet environment, the main technical content is shown in Example 21. Furthermore, the automatically decomposed subtasks, process routes, process information, optional machine tool equipment resources, basic processing time, auxiliary processing time, cutting power, no-load power and other information are shown in Table 2; the material transportation time between machine tool equipment resources in different regions is shown in Table 3, and the unit time transportation cost is set to 100 yuan; the service requirements of each manufacturing task are shown in Table 4.

[0308] Table 2 Task related information

[0309]

[0310] The parameters involved in the algorithm are set as follows: population size NS = 100, maximum number of iterations TS = 200, crossover probability P c =0.8, and Mutation probability P m =0.1, maximum search times NB = 20, maximum loop number TB = 50. The algorithm is implemented using MATLAB 2016a, the operating environment is CPU 2.8GHz, RAM 8GB, and the operating system is Windows 10. Figure 5 The fitness convergence trend shown in Figure 6 The optimal scheduling Gantt chart is shown.

[0311] Table 3 Transportation time between machine tool equipment resources in different regions

[0312]

[0313] Table 4 Service requirements for tasks

[0314]

[0315] Table 5 Optimal results of initial scheduling

[0316]

[0317] Table 6 Rescheduling results

[0318]

[0319] Depend on Figure 5 , Figure 6 It can be seen that a stable optimal scheduling solution is obtained after about 120 iterations, and the optimal process route is P 113 , P 121 , P 211 , P 221 , P 231 ,The results of each optimal index are shown in Table 5. From Table 5, we can see that each index meets the user requirements and the results are relatively good.

[0320] For the dynamic scheduling problem driven by production disturbance events during the operation of manufacturing tasks, it is assumed that when the service time is 38h, the machine tool equipment resource R4 is unavailable in the next 15h, which increases the maximum completion time of the task, thereby triggering the rescheduling program, and obtaining the new fitness function value convergence trend and scheduling Gantt chart, as shown in Figure 7 , Figure 8 The results of various indicators are shown in Table 6.

[0321] It can be seen from the rescheduling Gantt chart that the completion time of the manufacturing task increased by 3 hours, the corresponding service cost and service energy consumption increased, the service quality and reliability decreased slightly, but the service balance was better than before, and the dynamic deviation was 0.2083.

Claims

1. A manufacturing resource scheduling method for multi-task collaboration in an industrial Internet environment, characterized in that: The following steps are involved: 1) Combined with nonlinear process planning theory, a manufacturing resource scheduling model is established. 2) The improved genetic variable neighborhood algorithm is used to solve the manufacturing resource scheduling model and output the optimal solution for the static manufacturing resource scheduling scheme. 3) Execute the optimal solution of the static manufacturing resource scheduling scheme to determine whether the scheduling of all industrial manufacturing tasks has been completed. If so, output the optimal manufacturing resource scheduling scheme. If not, proceed to step 4); 4) Adjust the manufacturing resource scheduling model based on the dynamic scheduling strategy of event-driven and periodic rescheduling, and return the adjusted manufacturing resource scheduling model to step 2).

2. According to the method of manufacturing resource scheduling for multi-task collaboration in an industrial interconnection environment in claim 1, it is characterized in that: The objective function of the manufacturing resource scheduling model is as follows: s.t.f1≤T max (2) Where F is the objective function of the manufacturing resource scheduling model; b is the function number; λ b is the weight value; T max The maximum delivery time allowed by the party requesting the manufacturing task; is the maximum service cost; is the minimum service quality; i is the manufacturing task number, n is the total number of manufacturing tasks; k is the process route number, l is the total number of process routes; u is the machine tool equipment resource number, m is the total number of machine tool equipment resources; w is the process number, s is the total number of processes, and j is the subtask number; β u Loading degree of machine tool equipment resources; They represent the start processing time of the wth and w+1th processes on the machine tool equipment resources respectively; Indicates the processing time of the process on the machine tool equipment resources; Y ijkwu Indicates whether the process selects a certain machine tool from the machine tool resource set; X ijk Indicates whether the subtask selects a process from the process route set; Among them, the standardized formula f b 'As shown below: Among them, the service time function f1, the service cost function f2, the service quality function f3, the service reliability function f4, the service energy consumption function f5, and the resource load balancing function f6 are as follows: Where h is the total number of subtasks; It indicates the transportation time when two adjacent processes are processed on machine tool equipment resources in different regions; It indicates the unit time processing cost of the process on the machine tool equipment resources; represents the transportation cost of two adjacent processes when they are processed on machine tool equipment resources in different regions; ijkwu Indicates the process in the machine tool equipment resource R u Quality pass rate during processing; u ijkwu Indicates the process in the machine tool equipment resource R u On-time completion rate during processing; F u Indicates machine tool equipment resources R u The failure rate of ijkwu , P' ijkwu They respectively represent the cutting power and no-load power of the process when it is processed on the machine tool equipment resources; They represent auxiliary processing time and basic processing time respectively; Among them, each manufacturing task T i The service cost cannot exceed the expected value given by the platform users, that is: In the formula, To create task T i Cost of services; Each manufacturing task T i The service quality cannot be lower than the preset value given by the platform user, that is: In the formula, To create task T i service quality; Load degree of machine tool equipment resources β u As shown below: In the formula, Equipping machine tools with resources u Available service hours on the Industrial Internet Service Platform.

3. According to the method of manufacturing resource scheduling for multi-task collaboration in an industrial interconnection environment in claim 1, it is characterized in that: The step of using the improved genetic variable neighborhood algorithm to solve the manufacturing resource scheduling model and outputting the optimal solution of the static manufacturing resource scheduling scheme includes: 2.1) Encode the manufacturing task number, each subtask number, process route number and machine tool equipment resource number information on the gene to form the chromosome of each population individual, and initialize the chromosome of the population individual; 2.2) Decode and select the fitness of the initialized individuals in the population, and store the individuals with the highest fitness in the elite solution set; 2.3) Determine whether the termination condition is met. If so, output the optimal solution of the static manufacturing resource scheduling scheme. If not, proceed to step 2.4); 2.4) Perform crossover and mutation processing on the remaining population individuals after fitness selection; 2.5) Take each mutated offspring individual as the initial solution, construct the neighborhood structure, and obtain the optimized population individuals through the local variable neighborhood search algorithm; 2.6) Mix the optimized population individuals with the individuals in the elite solution set and return to step 2.2).

4. The manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment according to claim 3 is characterized in that: The chromosomes of the individuals of each population are encoded to be uniform in length; The length unification process is as follows: Assume that the k-th process route of the j-th subtask of the i-th manufacturing task contains at most The coding length of the chromosome is If a layer code is missing, fill it with 0, where n is the total number of manufacturing tasks and h is the total number of subtasks.

5. The manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment according to claim 3 is characterized in that: The steps of fitness selection are as follows: Taking the objective function of the manufacturing resource scheduling model as the fitness function, select some individuals from the population for fitness comparison, insert the individuals with the highest fitness into the mating pool, and iterate repeatedly until the mating pool is filled, and retain the top 10% of the individuals in the mating pool into the elite solution set; The number of the mating pools is a% of the population, where a is a positive integer.

6. The manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment according to claim 3 is characterized in that: The termination conditions include that the standard deviation of the fitness of the first 10% of the individuals in the elite solution set is less than 0.01 and the number of iterations is greater than or equal to a preset value.

7. The manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment according to claim 3 is characterized in that: The steps of crossover processing the remaining population individuals after fitness selection are as follows: 2.4.1) Define the crossover probability p c , for any two parent individuals Parent1 and Parent2, the probability p∈(0,1] is randomly generated. If p≤p c , then go to step 2.4.2), if p>p c , no cross processing is performed and jump to step 2.4.5); The crossover probability p c As shown below: In the formula, is the maximum crossover probability, is the minimum crossover probability, f max is the optimal fitness in the current population; is the average fitness of the two parent individuals; f avg is the average fitness of individuals in the population; 2.4.2) Randomly select two crossover positions cross1 and cross2 from the two parent individuals Parent1 and Parent2; 2.4.3) The gene fragments between the two crossover positions cross1 and cross2 of the parent individual Parent1 are reversely crossed and then copied to the offspring individual Children1; The gene fragments between the two crossover positions cross1 and cross2 of the parent individual Parent2 are crossed in reverse order and then copied to the offspring individual Children2; 2.4.4) Delete the genes of the parent individual Parent1 that are repeated in the offspring individual Children2 after the crossover position cross2, and copy the remaining genes in the parent individual Parent1 to the offspring individual Children2 after the crossover position cross2 in sequence; Delete the genes of the parent individual Parent2 that are repeated in the offspring individual Children1 after the crossover position cross2, and copy the remaining genes in the parent individual Parent2 to the offspring individual Children1 after the crossover position cross2 in sequence; 2.4.5) Output the current individual.

8. The manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment according to claim 3 is characterized in that: The neighborhood structure includes neighborhood structure D1, neighborhood structure D2, neighborhood structure D3, and neighborhood structure D4; The neighborhood structure D1 generates three different random numbers less than or equal to the chromosome length in the current solution, and arbitrarily exchanges the genes at the gene positions corresponding to these three numbers, and the resulting solution is the domain of the current solution; The neighborhood structure D2 is to randomly select several individuals from the elite solution set, randomly select two gene positions in the gene sequence of the current solution, and the genes located at these two gene positions in the randomly selected individuals constitute the neighborhood of the current solution; The neighborhood structure D3 randomly selects several individuals from the elite solution set, randomly selects two gene positions in the gene sequence of the current solution, and replaces the genes in the corresponding positions of the randomly selected individuals with the genes in the two gene positions of the current solution. The resulting solution is the domain of the current solution. The neighborhood structure D4 randomly selects several individuals from the elite solution set, randomly selects two gene positions in the gene sequence of the current solution, sorts the genes of the current solution in the two gene positions in reverse order and replaces the genes in the corresponding positions in the randomly selected individuals, and the resulting solution is the domain of the current solution.

9. The manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment according to claim 3 is characterized in that: The steps of taking each mutated offspring individual as the initial solution, constructing the neighborhood structure, and obtaining the optimized population individual through the local variable neighborhood search algorithm are as follows: 2.5.1) Let the initial solution be μ, and the constructed neighborhood structure is And initialize the domain number d = 1, where d max is the total number of fields; 2.5.2) Determine whether d>d max Is it true? If so, output the solution μ. If not, go to step 2.5.3); 2.5.3) In the solution of μ’s d-th neighborhood structure D d The solution μ' is randomly generated in 2.5.4) Taking solution μ' as the initial solution of local search, apply local search algorithm to obtain the local optimal solution μ"; 2.5.5) If the local optimal solution μ” is better than the solution μ’, set μ=μ” and return to step 2.5.3), otherwise set d=d+1 and return to step 2.5.2).

10. The manufacturing resource scheduling method for multi-task collaboration in an industrial interconnection environment according to claim 1, characterized in that: The objective function of the manufacturing resource scheduling model after adjustment based on the event-driven and periodic rescheduling dynamic scheduling strategy is as follows: Where F' is the objective function of the manufacturing resource scheduling model after the dynamic scheduling strategy based on event-driven and periodic rescheduling is adjusted; λ b is the weight value; f b ' is the standardized formula; Among them, when b = 7, the standardized formula f b 'As shown below: Where f7 is the deviation function; p w represents the production disturbance penalty coefficient, are the start processing time and processing duration of the process after the dynamic scheduling strategy is adjusted, and t is the time when the event occurs; Indicates the start time of the process on the machine tool equipment resources; Indicates the processing time of a process on machine tool equipment resources.

Images