Multi-variety, small-batch, multi-resource scheduling system and method based on digital twins

Abstract

The present invention discloses a multi-variety, small-batch, multi-resource scheduling system and method based on digital twins. The order optimization module splits the order and combines similar batches according to process similarity, and passes it to the scheduling optimization module as the basic task of production; the workshop data perception and disturbance event detection module, based on the workshop disturbance detection method of perception data fusion, realizes the real-time acquisition of the workshop's "man-machine-object-method-environment" state information, and passes the acquired information to the disturbance event processing module for disturbance factor judgment, and passes the disturbance event to the scheduling optimization module to realize the dynamic evolution of production; the scheduling optimization module receives and processes the information data transmitted by other modules to calculate the target scheduling plan. It not only solves the problem of many product varieties and small batches in the order, but also solves the problem of multi-target and multi-resource scheduling and allocation in the workshop, and improves the flexibility and robustness of the workshop for emergencies and the effectiveness of the scheduling plan.

Description

Technical Field

[0001] The present invention belongs to the technical field of digital twins, and specifically relates to a multi-variety, small-batch, and multi-resource scheduling method based on digital twins. Background Art

[0002] Faced with the wave of economic globalization, the manufacturing industry has undergone major changes in recent years, and the customer-oriented "customized, according to order" service model has become the development direction of manufacturing enterprises. In order to survive in such a competitive environment, enterprises must quickly adapt to the rapid changes in market demand by increasing operational flexibility, improving survival rate, shortening product cycles and enhancing fault tolerance. Due to the "customized, according to order" policy, the workshop produces a large variety of products, small production batches, and frequently switches production equipment and production routes, which not only increases the processing preparation time, but also reduces the stability of the workshop, increases the frequency of equipment failures, and reduces the yield rate. Therefore, it is urgent to effectively solve the current production status of "multi-variety and small batch".

[0003] Scheduling is a key and effective means in manufacturing. It is an optimization method that effectively utilizes the existing equipment, personnel, materials and other resources in the workshop to efficiently produce products. Traditional scheduling plans only consider a single resource (processing equipment) or two resources (processing equipment and operators). However, in the current workshop, processing equipment, operating staff and transport carts coexist. Comprehensively allocating these resources to minimize production costs and maximize production efficiency is a problem that needs to be solved urgently.

[0004] Secondly, the "customized, according to order" service model makes the dynamic nature of the workshop production process more complex and diverse. In a dynamic environment, the original scheduling plan may degenerate or even become unfeasible due to missed appropriate dates and deadlines, idle resources, high work-in-process inventory, and frequent rescheduling, which leads to increased costs. How to obtain disturbance events in a dynamic environment in real time and effectively schedule the evolution of the production process is also a pain point in workshop production planning.

[0005] Digital twins make full use of data such as physical models, sensor updates, and operation history, integrate multi-disciplinary, multi-physical, multi-scale, and multi-probability simulation processes, and complete mapping in virtual space to reflect the entire life cycle of the corresponding physical equipment. At the same time, it also has virtual-real interaction functions, can be updated and dynamically evolved in real time, thereby achieving evolutionary control and optimization of the physical world.

[0006] Currently, no effective solution has been proposed for the application of digital twin technology in the multi-resource scheduling of multiple varieties and small batches.

[0007] For example, the order problem considered in Chinese patent CN201911135963.1: An order scheduling and allocation dispatching method and system is how to achieve continuous production of the same order and the same model of products without interruption under reverse scheduling. A production line balancing algorithm with weight as the core is implemented, which not only maintains the basic requirements of reverse production scheduling according to delivery date but also meets the continuity requirements required for production. However, how to transform "multiple varieties and small batches" into "few varieties and large batches" cannot be solved. It is impossible to achieve unified batch production of products with different orders but similar processes.

[0008] Chinese Patent CN202011337391.8: A digital twin intelligent cloud scheduling method that meets personalized customized production. It formulates rules for pre-scheduling of orders before order execution to calculate order weights, clarifies the indicators that affect orders, and is a dynamic scheduling model based on two dynamic events: "arrival of new personalized orders" and "urgent and important customer orders". It considers the order priority production problem. It is a rescheduling under "arrival of personalized orders" and "urgent and important customer orders". It is a passive scheduling under the occurrence of disturbance events, and a periodic rescheduling without urgent orders. It is impossible to achieve accurate prediction of active evolution scheduling for the generated disturbance factors. Summary of the invention

[0009] In order to solve the technical problems existing in the above-mentioned related technologies to a certain extent, the purpose of the present invention is to provide a multi-variety, small-batch, multi-resource scheduling method based on digital twins, to solve the problem of many product varieties and small batches in orders, and to solve the problem of multi-objective and multi-resource scheduling and allocation in workshops. By combining digital twin technology, the workshop status can be perceived in real time, and evolutionary scheduling of disturbance events can be performed, which enhances the management and control of the actual workshop, and improves the flexibility and robustness of the workshop for emergencies and the effectiveness of the scheduling plan.

[0010] In order to achieve the above object, the present invention adopts the following technical solutions:

[0011] A multi-variety, small-batch, multi-resource scheduling system based on digital twins, characterized by:

[0012] It at least includes an order optimization processing module, a workshop data perception and disturbance event detection module, and a scheduling optimization processing module:

[0013] The order optimization processing module uses the order process similarity classification method based on fuzzy clustering analysis to split customer orders according to industry standard classification, obtain a total set of different products with the smallest unit, perform fuzzy clustering classification calculations on the products in the total set according to process similarity, merge products with the same similarity into similar batches, and use all similar batch products as the basic production targets for batch scheduling;

[0014] Workshop data perception and disturbance event detection module, a workshop disturbance event detection method based on perception data fusion, which realizes the real-time acquisition of the status information of "man-machine-material-method-environment" in the workshop, and transmits the acquired information to the disturbance event processing module for disturbance factor judgment, and processes the workshop status data in real time, including the judgment and classification of workshop equipment abnormalities, personnel leaving their posts, material shortages, processing method errors and safety accidents, and transmits the disturbance events to the scheduling optimization processing to achieve dynamic evolution scheduling of production;

[0015] Scheduling optimization processing module, including an initialization pre-scheduling module and a dynamic evolution module. The initialization pre-scheduling module is used to solve the multi-objective and multi-resource scheduling allocation problem, with minimizing the makespan and minimizing the production cost as the processing objectives. Considering the dependencies of processing equipment, personnel, and handling carts comprehensively, determine the processing sequence of N types of similar batch workpieces on M devices, the processing allocation of S workers on M devices, and the allocation of W handling carts on M devices, where W, S < M, and use the Grey Wolf Optimization Algorithm to solve the scheduling problem; the dynamic evolution module realizes effective evolution scheduling optimization for different disturbance events.

[0016] A multi-resource scheduling method for multi-variety and small-batch based on digital twin, characterized by including order optimization processing, workshop data perception and disturbance event detection, and scheduling optimization processing steps:

[0017] S1: Order optimization processing, based on the order process similarity classification method of fuzzy clustering analysis, split the customer orders according to the industry standard classification to obtain the total set of different products with the smallest units, perform fuzzy clustering classification calculation on the products in the total set according to the process similarity, merge the products with the same similarity into similar batches, and use all the similar batch products as the basic production objectives for batch scheduling;

[0018] S2: Workshop data perception and disturbance event detection, a workshop disturbance event detection method based on perception data fusion, which realizes the real-time acquisition of the status information of "man-machine-material-method-environment" in the workshop, and transmits the acquired information to the disturbance event processing module for disturbance factor judgment, and processes the workshop status data in real time, including the judgment and classification of workshop equipment abnormalities, personnel leaving their posts, material shortages, processing method errors and safety accidents, and transmits the disturbance events to the scheduling optimization processing to achieve dynamic evolution scheduling of production;

[0019] S3: Scheduling optimization processing, including the initialization pre-scheduling stage and the dynamic evolution stage; the initialization pre-scheduling model stage is used to realize the multi-objective and multi-resource scheduling and allocation problem, with minimizing the maximum completion time and minimizing the production cost as the processing goals, comprehensively considering the dependence of processing equipment, personnel and transport carts, and determining the processing sequence of N similar batches of workpieces on M equipment, the processing allocation of S workers on M equipment, and the allocation of W transport carts on M equipment, where W, S<M, and using the gray wolf optimization algorithm to solve the scheduling problem; the dynamic evolution stage realizes effective evolutionary scheduling optimization for different disturbance events.

[0020] In the above technical solution, in the S1 order optimization processing step, the order splitting is to split B customer orders according to the industry standard categories of products to obtain a total set P containing n products, where P i For the i-th category product.

[0021] In the above technical solution, in the S1 order optimization processing step, the fuzzy analysis method is used to split the order, and the product process similarity analysis and calculation are performed on the products in the total set P containing n kinds of products, and the products with the same process similarity are merged into similar batches to form a product set EP containing K types of similar batches, where EP k is the kth batch of similar products, where K≤n, and the fuzzy clustering analysis steps are:

[0022] S11: The basic data matrix is ​​composed of product processes m is the product process number;

[0023] S12: Use maximum normalization: for the feature index matrix X * The j-th column element calculates the maximum value M j =max{x 1j , x 2j , ..., x nj}, j = 1, 2, ..., m, and then do the transformation i=1, 2, ..., n;

[0024] S13: Normalized data, using the maximum and minimum method x il ∈[0,1], i=1,2,...,n; l=1,2,...m, similarity is constructed so that the similarity matrix is ​​R=(r ij ) n×n , r ij ∈[0, 1];

[0025] S14: Calculate the transfer package of the fuzzy similarity matrix R, when A = R a =R 2a, (1≤l≤n) is established for the first time, the fuzzy equivalent matrix A is obtained, the appropriate interception level λ∈[0,1] is selected, and the λ interception matrix A of A is obtained λ ; Based on A λ Classify the products by similarity and obtain the similar batch product set EP;

[0026] S15: Batch similar EP k The average processing time T of all product sets P′={p′1,p′2,…,b} in avg As EP k The processing time T of batch scheduling, c=1,2,3...,b,tp′ c is the processing time of the cth product in P′, sp′ c is the number of the cth product in P′.

[0027] In the above technical solution, in step S2 of workshop data perception and disturbance event detection, the workshop data perception senses the real-time operation information of the workshop equipment through sensors, including the processing speed, processing time, processing temperature, pressure, and energy consumption of the equipment, obtains the on-the-job information of the workers through cameras, obtains the type and quantity of materials through barcode / RFID equipment, senses the processing method status through special quality inspection equipment, and senses the on-site environment status through alarm equipment.

[0028] In the above technical solution, in step S2 of workshop data perception and disturbance event detection, workshop disturbance detection includes the following steps:

[0029] S21: Train the historical data and status of the equipment operation process information, build an equipment fault classification model, extract the facial features of employees, and build an employee facial feature library;

[0030] S22: Check the tags of all devices and divide the devices into a normal group and an abnormal group according to the tags;

[0031] S23: For normal group equipment, the real-time perceived equipment operation process data is used to predict equipment failures, and whether there is a failure in equipment m. If the prediction result is that equipment m is about to fail, the scheduling optimization processing module is informed to implement equipment failure evolution scheduling, and the equipment m constraint in the scheduling is set to 0, and the equipment m is marked as abnormal;

[0032] S24: For the abnormal group equipment, the real-time perceived equipment operation process data is used to predict equipment failures, and whether equipment m has recovered to normal is predicted. If the prediction result is that equipment m has recovered to normal, the scheduling optimization module is informed to implement equipment failure recovery evolution scheduling, and the equipment m constraint in the scheduling is set to 1, and the equipment m is marked as normal;

[0033] S25: Check all employee tags and divide the employees into on-the-job group and off-the-job group according to the tags;

[0034] S26: For the on-the-job group, the employee face photos perceived in real time are processed by feature F c Extract and calculate the feature data F in the feature library pre and F cur The Euclidean distance value V of max , V max →0, the scheduling optimization module is informed to implement employee off-duty evolution scheduling, set the employee S constraint in the scheduling to 0, and mark the employee S as off-duty;

[0035] S27: For the off-duty group, the employee face photos perceived in real time are processed by feature F c Extract and calculate the feature data F in the feature library pre and F cur The Euclidean distance value V of max Value, V max →0, the scheduling optimization module is informed to implement employee off-duty recovery evolutionary scheduling, and the employee S constraint in the scheduling is set to 1, and the employee S is marked as on-duty.

[0036] S28: Check the marks of all materials and divide the materials into material defect groups and material defect recovery groups according to the marks;

[0037] S29: For the material shortage group, the real-time perceived quantity of raw materials in inventory and the quantity of products to be processed are compared. i The quantity is less than that of processed product P i The number of products of this type is marked as material shortage, and the scheduling optimization module is informed to implement material shortage evolution scheduling and reduce P in the scheduling task. i The number (value is Δ=P i -Ps i ), and record the defect value Δ;

[0038] S210: For the material defect recovery group, the real-time perceived quantity of raw materials in inventory and the quantity of products to be processed are compared. i The quantity of processed product P i , mark this type of product as material defect recovery, inform the scheduling optimization module, implement material defect recovery evolution scheduling, and increase the P in the scheduling task i The number of (value is Δ), clear the missing value (Δ=0);

[0039] S211: Real-time perception to obtain quality inspection equipment information, when P iWhen product abnormal information is found, the corresponding processing equipment m and operator s are found, and the product P is marked i , equipment m and operator s are processing method errors, inform the scheduling optimization module, implement the processing method error evolution scheduling, and remove the mark when the scheduling process is completed;

[0040] S212: Real-time perception and acquisition of alarm device information of the on-site environment. When the alarm device information is acquired, the processing equipment m, operator s, and processed product P corresponding to the safety accident are found. i , and mark the product P i , equipment m and operator s are safety accidents, which inform the scheduling optimization module to implement safety accident evolution scheduling. When the scheduling processing is completed, the mark is removed.

[0041] The beneficial effects of the present invention are:

[0042] The method of the present invention improves the flexibility and robustness of workshop scheduling, can improve production efficiency, reduce production costs and personnel costs, better meet the changing needs of smart workshops at this stage, and has broad application prospects. Specifically including:

[0043] 1. Effective prediction of disturbance events is achieved by collecting, merging, classifying and judging the status data and historical data of personnel, equipment, materials and tasks during the processing.

[0044] 2. Construct a data-driven disturbance event evolution scheduling strategy to achieve efficient scheduling and optimization of disturbance events.

[0045] 3. Use the Grey Wolf Optimization Algorithm to calculate multi-objective and multi-resource batch scheduling problems and efficiently generate scheduling allocation plans.

[0046] 4. The fuzzy clustering method is used to classify products, which speeds up the grouping of products, reduces product varieties, increases product batches, optimizes the production problem of "multiple varieties and small batches" in the workshop, and improves the workshop's response speed to orders.

[0047] 5. The present invention solves the problem of how to transform "multiple varieties and small batches" (i.e., many orders but few products in the orders) into "few varieties and large batches" (in batch production task orders, the products in each order are similar products), and highlights the order optimization composition, i.e., how to achieve unified batch production of products with different orders but similar processes. At the same time, it can minimize the maximum completion time and minimize the maximum cost (personnel cost, equipment energy consumption cost, and truck energy consumption cost), so that the products can be delivered within the delivery period and the cost during the production process is low.

[0048] 6. It not only solves the problem of many product varieties and small batches in orders, but also solves the problem of multi-objective and multi-resource scheduling and allocation in the workshop. Combined with digital twin technology, it can perceive the workshop status in real time and perform evolutionary scheduling for disturbance events, which enhances the management and control of the actual workshop, and improves the flexibility and robustness of the workshop in response to emergencies and the effectiveness of the scheduling plan. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The accompanying drawings are used to provide a further understanding of the technical solution of the present application or the prior art, and constitute a part of the specification.

[0050] Figure 1 The figure is a diagram of the overall structure of the method of the present invention.

[0051] Figure 2 The figure is an overall flow chart of the method of the present invention. DETAILED DESCRIPTION

[0052] In order to make the objectives, technical solutions and advantages of the present application more clear, the present invention is further described in detail below with reference to the accompanying drawings.

[0053] The purpose of the present invention is to provide a multi-variety, small-batch, multi-resource scheduling method based on digital twins, which solves the problem of many product varieties and small batches in orders, and solves the problem of multi-objective and multi-resource scheduling and allocation in workshops. By combining digital twin technology, the workshop status can be perceived in real time, and evolutionary scheduling can be performed on disturbance events, which enhances the management and control of the actual workshop, and improves the flexibility and robustness of the workshop for emergencies and the effectiveness of the scheduling plan.

[0054] The overall structure diagram of the present invention is as follows Figure 1 As shown, it includes order optimization processing, workshop data perception and disturbance event detection and scheduling optimization processing. The overall flow chart of the present invention is as follows Figure 2 As shown in the figure, in the workshop, order optimization is carried out for the orders of "multiple varieties and small batches" to obtain the production goal of "few varieties and large batches". Taking the current status information of "man-machine-object-method-environment" in the workshop as a constraint, a scheduling model is constructed with the goal of minimizing the maximum completion time and total processing cost. The gray wolf optimization algorithm is used to calculate the scheduling plan, and the scheduling plan is implemented in the actual workshop. At the same time, the disturbance events of "man-machine-object-method-environment" in the workshop are perceived in real time, and evolutionary scheduling optimization is performed according to the disturbance type.

[0055] against Figure 1 The specific content of order optimization is as follows: Order optimization processing includes order splitting and order aggregation. Order splitting is to split B customer orders according to the industry standard categories of products to obtain a total set P containing n products, where P iis the i-th product. The fuzzy analysis method is used to perform product process similarity analysis and calculation on the products in the total set P containing n products. Products with the same process similarity are merged into similar batches to form a product set EP containing K similar batches, where EP k is the kth batch of similar products, where K≤n, and the fuzzy clustering analysis steps are:

[0056] a) The basic data matrix is ​​composed of product processes m is the product process number;

[0057] b) Use maximum normalization: for the feature index matrix X * The j-th column element calculates the maximum value M j =max{x 1j , x 2j , ..., x nj}, j = 1, 2, ..., m, and then do the transformation i=1, 2, ..., n;

[0058] c) After normalization, use the maximum and minimum method x il ∈[0,1], i=1,2,...,n; l=1,2,...m, similarity is constructed so that the similarity matrix is ​​R=(r ij ) n×n , r ij ∈[0, 1];

[0059] d) Calculate the transfer package of the fuzzy similarity matrix R, when A = R a =R 2a , (1≤l≤n) is established for the first time, the fuzzy equivalent matrix A is obtained, the appropriate interception level λ∈[0,1] is selected, and the entry interception matrix A of A is obtained λ ; Based on A λ Classify the products by similarity and obtain the similar batch product set EP.

[0060] e) Similar batches of EP k The average processing time T of all product sets P′={p′1,p′2,…,b} in avg As EP k The processing time T of batch scheduling, c=1,2,3...,b,tp′ c is the processing time of the cth product in P′, sp′ c is the number of the cth product in P′.

[0061] against Figure 1The data status perception and transmission of the workshop is realized by sensing the real-time operation information of the workshop equipment through sensors, including the processing speed, processing time, processing temperature, pressure, and energy consumption of the equipment. The on-the-job information of the workers is obtained through cameras, the type and quantity of materials are obtained through barcode / RFID equipment, the processing method status is perceived through special quality inspection equipment, and the on-site environmental status is perceived through alarm equipment.

[0062] against Figure 1 The specific implementation of the disturbance detection in the workshop is as follows:

[0063] (1) Train the historical data of equipment operation process information and its status to build an equipment fault classification model, extract the facial features of employees, and build an employee facial feature library;

[0064] (2) Check the tags of all devices and divide the devices into normal group and abnormal group according to the tags;

[0065] (3) For normal group equipment, the real-time perceived equipment operation process data is used to predict equipment failures and predict whether there is a failure in equipment m. If the prediction result is that equipment m is about to fail, the scheduling optimization module is informed to implement equipment failure evolution scheduling, and the equipment m constraint in the scheduling is set to 0, and equipment m is marked as abnormal;

[0066] (4) For the abnormal group of equipment, the real-time perceived equipment operation process data is used to predict equipment failures and predict whether equipment m has recovered to normal. If the prediction result is that equipment m has recovered to normal, the scheduling optimization module is informed to implement equipment failure recovery evolution scheduling, and the equipment m constraint in the scheduling is set to 1, and equipment m is marked as normal;

[0067] (5) Check all employee tags and divide the employees into on-the-job group and off-the-job group according to their tags;

[0068] (6) For the on-the-job group, the employee’s face photos perceived in real time are processed by feature F c Extract and calculate the feature data F in the feature library pre and F cur The Euclidean distance value V of max , V max →0, the scheduling optimization module is informed to implement employee off-duty evolution scheduling, and the employee S constraint in the scheduling is set to 0, and the employee S is marked as off-duty;

[0069] (7) For the off-duty group, the employee’s face photos perceived in real time are processed by feature F c Extract and calculate the feature data F in the feature library pre and F cur The Euclidean distance value V of max Value, Vmax →0, the scheduling optimization module is informed to implement employee off-duty recovery evolutionary scheduling, and the employee S constraint in the scheduling is set to 1, and the employee S is marked as on-duty.

[0070] (8) Check the markings of all materials and classify the materials into material defect groups and material defect recovery groups according to the markings;

[0071] (9) For the material shortage group, the real-time perceived quantity of raw materials in inventory and the quantity of products to be processed are compared. i The quantity is less than that of processed product P i The number of products of this type is marked as material shortage, and the scheduling optimization module is informed to implement material shortage evolution scheduling and reduce P in the scheduling task. i The number (value is Δ=P i -Ps i ), and record the defect value Δ;

[0072] (10) For the material defect recovery group, the real-time perceived quantity of raw materials in inventory and the quantity of products to be processed are compared. i The number of processed products Pi is not less than the number of processed products Pi, and the products are marked as material defect recovery, and the scheduling optimization module is informed to implement material defect recovery evolution scheduling, and increase P in the scheduling task i The number of (value is Δ), clear the missing value (Δ=0);

[0073] (11) Real-time perception to obtain quality inspection equipment information. When P i When product abnormal information is found, the corresponding processing equipment m and operator s are found, and the product P is marked i , equipment m and operator s are processing method errors, inform the scheduling optimization module, implement the processing method error evolution scheduling, and remove the mark when the scheduling process is completed;

[0074] (12) Real-time perception and acquisition of alarm device information in the on-site environment. When the alarm device information is acquired, the processing equipment m, operator s, and processed product P corresponding to the safety accident are found. i , and mark the product P i , equipment m and operator s are safety accidents, inform the scheduling optimization module, implement the safety accident evolution scheduling, and remove the mark when the scheduling processing is completed;

[0075] against Figure 1 The mathematical model of the scheduling optimization process is modeled by mathematical programming, and the specific implementation method is as follows:

[0076] (1) Constructing the objective function

[0077] F=min{αF1+(1-α)F2}, where F1=max(C i ) represents the maximum completion time, Among them, f2 represents the personnel cost and equipment energy consumption cost, f3 represents the transport vehicle energy consumption cost, and F2 represents the total cost of the production process;

[0078] (2) Adding constraints

[0079] For the same product, the next process can only be carried out after the previous process is completed;

[0080] Only one piece of equipment can be selected for processing in the same process, and only one worker and one transport vehicle can be selected for operation at the same time;

[0081] C ij -Cgh+H(1-c ijk )+H(1-x ghk )+H×y ijghk ≥t ijk ,

[0082] C gh -C ij +H(1-c ijk )+H(1-x ghk )+H×(1-y ijghk )≥t ghk , indicating that there is a sequential relationship between the processing operations of the same equipment and they cannot be carried out at the same time;

[0083] Among them, x ijk ∈{0, 1}, x ijks ∈{0, 1}, y ijghk ∈{0,1},i,g=1,2,...,N,j=1,2,...r i , h = 1, 2, ... r g , k = 1, 2, ..., M, s = 1, 2, ... S, random coefficients of α∈(0, 1), C i represents the final processing completion time of similar batch i, N represents the total number of similar batches of workpieces to be processed in the production workshop, M represents the total number of processing equipment, S represents the total number of operating employees, and r i represents the total number of processes of workpiece i, O ij represents the jth processing step of workpiece i, t ijk Representative process O ij Processing time at processing equipment k, C ijks Representative process O ij The processing cost of processing by worker s on equipment k, Eijks Representative process O ij Energy consumption of processing by worker s on equipment k, E ijkw Representative process O ij Energy consumption of AGVw for handling on equipment k, x ijks Representative process O ij If the variable value is 1 when the machine is processed on equipment k and by operator s, otherwise it is 0. ijkw Representative process O ij If the variable value is 1 when it is processed on equipment k and transported by transport vehicle w, otherwise it is 0. ijghk Representative process O ij Is it prioritized over O? gh Executed on device k, if yes, the variable value is 1, otherwise 0;

[0084] against Figure 1 The dynamic scheduling strategy for scheduling optimization processing is specifically implemented as follows:

[0085] Machine fault evolution scheduling: Obtain all processing procedures and processing times of the unprocessed workpieces assigned to the machine and the unfinished procedures and processing time information of the workpieces to be processed as the scheduling target, and use the last scheduling plan as the initial solution of the Grey Wolf Optimization Algorithm, and record the scheduling plan calculated by the algorithm;

[0086] Evolutionary scheduling of staff leaving their jobs: Find all the employees with the same operating capabilities as the employee from the historical allocation plans and record the allocation plans, select the employee with the least processing volume as the replacement for the employee, and record the current allocation plan.

[0087] Material defect evolution scheduling: obtain all products related to defective raw materials, reduce the number of unprocessed products Δ, and move the processing order of products after this type of products forward.

[0088] Scheduling of the evolution of processing method errors: Urgently remind operators to correct processing methods and equipment processing parameters, obtain the raw materials corresponding to processing defective products, and temporarily insert them into the production equipment for processing according to the original scheduling plan.

[0089] Safety accident evolution scheduling: For employees involved in safety accidents, their health status is judged. If they cannot continue to work, they will be scheduled to leave their posts. For the equipment involved, their operating status is judged. If a failure occurs, equipment failure scheduling is carried out. For the products involved, the corresponding raw materials are obtained and they will be temporarily inserted into the production equipment for processing according to the original scheduling plan.

[0090] The contents not described in detail in the specification of the present invention belong to the prior art known to professionals in the field, and the above description is only a preferred embodiment of the present invention.

[0091] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all these improvements and changes should fall within the scope of protection of the appended claims of the present invention.

Claims

1. A multi-variety, small-batch, multi-resource scheduling method based on digital twins, characterized in that Including order optimization processing, workshop data perception and disturbance event detection, and scheduling optimization processing steps: S1: Order optimization processing, based on the order process similarity classification method of fuzzy clustering analysis, customer orders are split according to industry standard classification to obtain a total set of different products with the smallest unit. Fuzzy clustering classification calculation is performed on the products in the total set according to process similarity, products with the same similarity are merged into similar batches, and all similar batches of products are used as the basic production targets for batch scheduling; S2: Workshop data perception and disturbance event detection. The workshop disturbance event detection method based on perception data fusion realizes the real-time acquisition of the workshop "man-machine-object-method-environment" status information, and transmits the acquired information to the disturbance event processing module for disturbance factor judgment. The workshop status data is processed in real time, including the judgment and classification of workshop equipment abnormalities, personnel leaving the post, material defects, processing method errors and safety accidents, and the disturbance events are transmitted to the scheduling optimization processing to realize the dynamic evolution scheduling of production; S3: Scheduling optimization processing, including the initialization pre-scheduling stage and the dynamic evolution stage; the initialization pre-scheduling model stage is used to realize the multi-objective multi-resource scheduling allocation problem, minimizing the maximum completion time and minimizing the production cost as the processing goals, comprehensively considering the dependence of processing equipment, personnel and transport vehicles, and determining Similar batches of workpieces The processing sequence of each device, Workers in The distribution of processing on each piece of equipment, and A transporter in The distribution on the devices, where , The scheduling problem is solved using the Grey Wolf Optimization Algorithm; the dynamic evolution stage realizes effective evolution scheduling optimization for different disturbance events; The fuzzy cluster analysis method is used to split the order. Total product collection The products in the process are analyzed and calculated for similarity, and the products with the same process similarity are combined into similar batches to form a batch containing Similar batch product set ,in For the Batch of similar products, including , the fuzzy cluster analysis steps are: S11: The basic data matrix is ​​composed of product processes , is the product process number; S12: Use maximum normalization: for the characteristic index matrix No. Calculate the maximum value of column elements ; Then do the transformation ;in ; S13: Normalized data, using the maximum and minimum method , , , Perform similar construction so that the similarity matrix is , ; S14: Calculate the transfer package of the fuzzy similarity matrix R, when ,and When it is established for the first time, the fuzzy equivalence matrix is ​​obtained , select the appropriate intercept level , and find of Truncation Matrix ;in accordance with Classify products similarly to obtain similar batch product sets ; S15: Similar batches All product collections Average processing time As Processing time for batch scheduling ,in ,in ,in for The Processing time of products, for Middle The number of products, Representative product collection The last element of is the index number of the last element, .

2. The multi-variety, small-batch, multi-resource scheduling method based on digital twins according to claim 1 is characterized by: The order is split into Customer orders are split according to the industry standard categories of products, and the Total product collection ,in For the Category products.

3. The multi-variety, small-batch, multi-resource scheduling method based on digital twins according to claim 2 is characterized in that: The method specifically comprises the following steps: S21: Train the historical data and status of the equipment operation process information, build an equipment fault classification model, extract the facial features of employees, and build an employee facial feature library; S22: Check the tags of all devices and divide the devices into a normal group and an abnormal group according to the tags; S23: For normal group equipment, the real-time perceived equipment operation process data is used to predict equipment failures to predict whether there is equipment failure. Failure, if the prediction result is equipment If a fault is about to occur, the scheduling optimization processing module is informed to implement the equipment fault evolution scheduling and schedule the equipment in the scheduling Set the constraints to 0 and set the device Mark as abnormal; S24: For abnormal group equipment, the real-time perceived equipment operation process data is used to predict equipment failures and predict whether there is any equipment failure. Return to normal if the prediction result is device If the fault is restored to normal, the scheduling optimization module is informed to implement the equipment failure recovery evolution scheduling and allocate the equipment in the scheduling Set the constraint to 1 and set the device Mark as normal; S25: Check all employee tags and divide the employees into on-the-job group and off-the-job group according to the tags; S26: For the on-the-job group, the real-time perceived employee face photos are characterized Extract and calculate feature data in the feature library and The Euclidean distance value ,if , , then inform the scheduling optimization module to implement employee off-duty evolution scheduling and transfer the employees in the schedule Set the constraint to 0 and set the personnel Mark as off-post; S27: For the off-duty group, the real-time perceived employee face photos are characterized Extract and calculate feature data in the feature library and The Euclidean distance value ,if , , the scheduling optimization module is informed to implement the employee off-duty recovery evolution scheduling and schedule the employees in the schedule The constraint is set to 1, and the personnel Mark as in-service; S28: Check the marks of all materials and divide the materials into material defect groups and material defect recovery groups according to the marks; S29: For the material shortage group, the real-time perceived quantity of raw materials in inventory and the quantity of products to be processed are compared. The quantity is less than that of processed products The quantity of raw materials will appear The quantity is less than that of processed products The number of products is marked as material shortage, and the scheduling optimization module is informed to implement material shortage evolution scheduling and reduce the number of scheduling tasks. The number of missing values ,in ; S210: For the material defect recovery group, the real-time perceived quantity of raw materials in inventory and the quantity of products to be processed are compared. The quantity is not less than the processed products The quantity of raw materials The quantity is not less than the processed products The quantity of products is marked as material defect recovery, and the scheduling optimization module is informed to implement material defect recovery evolution scheduling and increase the scheduling task The amount of increase is , and clear the missing values ; S211: Real-time perception to obtain quality inspection equipment information. When product abnormal information is detected, find the corresponding processing equipment and operator , and mark the product ,equipment and operator In case of processing method error, the scheduling optimization module is informed to implement the processing method error evolution scheduling. When the scheduling process is completed, the mark is removed; S212: Real-time perception and acquisition of alarm device information of the on-site environment. When the alarm device information is acquired, the processing equipment corresponding to the safety accident is found. , Operator and processed products , and mark the product ,equipment and operator For safety accidents, the scheduling optimization module is informed to implement safety accident evolution scheduling. When the scheduling processing is completed, the mark is removed.

4. The multi-variety, small-batch, multi-resource scheduling method based on digital twins according to claim 3 is characterized in that The specific implementation is as follows: Machine fault evolution scheduling: Obtain the processing procedures and processing time of all unprocessed workpieces assigned to the target machine, as well as the unfinished procedures and processing time information of the workpieces to be processed, as the scheduling target; and use the last scheduling plan as the initial solution of the Grey Wolf Optimization Algorithm, calculate and record the scheduling plan generated by the algorithm; Evolutionary scheduling of staff leaving their jobs: Find all the allocation records of employees with the same operating ability as the employee with the problem from the historical allocation plan, select the employee with the least processing volume as a replacement, and record the current allocation plan; Material defect evolution scheduling: obtain all products related to defective raw materials and reduce the number of unprocessed products , move the processing order of all products after the products related to the defective raw materials forward; Scheduling of the evolution of processing method errors: Urgently remind the operator to correct the processing method and equipment processing parameters, obtain the raw materials corresponding to the defective products, and temporarily insert them into the production equipment for processing according to the original scheduling plan; Scheduling of the evolution of safety accidents: For employees involved in safety accidents, judge their health status. If they cannot continue to work, schedule them to leave their posts; for the equipment involved, judge the operating status. If a failure occurs, conduct equipment failure scheduling. For the products involved, obtain the raw materials corresponding to the products, and temporarily insert them into the production equipment for processing according to the original scheduling plan.

5. A multi-variety, small-batch, multi-resource scheduling system based on digital twins, characterized by The multi-variety, small-batch, multi-resource scheduling method based on digital twins as described in any one of claims 1 to 4 is adopted; at least comprising: an order optimization module, a workshop data perception and disturbance event detection module, and a scheduling optimization module; The order optimization module splits customer orders according to industry standard classification, combines similar batches according to the process similarity of the products, and passes the combination as the basic task of production to the scheduling optimization module; The workshop data perception and disturbance event detection module, based on the workshop disturbance detection method of perception data fusion, realizes the real-time acquisition of the workshop "man-machine-object-method-environment" status information, and transmits the acquired information to the disturbance event processing module to judge the disturbance factors, processes the workshop status data in real time, and transmits the determined disturbance events to the scheduling optimization module to realize the dynamic evolution scheduling of production; The scheduling optimization module receives and processes the information data transmitted by other modules, and performs iterative optimization to obtain an excellent scheduling solution.

6. The multi-variety, small-batch, multi-resource scheduling system based on digital twins according to claim 5 is characterized by: (1) Order optimization processing module: Based on the order process similarity classification method of fuzzy clustering analysis, customer orders are split according to industry standard classification to obtain a total set of different products with the smallest unit. Fuzzy clustering classification calculation is performed on the products in the total set according to process similarity, and products with the same similarity are merged into similar batches. All similar batches of products are used as the basic production targets for batch scheduling; (2) Workshop data perception and disturbance event detection module: Based on the workshop disturbance event detection method of perception data fusion, it realizes the real-time acquisition of the workshop "man-machine-object-method-environment" status information, and transmits the acquired information to the disturbance event processing module to judge the disturbance factors. It processes the workshop status data in real time, including the judgment and classification of workshop equipment abnormalities, personnel leaving the post, material defects, processing method errors and safety accidents, and transmits the disturbance events to the scheduling optimization processing to realize the dynamic evolution scheduling of production; (3) Scheduling optimization processing module, including initialization pre-scheduling module and dynamic evolution module. The initialization pre-scheduling module is used to solve the multi-objective multi-resource scheduling allocation problem, with the maximum completion time and production cost minimized as the processing objectives. The module comprehensively considers the dependence of processing equipment, personnel and transport vehicles to determine Similar batches of workpieces The processing sequence of each device, Workers in The distribution of processing on each piece of equipment, and A transporter in The distribution on the devices, where , The scheduling problem is solved using the Grey Wolf Optimization Algorithm. The dynamic evolution module implements effective evolutionary scheduling optimization for different disturbance events.

Images