Flexible production management method and system for multi-variety small-batch production order

Abstract

The invention provides a flexible production management method and system for multi-variety small-batch production orders. The method comprises the following steps: when a new order is received by using an SAP system, obtaining order adjustment information; inputting the order adjustment information into a decision recommendation model to obtain decision recommendation information; wherein the decision recommendation model is integrated in an MES system, and the decision recommendation information comprises configuration parameter information of a production line; adjusting a production plan according to the decision recommendation information, and performing flexible production according to the adjusted production plan; wherein the order adjustment information comprises new order information, production line information and material management information of the WMS. Through analysis and deep learning of production big data, the decision recommendation information is automatically generated for multi-variety and small-batch production characteristics and frequent change of customer orders, so that the production plan is adjusted, the production efficiency is improved, and the purpose of intelligent flexible production is achieved.

Description

Technical Field

[0001] The present application relates to the field of industrial production technology, and in particular to a flexible production management method and system for multi-variety small-batch production orders. Background Art

[0002] Flexible production is a production method that focuses on the construction of "the ability of the manufacturing system to respond to changes in the internal and external environment". In discrete manufacturing, it refers to the results of relying on the ability of a highly flexible computer manufacturing system to respond to changes in the internal and external environment to reduce production costs and improve efficiency. At present, most discrete manufacturing companies use the ERP+MES method to handle the conversion and scheduling of sales orders to production orders. The basis for scheduling production is mainly customers, order quantity and delivery date. At the production end, the production plan will be arranged in the MES system according to the product type, delivery date and production capacity of each production line of the production order.

[0003] However, the above-mentioned SAP method of issuing production work orders to the MES system and then arranging production plans for the MES system is more suitable for production companies with large batches and small order fluctuations. For the production model of multiple varieties and small batches, since customer orders fluctuate greatly and change frequently, the above-mentioned traditional production method will cause constant modifications to the production plan, resulting in waste of resources and reduced efficiency.

[0004] Based on this, the present application provides a flexible production management method and system for multi-variety small batch production orders to improve the existing technology and meet the needs of practical applications. Summary of the invention

[0005] The purpose of this application is to provide a flexible production management method and system to solve the problem of constantly modifying the production plan for multi-variety small batch production orders.

[0006] The purpose of this application is achieved by the following technical solutions:

[0007] In a first aspect, the present application provides a flexible production management method for multi-variety small batch production orders, the method comprising:

[0008] When a new order is received using the SAP system, order adjustment information is obtained;

[0009] Inputting the order adjustment information into a decision recommendation model to obtain decision recommendation information; wherein the decision recommendation model is integrated into an MES system, and the decision recommendation information includes configuration parameter information of a production line;

[0010] Adjusting the production plan according to the decision recommendation information, and performing flexible production according to the adjusted production plan;

[0011] The order adjustment information includes new order information, production line information and material management information of the WMS system.

[0012] In some possible implementations, the construction of the decision recommendation model includes:

[0013] Use big data technology to obtain multiple historical production information from the MES system, ERP system, and WMS system, and build a training set;

[0014] Based on the training set, a regression algorithm is used for training to obtain a decision recommendation model.

[0015] In some possible implementations, the training based on the training set using a regression algorithm to obtain a decision recommendation model includes:

[0016] Based on the training set, a decision tree regression model is used to calculate variables, and a linear regression algorithm is used to obtain a decision prediction result;

[0017] By comparing the difference optimization algorithms, a decision recommendation model is obtained.

[0018] In some possible implementations, the decision tree regression model is a model constructed by a decision tree regression algorithm, and the decision tree algorithm includes an ID3 algorithm, a C4.5 algorithm, or a CART algorithm.

[0019] In some possible implementations, the construction of the decision tree regression model includes:

[0020] S1, obtaining a data set for training a decision tree, wherein the data set includes a plurality of samples;

[0021] S2, check whether all samples belong to the same category. If all samples of the current node belong to the same category, no further division is required, and leaf nodes are directly generated for category labeling;

[0022] S3, when all samples are not of the same class, extract all possible features from the training set to form a feature candidate set, and determine whether all feature values ​​in the feature candidate set are unique;

[0023] S4, if the values ​​of all features are unique, put them into the candidate data set until the feature set is extracted, generate leaf nodes and select the class with the largest proportion in the candidate data set for category labeling;

[0024] S5, if the values ​​of all features are not unique, select the optimal partitioning feature according to CART and generate a decision node;

[0025] S6, determining whether all features in the feature candidate set have been considered, if not, continuing to generate data subsets, and deleting the optimal partitioning feature to obtain a new data subset; using the new data subset as a data set for training the decision tree and executing S1;

[0026] S7, if yes, stop splitting and generate leaf nodes, and obtain the decision tree regression model through pruning, evaluation and tuning processes.

[0027] In some possible implementations, the new order information includes material number, BOM, quantity, delivery date, UPH, material shortage status and production status; the production line information includes workshop identification, line number, estimated total capacity and proficiency; the material management information includes bill of materials, inventory level and material storage location.

[0028] In a second aspect, the present application also provides a flexible production management system, the system comprising:

[0029] An information acquisition module, used to acquire order adjustment information when a new order is received, wherein the order adjustment information includes production order information, production line information and real-time production line information of the new order;

[0030] A model calculation module, used for inputting the order adjustment information into a decision recommendation model to obtain decision recommendation information; wherein the decision recommendation model is integrated into the MES system;

[0031] The plan adjustment module is used to adjust the production plan according to the decision recommendation information and to produce according to the adjusted production plan.

[0032] Compared with the prior art, the advantages and positive effects of the present application are: it can quickly adapt to order changes, realize efficient production of small batches of multiple varieties, and improve the flexibility and adaptability of the production line. Through intelligent decision-making recommendations, the configuration of production lines and resources is optimized, resource utilization is improved, and waste is reduced. The decision model integrated in the MES system can quickly respond to new order information and production changes, and shorten the decision-making time. Using the data in the SAP system, WMS system and MES system, data-based decisions are made to improve the accuracy and reliability of decisions. The present application provides a data-driven method and a data-driven system. Through the analysis and deep learning of production big data, the production plan is adjusted according to the production characteristics of small batches of multiple varieties and the frequent changes in customer orders, so as to improve production efficiency and achieve the purpose of intelligent flexible production. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The present application is further described below in conjunction with the accompanying drawings and implementation methods.

[0034] Figure 1It is a flow chart of a flexible production management method provided in an embodiment of the present application.

[0035] Figure 2 It is a flowchart of building a decision recommendation model provided in an embodiment of the present application.

[0036] Figure 3 It is a flowchart of a training decision tree regression model provided in an embodiment of the present application.

[0037] Figure 4 It is a structural diagram of a flexible production management system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0038] The technical solution in the present application will be described below in conjunction with the drawings and specific implementation methods of the specification of the present application. It should be noted that, under the premise of no conflict, the various implementation methods or technical features described below can be arbitrarily combined to form a new implementation method.

[0039] In the embodiments of the present application, the words "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any implementation or design scheme described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other implementations or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.

[0040] The technical solution in the present application will be described below in conjunction with the drawings and specific implementation methods of the specification of the present application. It should be noted that, under the premise of no conflict, the various implementation methods or technical features described below can be arbitrarily combined to form a new implementation method.

[0041] SAP (Systems Applications and Products in Data Processing, Enterprise Management Solution) system. MES (Manufacturing Execution System) is a software-based solution for monitoring and controlling the production process of the workshop during the manufacturing process. In manufacturing operations management, the MES system acts as a bridge between the enterprise's plans and SAP and the actual manufacturing operations. The main purpose of the MES system is to track and record the transformation of raw materials to finished products in real time. It captures data from various sources (including machines, sensors, and operators) to provide accurate and up-to-date information about the status of production activities. WMS (Warehouse Management System) provides visibility into the entire inventory of the enterprise.

[0042] Multi-variety small batch production orders refer to a production model in which a manufacturing enterprise accepts and processes production orders for a variety of different products, with a relatively small order volume for each product. For example, in the production of auto parts, different models of engine parts need to be produced, and the order volume for each part is not large, but the variety is complex. Another example is clothing manufacturing, which produces clothing of various styles and sizes, with a small order volume for each style, but the styles change quickly.

[0043] In the related technology, after receiving the customer's purchase order, SAP issues a production order, WMS prepares materials and then MES schedules production, produces according to the process flow, and finally completes the warehousing. If there is a SAP order modification in the middle, the production plan needs to be adjusted, and then WMS prepares materials to continue the subsequent steps. However, the above production plan arrangement method, for the production mode of multi-variety and small batch, needs to constantly modify the production plan, resulting in waste of resources and reduced efficiency.

[0044] Application No. CN202010004389.2 discloses a flexible system that can be quickly switched, which is suitable for the production of multiple varieties and small batches. Although it is recognized that the production of multiple varieties and small batches has gradually become the main production form of small and medium-sized production and processing enterprises, the technical solution provided by it focuses on the hardware-improved flexible system that can be quickly switched. It is composed of a zero-point quick-change module, a machine tool automatic door module, a mobile robot platform module, and an adjustable equipment pad. Several modules can quickly build a flexible production system for the automated production of parts. At the same time, the flexible production system can be quickly removed under working conditions that require traditional manual production mode, and the normal operation of the manual production mode will not be affected. The flexible system can not only ensure the processing quality of the products and greatly improve the processing efficiency, but also the characteristics of rapid switching are particularly suitable for the current production status of small and medium-sized production and processing enterprises. In other words, it mainly starts with the improvement of the physical system.

[0045] With the promotion of SAP systems, MES systems, and WMS systems, we will face application scenarios with large amounts of data, and the improvement effect of the above physical systems will be limited. Based on this, the present application provides a flexible production management method and system, which, as a data-driven method and data-driven system, automatically generates decision-making recommendation information to adjust production plans and improve production efficiency, thereby achieving the purpose of intelligent flexible production through analysis and deep learning of production big data, targeting the production characteristics of multi-variety small batches and frequent changes in customer orders, in order to achieve the purpose of intelligent flexible production. The following will first explain the method and then the system.

[0046] Method embodiment.

[0047] See also Figure 1 , Figure 1 It is a flow chart of a flexible production management method provided in an embodiment of the present application.

[0048] An embodiment of the present application provides a flexible production management method for multi-variety small-batch production orders, the method comprising steps S101 to S103.

[0049] S101, when a new order is received by using the SAP system, order adjustment information is obtained, wherein the order adjustment information includes new order information, production line information and material management information of the WMS system.

[0050] New production orders are received using the SAP system. As an enterprise resource planning (ERP) tool, the SAP system can process and store detailed information related to orders. The newly received order information is integrated with the existing production line information and material management information to form order adjustment information. It can be considered that the order adjustment information includes the detailed requirements of the new order, the configuration and capacity of the existing production line, the material supply capacity of the warehouse, and the real-time operation status of the production line.

[0051] Among them, the new order information may include material number, BOM (bill of materials), quantity, delivery time, UPH (unit production hour), material shortage status and production status; the production line information may include workshop identification, line number, estimated total production capacity and proficiency; the material management information may include bill of materials, inventory level and material storage location.

[0052] As an example, an electronic product manufacturer receives a new customer order during the production process. New order information includes, for example: material number, XD123; BOM (Bill of Materials), including components such as motherboard, display, battery, etc.; quantity, 500 pieces; delivery time, 30 days; UPH (Unit Production Hour), 0.5 hours / piece; shortage status, battery shortage; production status, to be arranged. Production line information includes, for example: workshop identification, electronic assembly workshop; line number, assembly line 1#; estimated total capacity: 1,000 pieces / day; proficiency. Material management information includes, for example: bill of materials, detailed list of components and raw materials; inventory level, 100 motherboards, 80 display screens, 30 batteries (shortage); material storage location, motherboard-shelf A1, display screen-shelf A2, battery-shelf B1.

[0053] S102, inputting the order adjustment information into a decision recommendation model to obtain decision recommendation information; wherein the decision recommendation model is integrated into an MES system, and the decision recommendation information includes configuration parameter information of the production line.

[0054] The order adjustment information is input into the decision recommendation model integrated into the MES system. The decision recommendation model can use an algorithm to analyze the input order adjustment information and generate decision recommendations. The decision recommendation model outputs decision recommendation information including configuration parameter information of the production line. The decision recommendation information can be used for reconfiguration of the production line, reallocation of resources, adjustment of production sequence, etc.

[0055] The decision recommendation information may include one or more of: line configuration, resource allocation, production sequence, and time planning. Line configuration may include the selection and settings of machines to recommend which lines to use to process specific orders. Resource allocation may include the number of workers recommended to be assigned to each line, skill requirements, raw material quantities, etc. The production sequence is used to indicate the priority order of order processing. Time planning includes recommendations for the start and end times of each task or order.

[0056] S103, adjusting the production plan according to the decision recommendation information, and performing flexible production according to the adjusted production plan.

[0057] Production managers can adjust production plans directly based on decision-making recommendation information. Specifically, they can adjust the configuration of the production line based on the recommendation information, such as changing machine settings or adjusting the layout of the production line; reallocate workers and raw materials based on the recommendation to ensure that each production line has sufficient resources; adjust the priority of production tasks to meet delivery requirements or optimize production processes and update production sequences; and formulate or update production schedules to ensure that tasks are completed on time.

[0058] Therefore, the original production plan can be adjusted in real time according to the decision recommendation information to adapt to the new order requirements and changes in production conditions. According to the adjusted production plan, flexible production is performed. The technical solution provided by this embodiment can quickly respond to order changes and adapt to the production needs of small batches of multiple varieties, avoiding the situation where the traditional production method causes the production plan to be constantly modified, wastes resources, and reduces efficiency.

[0059] The flexible production management method provided in the present application can be run on an electronic device, and the electronic device can be a computer, a server (including a cloud server), or other device with computing capabilities.

[0060] Therefore, the flexible production management method provided in this embodiment can quickly adapt to order changes, realize efficient production of small batches of multiple varieties, and improve the flexibility and adaptability of the production line. Through intelligent decision-making recommendations, the configuration of production lines and resources is optimized, resource utilization is improved, and waste is reduced. The decision model integrated in the MES system can quickly respond to new order information and production changes, shortening the decision-making time. Using data in the SAP system, WMS system, and MES system, data-based decision-making is achieved to improve the accuracy and reliability of decision-making.

[0061] In summary, by supporting production through intelligent decision-making, rapid response and efficient management of multi-variety small-batch production orders are achieved.

[0062] See also Figure 2 , Figure 2 It is a flowchart of building a decision recommendation model provided in an embodiment of the present application.

[0063] In some embodiments, the construction of the decision recommendation model includes:

[0064] S201, using big data technology to obtain multiple historical production information from the MES system, ERP system and WMS system, and build a training set;

[0065] Using big data technology, historical production information is collected from various information systems of manufacturing enterprises (such as MES, ERP, WMS). Historical production information includes, for example, order details, production line performance, production efficiency, quality control data, material usage, etc. corresponding to order information, production line information and material management information, as well as decision-making recommendation information composed of production line configuration, resource allocation, production sequence, time planning, etc. Big data technology refers to a series of technologies used to collect, store, process and analyze large-scale, diverse and rapidly growing data sets.

[0066] The process of constructing a training set can be: preprocessing the collected historical production information by cleaning, integrating and formatting the data for subsequent processing and analysis; using the preprocessed historical production information as a training set, and using the training set for subsequent model training.

[0067] S202: Based on the training set, a regression algorithm is used for training to obtain a decision recommendation model.

[0068] Therefore, by using big data technology, historical production information is collected from various information systems of production enterprises and then trained to obtain a decision recommendation model, which can adapt to different production scenarios and demand changes and provide flexible production plan adjustment suggestions. The construction method of the decision recommendation model optimizes production management decisions by combining big data technology and machine learning algorithms, and has significant economic benefits and technical value.

[0069] In some embodiments, the training based on the training set using a regression algorithm to obtain a decision recommendation model includes:

[0070] Based on the training set, a decision tree regression model is used to calculate variables, and a linear regression algorithm is used to obtain a decision prediction result;

[0071] By comparing the difference optimization algorithms, a decision recommendation model is obtained.

[0072] Use the decision tree regression model to analyze the training set and identify the key features that affect the production plan. For example, order quantity, product type, delivery deadline, production resource availability, etc. The decision tree regression model recursively splits the data set, identifies the optimal splitting features, generates a series of decision nodes and leaf nodes, and thus extracts variables that have a significant impact on the production plan. Based on the key features extracted by the decision tree regression model, a linear regression model is constructed. For example, the linear regression model fits the data through the least squares method and predicts continuous output values. The continuous output value can be the expected production time, expected cost, etc.

[0073] The comparison difference optimization algorithm refers to comparing the prediction results of the linear regression model with the actual production data in the training set to evaluate the prediction accuracy and reliability of the model. Based on the results of the above comparison, the error analysis and parameter adjustment optimization algorithm are used to improve the prediction accuracy of the model. The specific methods include, for example, adjusting feature weights, optimizing regression coefficients, etc., which are not described here. Through repeated training and verification, iterative training is performed to continuously optimize the model parameters.

[0074] By combining the optimized linear regression model with the decision tree regression model, a comprehensive decision recommendation model can be constructed. The trained and evaluated decision recommendation model is integrated into the MES system so that it can receive new production data in real time and provide decision recommendations.

[0075] Therefore, the decision recommendation model obtained by the above method can accurately formulate production plans by extracting key features using the decision tree regression model and combining it with the linear regression algorithm for prediction. By combining the decision tree regression and linear regression algorithms, using big data analysis and machine learning technology, intelligent and flexible management of multi-variety small batch production modes is achieved. This solution not only improves the accuracy and management efficiency of production plans, but also has highly flexible adaptability, which can significantly improve the production efficiency and customer satisfaction of enterprises.

[0076] In some embodiments, the decision tree regression model is a model constructed by a decision tree regression algorithm, and the decision tree algorithm includes an ID3 algorithm, a C4.5 algorithm, or a CART algorithm.

[0077] In some embodiments, the construction of the decision tree regression model includes:

[0078] S1, obtaining a data set for training a decision tree, wherein the data set includes a plurality of samples;

[0079] S2, check whether all samples belong to the same category. If all samples of the current node belong to the same category, no further division is required, and leaf nodes are directly generated for category labeling;

[0080] S3, when all samples are not of the same class, extract all possible features from the training set to form a feature candidate set, and determine whether all feature values ​​in the feature candidate set are unique;

[0081] S4, if the values ​​of all features are unique, put them into the candidate data set until the feature set is extracted, generate leaf nodes and select the class with the largest proportion in the candidate data set for category labeling;

[0082] S5, if the values ​​of all features are not unique, select the optimal partitioning feature according to CART and generate a decision node;

[0083] S6, determining whether all features in the feature candidate set have been considered, if not, continuing to generate data subsets, and deleting the optimal partitioning feature to obtain a new data subset; using the new data subset as a data set for training the decision tree and executing S1;

[0084] S7, if yes, stop splitting and generate leaf nodes, and obtain the decision tree regression model through pruning, evaluation and tuning processes.

[0085] See also Figure 3 , Figure 3 It is a flowchart of a training decision tree regression model provided in an embodiment of the present application.

[0086] As an example, the process of training the model to be trained based on the training set to obtain the decision tree regression model includes:

[0087] Step 1, start;

[0088] Step 2, obtain the data set; collect and prepare the data set for training the decision tree.

[0089] Step 3: Check whether all samples belong to the same category. If all samples of the current node belong to the same category, no further division is required, and a leaf node is directly generated and labeled with the category.

[0090] For example, at a certain node in the model, check whether all orders have the same production characteristics, such as whether they are all urgent orders. If so, no further adjustment is required, and a leaf node is directly generated and marked as "maintain current production status".

[0091] Step 4: When all samples are not of the same class, all possible features are extracted from the training set to form a feature candidate set. In this stage, the algorithm considers all possible features in the training set for subsequent node partitioning.

[0092] If the orders have different characteristics, extract all the features that may affect the production decision, such as order quantity, product type, delivery deadline, etc.

[0093] Step 5: Determine whether all feature values ​​are unique. If all feature values ​​are unique, then these features are not suitable for further partitioning the data set, and the algorithm will select the class with the largest proportion in the data set to generate a leaf node.

[0094] Step 6: If the values ​​of all features are not unique, select the optimal partition feature a to generate a decision node according to CART. CART selects the optimal feature based on criteria such as the Gini index or information gain rate. After selecting the optimal feature a, the algorithm will use it to generate a decision node. The optimal feature a is, for example, "delivery deadline urgency".

[0095] In specific applications, the ID3 algorithm can be used to calculate the information gain, and the feature with the largest information gain can be selected as the optimal partition feature; the C4.5 algorithm can be used to calculate the information gain ratio, and the feature with the largest information gain ratio can be selected as the optimal partition feature; the CART algorithm can also be used to calculate the Gini index, and the feature with the smallest Gini index can be selected as the optimal partition feature.

[0096] Step 7: Determine whether the traversal of feature values ​​is complete. If all possible values ​​of feature a have been used to generate child nodes, the traversal is complete. If there are still values ​​that have not been considered, the algorithm will continue to generate data subsets.

[0097] For "Delivery Deadline Urgency", check if all urgency levels have been considered. If not, continue splitting.

[0098] Step 8: Generate data subsets. For each value of feature a, the algorithm will generate a new data subset that contains all samples with that feature value.

[0099] According to different levels of "delivery deadline urgency", the orders are divided into several subsets, such as "very urgent", "moderately urgent" and "not urgent".

[0100] Step 9, delete feature a. After generating the data subset, remove the used feature a from the subset, and do not consider the used feature a in the further splitting of the child node. After deleting feature a, get the (new) data subset, and then execute step 1.

[0101] Step 10: When all samples in step 3 are of the same class, select this class. That is, if all samples are of the same class, select this class and execute step 13.

[0102] Step 11, after step 4, determine whether all features in the feature candidate set have been considered (determine whether all features have been considered), that is, determine whether the feature candidate set is empty. If the feature candidate set is empty, that is, there are no more features to choose from, then the algorithm will stop splitting and generate leaf nodes. Specifically, after selecting "delivery deadline urgency", check whether there are other features that can be used to further split the node.

[0103] Step 12: Select the class with the largest proportion in the data set. If the samples of the current node are not of the same class, the algorithm will select the class with the largest proportion in the data set as the predicted class of the current node and execute step 13.

[0104] Among them, selecting the class / selecting the class with the largest proportion in the data set in step 10 and step 12 is the basis for generating leaf nodes in the decision tree. If all samples of a node belong to the same class, a leaf node representing the class is directly generated. If not, the class with the largest proportion in the data set is selected to generate a leaf node.

[0105] For example, on each subset, “delivery deadline urgency” is no longer considered, and other features such as “product complexity” may be considered.

[0106] Step 13, generating leaf nodes of the selected class, that is, generating leaf nodes after determining the category and marking them as the selected category.

[0107] Step 14, end. After the generation of leaf nodes is completed, the decision tree construction for this branch is completed.

[0108] In a specific application, a flexible production management method is provided, which collects data from SPA, MES and WMS systems respectively, and then performs data processing including data cleaning and feature extraction, performs model training, gives predictions and algorithm optimization intermediate algorithm processes, and applies the algorithm to actual production. It includes the following steps:

[0109] 1. Define key parameters. For example, it includes production order information (material number, BOM, quantity, delivery date, UPH, material shortage status, production status, etc.), production line information (workshop, line number, estimated total capacity, proficiency), etc. Defining key parameters can clarify the model input variables. Key parameters determine the input characteristics of the model, that is, which data will be used for model training and prediction.

[0110] 2. Data collection: Connect SAP, MES, and WMS systems to collect required data.

[0111] 3. Big data analysis: Process, filter and analyze large amounts of historical and real-time data, using Hadoop and Storm as processing frameworks.

[0112] 4. Use regression algorithms (linear + decision tree regression models) to train models, calculate variables, make production decisions, and predict results based on decisions, compare differences, and then optimize the algorithm.

[0113] 5. Generate applications based on the trained algorithm model and integrate them with the MES system.

[0114] Therefore, for the production mode of multiple varieties and small batches, and the situation where order information changes frequently, the AI ​​algorithm is combined with the MES system to give the best decision in real time for changes in order parameters, reduce the waste of production resources caused by frequent order switching, improve production efficiency, and maximize the satisfaction of shipment needs.

[0115] In specific applications, the software tools that can be used in this application include Visual Studio, JavaScript, Python, SQL Sever, PowerBI, Hadoop, Storm, and TensorFlow, etc. Among them, SQL Server is used to store structured data; Python is used to develop and train machine learning models; Tensor Flow is used as a machine learning framework to build and train complex models; JavaScript is used in combination with PowerBI to develop custom data visualization components and create interactive data visualization reports; Visual Studio is used as an integrated development environment (IDE) to develop applications; Hadoop and Storm are used as processing frameworks for big data analysis.

[0116] Device Embodiments.

[0117] See also Figure 4 , Figure 4 It is a structural diagram of a flexible production management system provided in an embodiment of the present application.

[0118] The embodiment of the present application provides a flexible production management system, and its specific embodiment is consistent with the embodiment recorded in the above method embodiment and the technical effects achieved, and some contents will not be repeated here.

[0119] Flexible production management system includes:

[0120] An information acquisition module, used to acquire order adjustment information when a new order is received, wherein the order adjustment information includes production order information, production line information and real-time production line information of the new order;

[0121] A model calculation module, used for inputting the order adjustment information into a decision recommendation model to obtain decision recommendation information; wherein the decision recommendation model is integrated into the MES system;

[0122] The plan adjustment module is used to adjust the production plan according to the decision recommendation information and to produce according to the adjusted production plan.

[0123] In some embodiments, the construction of the decision recommendation model includes:

[0124] Use big data technology to obtain multiple historical production information from the MES system, ERP system, and WMS system, and build a training set;

[0125] Based on the training set, a regression algorithm is used for training to obtain a decision recommendation model.

[0126] In some embodiments, the training based on the training set using a regression algorithm to obtain a decision recommendation model includes:

[0127] Based on the training set, a decision tree regression model is used to calculate variables, and a linear regression algorithm is used to obtain a decision prediction result;

[0128] By comparing the difference optimization algorithms, a decision recommendation model is obtained.

[0129] In some embodiments, the construction of the decision tree regression model includes:

[0130] S1, obtaining a data set for training a decision tree, wherein the data set includes a plurality of samples;

[0131] S2, check whether all samples belong to the same category. If all samples of the current node belong to the same category, no further division is required, and leaf nodes are directly generated for category labeling;

[0132] S3, when all samples are not of the same class, extract all possible features from the training set to form a feature candidate set, and determine whether all feature values ​​in the feature candidate set are unique;

[0133] S4, if the values ​​of all features are unique, put them into the candidate data set until the feature set is extracted, generate leaf nodes and select the class with the largest proportion in the candidate data set for category labeling;

[0134] S5, if the values ​​of all features are not unique, select the optimal partitioning feature according to CART and generate a decision node;

[0135] S6, determining whether all features in the feature candidate set have been considered, if not, continuing to generate data subsets, and deleting the optimal partitioning feature to obtain a new data subset; using the new data subset as a data set for training the decision tree and executing S1;

[0136] S7, if yes, stop splitting and generate leaf nodes, and obtain the decision tree regression model through pruning, evaluation and tuning processes.

[0137] It should be noted that, in the embodiments of the present application, "at least one" refers to one or more, and "plurality" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can represent: a, b, c, a and b, a and c, b and c or a and b and c, where a, b and c can be single or multiple. It is worth noting that "at least one" can also be interpreted as "one or more items".

[0138] The terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are configured to distinguish similar objects, and need not be configured to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable in appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, the process, method, system, product or equipment comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0139] This application is explained from the perspectives of purpose of use, effectiveness, progress and novelty, and has met the functional enhancement and usage requirements emphasized by the Patent Law. The above description and drawings of this application are only the preferred embodiments of this application, and are not intended to limit this application. Therefore, all structures, devices, features, etc. that are similar or identical to this application, that is, all equivalent replacements or modifications made in accordance with the scope of the patent application of this application, should fall within the scope of protection of the patent application of this application.

Claims

1. A flexible production management method for multi-variety small batch production orders, characterized in that: The method comprises: When a new order is received using the SAP system, order adjustment information is obtained; Inputting the order adjustment information into a decision recommendation model to obtain decision recommendation information; wherein the decision recommendation model is integrated into an MES system, and the decision recommendation information includes configuration parameter information of a production line; Adjusting the production plan according to the decision recommendation information, and performing flexible production according to the adjusted production plan; The order adjustment information includes new order information, production line information and material management information of the WMS system.

2. The flexible production management method according to claim 1, characterized in that: The construction of the decision recommendation model includes: Use big data technology to obtain multiple historical production information from the MES system, ERP system, and WMS system, and build a training set; Based on the training set, a regression algorithm is used for training to obtain a decision recommendation model.

3. The flexible production management method according to claim 2, characterized in that: The step of training based on the training set using a regression algorithm to obtain a decision recommendation model includes: Based on the training set, a decision tree regression model is used to calculate variables, and a linear regression algorithm is used to obtain a decision prediction result; By comparing the difference optimization algorithms, a decision recommendation model is obtained.

4. The flexible production management method according to claim 3, characterized in that: The decision tree regression model is a model constructed by a decision tree regression algorithm, and the decision tree algorithm includes an ID3 algorithm, a C4.5 algorithm or a CART algorithm.

5. The flexible production management method according to claim 3, characterized in that: The construction of the decision tree regression model includes: S1, obtaining a data set for training a decision tree, wherein the data set includes a plurality of samples; S2, check whether all samples belong to the same category. If all samples of the current node belong to the same category, no further division is required, and leaf nodes are directly generated for category labeling; S3, when all samples are not of the same class, extract all possible features from the training set to form a feature candidate set, and determine whether all feature values ​​in the feature candidate set are unique; S4, if the values ​​of all features are unique, put them into the candidate data set until the feature set is extracted, generate leaf nodes and select the class with the largest proportion in the candidate data set for category labeling; S5, if the values ​​of all features are not unique, select the optimal partitioning feature according to CART and generate a decision node; S6, determining whether all features in the feature candidate set have been considered, if not, continuing to generate data subsets, and deleting the optimal partitioning feature to obtain a new data subset; using the new data subset as a data set for training the decision tree and executing S1; S7, if yes, stop splitting and generate leaf nodes, and obtain the decision tree regression model through pruning, evaluation and tuning processes.

6. The flexible production management method according to claim 1, characterized in that: The new order information includes material number, BOM, quantity, delivery date, UPH, material shortage status and production status; the production line information includes workshop identification, line number, estimated total production capacity and proficiency; the material management information includes bill of materials, inventory level and material storage location.

7. A flexible production management system, characterized in that: The system comprises: An information acquisition module, used to acquire order adjustment information when a new order is received, wherein the order adjustment information includes production order information, production line information and real-time production line information of the new order; A model calculation module, used for inputting the order adjustment information into a decision recommendation model to obtain decision recommendation information; wherein the decision recommendation model is integrated into the MES system; The plan adjustment module is used to adjust the production plan according to the decision recommendation information and to produce according to the adjusted production plan.

8. The flexible production management system according to claim 7, characterized in that: The construction of the decision recommendation model includes: Use big data technology to obtain multiple historical production information from the MES system, ERP system, and WMS system, and build a training set; Based on the training set, a regression algorithm is used for training to obtain a decision recommendation model.

9. The flexible production management system according to claim 8, characterized in that: The step of training based on the training set using a regression algorithm to obtain a decision recommendation model includes: Based on the training set, a decision tree regression model is used to calculate variables, and a linear regression algorithm is used to obtain a decision prediction result; By comparing the difference optimization algorithms, a decision recommendation model is obtained.

10. The flexible production management system according to claim 8, characterized in that: The construction of the decision tree regression model includes: S1, obtaining a data set for training a decision tree, wherein the data set includes a plurality of samples; S2, check whether all samples belong to the same category. If all samples of the current node belong to the same category, no further division is required, and leaf nodes are directly generated for category labeling; S3, when all samples are not of the same class, extract all possible features from the training set to form a feature candidate set, and determine whether all feature values ​​in the feature candidate set are unique; S4, if the values ​​of all features are unique, put them into the candidate data set until the feature set is extracted, generate leaf nodes and select the class with the largest proportion in the candidate data set for category labeling; S5, if the values ​​of all features are not unique, select the optimal partitioning feature according to CART and generate a decision node; S6, determining whether all features in the feature candidate set have been considered, if not, continuing to generate data subsets, and deleting the optimal partitioning feature to obtain a new data subset; using the new data subset as a data set for training the decision tree and executing S1; S7, if yes, stop splitting and generate leaf nodes, and obtain the decision tree regression model through pruning, evaluation and tuning processes.

Images