Economic batch decision-making method, service platform and medium for intermittent production system based on industrial big data

Based on industrial big data, using SVM and LSTM models combined with genetic algorithms to optimize production batch decisions, the problem of insufficient application of industrial big data in production logistics systems is solved, and the accuracy and intelligence of production decisions are improved. Reduced costs.

CN118278693BActive Publication Date: 2024-11-01JINAN UNIVERSITY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410466365.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-04-18
Publication Date
2024-11-01
Estimated Expiration
2044-04-18

AI Technical Summary

Technical Problem

In the prior art, industrial big data is insufficiently used in decision-making of production logistics systems, especially in the lack of effective integration of mass production planning and flexible production scheduling, resulting in the failure of sufficient improvement of the accuracy and intelligence level of decision-making models.

Method used

Adopting the economic batch decision-making method of batch production systems based on industrial big data, the multi-classification model of support vector machine (SVM) and the long and short-term memory network (LSTM) timing prediction model are constructed by combining genetic algorithms and iterative rules. , optimize production plans to cope with changes in product demand.

Benefits of technology

It has improved the intelligence level of multi-level and multi-scale management and control decisions in the production logistics system, enhanced the accuracy of model solving, reduced production and warehousing costs, and enhanced the core competitiveness of the enterprise.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118278693B_ABST
    Figure CN118278693B_ABST
Patent Text Reader

Abstract

The present application relates to an economic batch decision method, service platform and medium for an intermittent production system based on industrial big data. The method includes: obtaining currently received demand order information; processing the current product production information according to a preset support vector machine (SVM) multi-classification model to determine the product type; inputting the current product production information of the determined product type into a preset demand time series forecasting model to obtain first product demand forecast information; after importing the first product demand forecast information into the economic production batch decision model corresponding to the product type, using a preset genetic algorithm and iteration rules to calculate the economic production batch result, by introducing big data analysis technology to determine the key variables in the decision model, enhancing the accuracy of the model solution, and improving the intelligence level of multi-level and multi-scale management and control decision-making of the production logistics system under a complex random environment.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of intermittent production management, and in particular to an economic batch decision-making method, service platform, and medium for an intermittent production system based on industrial big data. Background Art

[0002] In recent years, as customer demands have become increasingly personalized, companies' production methods are shifting toward high-variety, small-batch, and even customized production. Decision-making is shifting from supplier-driven planning to customer-driven demand, with production plans based on customer orders. In the context of economic globalization, market competition among companies is intensifying. Relying solely on models or empirical experience for production planning is no longer sufficient to address the random fluctuations in customer demand. Rapidly understanding customer consumption trends, accurately determining product demand, and developing proactive, reliable, and robust flexible production plans to quickly meet customer needs and maximize overall profits have become crucial to enhancing a company's core competitiveness.

[0003] The deep integration of next-generation information technology and manufacturing is leading the global manufacturing industry towards a major industrial transformation characterized by informatization, digitalization, and intelligentization. With the widespread application of these information technologies, enterprises have rapidly accumulated large-scale, high-dimensional, and low-value industrial big data. However, due to long-standing limitations such as unclear analysis of business problems and the difficulty in closely integrating data analysis with model building, industrial big data has been far from fully developed and utilized. Therefore, how to closely integrate practical business problems with data analysis methods, optimize models through knowledge-driven methods, and effectively improve the quality of operational decisions, thereby achieving system cost reduction and efficiency improvement, has become a current research hotspot.

[0004] Among related technologies, although industrial big data technology has provided certain technical support for typical business scenarios such as production, logistics or supply chain, its application often only considers the prediction or target analysis of specific parameter variables, but ignores the relationship between big data and decision-making models, especially the lack of integration research with models such as batch production planning and flexible production scheduling; on the other hand, a comprehensive literature review represented by economic production batch problems found that most studies focused on the adjustment and improvement of decision-making model structures and target constraints, ignoring the role of industrial big data in improving the quality of decision-making in traditional production logistics systems.

[0005] How to apply industrial big data to decision-making models for specific business problems in related technologies and optimize management and control decisions through knowledge mining is an urgent problem that needs to be solved. Summary of the Invention

[0006] The embodiments of the present application provide an economic batch decision-making method, service platform and medium for an intermittent production system based on industrial big data, so as to at least solve the problem of how to apply industrial big data to the decision-making model of specific business problems in the related technology and optimize management and control decisions through knowledge mining.

[0007] The embodiment of the present application provides an economic batch decision-making method for an intermittent production system based on industrial big data, comprising: obtaining currently received demand order information, wherein the demand order information includes current product production information; processing the current product production information according to a preset support vector machine (SVM) multi-classification model to determine the product type, wherein the product type is used to characterize the sales popularity of the product corresponding to the current product production information; inputting the current product production information of the determined product type into a preset demand time series forecasting model to obtain first product demand forecast information, wherein the demand time series forecasting model is based on a long short-term memory network (LSTM) and preset product demand information data. A neural network trained by the set is trained to predict and output second product demand forecast information corresponding to the product production information based on the corresponding product production information, wherein the first product demand forecast information is used to characterize the demand for the product corresponding to the product type within a preset planning period; after the first product demand forecast information is imported into the economic production batch decision model corresponding to the product type, the economic production batch result is calculated using a preset genetic algorithm and iterative rules, wherein the economic production batch decision model is constructed based on the product's sales popularity and a preset production and inventory cost objective function, and the economic production batch result includes at least the production volume of the product corresponding to the product type within the planning period.

[0008] In a second aspect, an embodiment of the present application provides a service platform, a service platform comprising a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps of the economic batch decision-making method for intermittent production systems based on industrial big data as described in the first aspect above.

[0009] In a third aspect, an embodiment of the present application provides a storage medium on which a computer program is stored. When the program is executed by a processor, the steps of the economic batch decision-making method for an intermittent production system based on industrial big data as described in the first aspect above are implemented.

[0010] Compared with the related art, the economic batch decision-making method, service platform and medium for an intermittent production system based on industrial big data provided by the embodiment of the present application adopt the method of obtaining the currently received demand order information, wherein the demand order information includes the current product production information; the current product production information is processed according to the preset support vector machine SVM multi-classification model to determine the product type, wherein the product type is used to characterize the sales popularity of the product corresponding to the current product production information; the current product production information with the determined product type is input into the preset demand time series forecasting model to obtain the first product demand forecast information, wherein the demand time series forecasting model is a neural network trained based on the long short-term memory network LSTM and a preset product demand information data set, and is trained to predict and output the second product demand forecast information corresponding to the product production information according to the corresponding product production information, The first product demand forecast information is used to characterize the demand for products corresponding to the product type within a preset planning period; after the first product demand forecast information is imported into the economic production batch decision model corresponding to the product type, the economic production batch result is calculated using a preset genetic algorithm and iterative rules, wherein the economic production batch decision model is constructed based on product sales and a preset production and inventory cost objective function, and the economic production batch result includes at least the production volume of the product corresponding to the product type within the planning period, which solves the problem of how to apply industrial big data to the decision model of specific business problems in related technologies, optimizes management and control decisions through knowledge mining, determines key variables in the decision model by introducing big data analysis technology, enhances the accuracy of model solution, and improves the intelligence level of multi-level and multi-scale management and control decisions of production logistics systems in complex random environments.

[0011] The details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0013] Figure 1 This is a hardware structure block diagram of a terminal for an economic batch decision method for an intermittent production system based on industrial big data according to an embodiment of the present application;

[0014] Figure 2 This is a flow chart of an economic batch decision-making method for an intermittent production system based on industrial big data according to an embodiment of the present application;

[0015] Figure 3This is a structural block diagram of an economic batch decision-making device for an intermittent production system based on industrial big data according to an embodiment of the present application. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is described and illustrated below in conjunction with the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. Based on the embodiments provided in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application. In addition, it can also be understood that although the efforts made in this development process may be complex and lengthy, for ordinary technicians in the field related to the contents disclosed in the present application, some changes such as design, manufacturing or production based on the technical contents disclosed in the present application are only conventional technical means and should not be understood as the contents disclosed in the present application being insufficient.

[0017] References to "embodiments" in this application mean that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it refer to independent or alternative embodiments that are mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described in this application may be combined with other embodiments unless there is a conflict.

[0018] Unless otherwise defined, the technical terms or scientific terms involved in this application should be understood by people with ordinary skills in the technical field to which this application belongs. The words "one", "a", "the" and similar words involved in this application do not indicate quantitative restrictions and can represent the singular or plural. The terms "include", "comprise", "have" and any of their variations involved in this application are intended to cover non-exclusive inclusions; for example, a process, method, system, product or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units that are not listed, or may also include other steps or units inherent to these processes, methods, products or devices. The "multiple links" involved in this application refer to links that are greater than or equal to two. "And / or" describes the association relationship of associated objects, indicating that three relationships can exist. For example, "A and / or B" can represent: A exists alone, A and B exist at the same time, and B exists alone. The terms "first", "second", "third" and the like involved in this application are merely to distinguish similar objects and do not represent a specific ordering of objects.

[0019] The method embodiment provided in this embodiment can be executed in a terminal, a computer or a similar computing device. Taking running on a terminal as an example, Figure 1 This is a hardware block diagram of the terminal of the storage location planning method in the batch production mode of the embodiment of the present application. Figure 1 As shown, the terminal may include one or more ( Figure 1 Only one is shown) a processor 102 (the processor 102 may include but is not limited to a microprocessor MCU or a programmable logic device FPGA and other processing devices) and a memory 104 for storing data. Optionally, the terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It will be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above terminal. Figure 1 More or fewer components than shown, or with Figure 1 Different configurations shown.

[0020] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the method for inventory location planning in batch production mode in the embodiment of the present invention. The processor 102 executes the computer program stored in the memory 104 to execute various functional applications and data processing, thereby implementing the above-mentioned method. The memory 104 may include high-speed random access memory and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories may be connected to the terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0021] Transmission device 106 is used to receive or transmit data via a network. A specific example of the aforementioned network may include a wireless network provided by the communications provider of terminal 10. In one embodiment, transmission device 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to enable communication with the Internet. In another embodiment, transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0022] This embodiment provides a method for planning storage locations in batch production mode running on the above terminal. Figure 2 Flowchart of the method for warehouse location planning in batch production mode according to an embodiment of the present application. Figure 2As shown, the process includes the following steps:

[0023] Step S201: Acquire currently received demand order information, wherein the demand order information includes current product production information.

[0024] In this embodiment, the first problem to be solved is how to effectively integrate and process data sets from multiple different sources, formats, and attributes, such as order data, product data, and other current product production information; using endogenous association logic to construct a flat data set, and realizing time-series integration based on timestamps to form the original data set; considering that there may be a lot of noise in the data set, it is necessary to adopt data cleaning techniques, such as processing missing values ​​and outliers, deduplication, etc., to improve the accuracy, completeness and consistency of the data set and ensure data quality. Since big data has the characteristics of high dimensionality and large quantity, in order to ensure the efficiency and feasibility of data analysis, dimensionality reduction techniques (such as principal component analysis, local linear embedding, etc.) and feature selection techniques (such as recursive feature elimination, chi-square test, etc.) are adopted to achieve data compression; finally, according to actual business needs and analysis perspectives, the data set is segmented by business, time, and service scope dimensions to obtain a high-quality data set that can be used for analysis, that is, the currently received demand order information.

[0025] Step S202 : Process the current product production information according to a preset support vector machine (SVM) multi-classification model to determine the product type, wherein the product type is used to characterize the sales popularity of the product corresponding to the current product production information.

[0026] In this embodiment, product types are divided into best-selling, ordinary and slow-selling types according to the popularity of the products corresponding to the current product production information. Among them, best-selling products have the characteristics of large order demand, short average inventory time and frequent orders. When constructing their batch models, full equipment production is considered, and it is guaranteed that there will be no shortage of goods in each period during the planning period. The remaining finished products in the period are placed in the warehouse; ordinary products have the characteristics of moderate demand, long inventory time and slightly long order time. Since their order frequency is not high and the demand is large, in order to avoid a large amount of finished products piling up, the equipment can be not fully produced, and out of stock is not allowed; slow-selling products have the characteristics of small demand and long order intervals. The planning decision model of this type of products gives priority to inventory allocation, and allows batch production by backlog orders, which is reflected in the fact that there may be out of stock in a certain period during the planning period.

[0027] In step S203, the current product production information of the determined product type is input into a preset demand time series forecasting model to obtain first product demand forecast information, wherein the demand time series forecasting model is a neural network trained based on a long short-term memory network LSTM and a preset product demand information data set, and is trained to predict and output second product demand forecast information corresponding to the product production information based on the corresponding product production information. The first product demand forecast information is used to characterize the demand for the product corresponding to the product type within a preset planning period.

[0028] In this embodiment, based on the business needs of economic production batch decision-making, a product demand forecast set and a product sales classification set are obtained, and the product sales classification data is imported into a multi-classification model based on a support vector machine (SVM) for training. The classification results will be transmitted to the batch decision-making mechanism to provide data support for the model selection of the production plan; in addition, the product demand data set is imported into a time series data prediction model based on a long short-term memory network (LSTM) for iterative training. After the prediction results are obtained, they are imported into the production plan decision model and become the key parameters of the decision model.

[0029] Step S204, after importing the first product demand forecast information into the economic production batch decision model corresponding to the product type, the economic production batch result is calculated using the preset genetic algorithm and iteration rules, wherein the economic production batch decision model is constructed based on the product's sales popularity and the preset production and inventory cost objective function, and the economic production batch result at least includes the production volume of the product corresponding to the corresponding product type within the planning period.

[0030] In this embodiment, after determining the category to which the product belongs, the product demand estimate of the LSTM-based time series data prediction model is respectively brought into the economic production batch model of the three product types, and combined with the basic information collected in the early stage, the genetic algorithm is called for solution. By calculating the corresponding production, warehousing and other costs, the equipment combination for each time period in the planning cycle is obtained. Under the condition of meeting the number of iterations, the optimal economic production batch result is solved.

[0031] Through the above steps S201 to S204, the currently received demand order information is obtained, wherein the demand order information includes the current product production information; the current product production information is processed according to the preset support vector machine SVM multi-classification model to determine the product type, wherein the product type is used to characterize the sales popularity of the product corresponding to the current product production information; the current product production information of the determined product type is input into the preset demand time series forecasting model to obtain the first product demand forecast information, wherein the demand time series forecasting model is a neural network trained based on the long short-term memory network LSTM and the preset product demand information data set, and is trained to predict and output the second product demand forecast information corresponding to the product production information according to the corresponding product production information, and the first product demand forecast information is used to characterize The demand for products corresponding to the product type within a preset planning cycle; after importing the first product demand forecast information into the economic production batch decision model corresponding to the product type, the economic production batch result is calculated using the preset genetic algorithm and iterative rules, wherein the economic production batch decision model is constructed based on the product's sales popularity and the preset production and inventory cost objective function, and the economic production batch result at least includes the production volume of the products corresponding to the product type within the planning cycle, which solves the problem of how to apply industrial big data to the decision model of specific business problems in related technologies, optimizes management and control decisions through knowledge mining, determines the key variables in the decision model by introducing big data analysis technology, enhances the accuracy of model solution, and improves the intelligence level of multi-level and multi-scale management and control decisions of production logistics systems in complex random environments.

[0032] In some embodiments, the current product production information is processed according to a preset support vector machine (SVM) multi-classification model to determine the product type, which is achieved through the following steps:

[0033] Step 21 : Generate an original data set based on current product production information, wherein the original data set is production data including at least order data, product data, and equipment data.

[0034] In this example, the first challenge is how to effectively integrate and process datasets from diverse sources, formats, and attributes, such as order data, product data, and other current product production information. This involves constructing a flat dataset using inherent associative logic and implementing time-series integration based on timestamps to form a raw data set.

[0035] Step 22: clean, compress, and segment the original data set to obtain a first data set, wherein the first data set includes at least one product classification set based on best-selling performance.

[0036] In this embodiment, considering that there may be a large amount of noise in the data set, it is necessary to adopt data cleaning techniques, such as processing missing values ​​and outliers, deduplication, etc., to improve the accuracy, completeness and consistency of the data set and ensure data quality. Since big data has the characteristics of high dimensionality and large quantity, in order to ensure the efficiency and feasibility of data analysis, dimensionality reduction techniques (such as principal component analysis, local linear embedding, etc.) and feature selection techniques (such as recursive feature elimination, chi-square test, etc.) are adopted to achieve data compression. Finally, according to actual business needs and analysis perspectives, the data set is segmented by dimensions such as business, time and service scope to obtain a high-quality data set that can be used for analysis, that is, the demand order information currently received.

[0037] Step 23 : According to a preset selection rule, a product category set and a corresponding best-selling degree are selected from the first data set to obtain a product type, wherein the selection rule is generated based on the economic production batch decision requirement.

[0038] In this embodiment, the corresponding selection rule indicates that the first data set is classified according to different segmentation dimensions, and after the classification is completed, at least a best-selling classification set is generated.

[0039] Through the above steps, an original data set is generated based on the current product production information, and the original data set is production data that includes at least order data, product data and equipment data; the original data set is cleaned, compressed and segmented to obtain a first data set, and the first data set includes at least one product classification set with best-selling degree; according to preset selection rules, product classification sets and corresponding best-selling degrees are selected from the first data set to obtain product types, and the selection rules are generated according to the economic production batch decision-making requirements, so as to realize data processing and mining of current product production information to become a high-quality data set that provides a decision-making model.

[0040] It should be noted that after obtaining the current product production information, it is necessary to solve the problem of effectively integrating and processing data sets from various sources, formats, and attributes, such as order data and product data; construct a flat data set by adopting endogenous association logic, and realize time-series integration according to timestamps to form the original data set; considering that there is a lot of noise in the original data set, data cleaning techniques such as missing values, outlier processing, and deduplication are adopted to improve the accuracy, completeness and consistency of the data set and ensure data quality; at the same time, since big data has the characteristics of high dimension and large quantity, in order to ensure the efficiency and feasibility of data analysis, dimensionality reduction techniques (such as principal component analysis, local linear embedding) and feature selection techniques (such as recursive feature elimination, chi-square test, etc.) are adopted to achieve data compression; finally, according to actual business needs and analysis perspectives, the data set is divided into dimensions such as business, time and service scope to obtain a high-quality first data set that can be used for analysis.

[0041] In some embodiments, generating an original data set based on current product production information is achieved through the following steps:

[0042] Step 31: Flatten the current product production information to obtain a flattened data set.

[0043] Step 32: Perform time-series integration on the flattened dataset according to timestamps to obtain the original dataset.

[0044] By flattening the current product production information in the above steps, a flat data set is obtained; the flat data set is time-series integrated according to the timestamp to obtain the original data set. By adopting the endogenous associative logic to construct the flat data set and realizing time-series integration according to the timestamp, the original data set is formed to provide accurate data for product demand forecasting.

[0045] In some embodiments, data cleaning, data compression, and data segmentation are performed on the original data to obtain a first data set, which is achieved by the following steps:

[0046] Step 41 : performing a data cleaning operation on the original data set to obtain a candidate data set, wherein the data cleaning includes one of the following: missing value interpolation, outlier removal, and duplicate removal.

[0047] Step 42 , sequentially performing dimensionality reduction and feature selection processing on the candidate dataset to generate a compressed dataset, wherein the dimensionality reduction operation includes one of the following: principal component analysis, local linear embedding, and the feature selection processing includes one of the following: recursive feature elimination, chi-square test.

[0048] Step 43 : Segment the compressed data set according to a preset segmentation dimension to generate a first data set, wherein the segmentation dimension includes at least one of the following: business category, time, and service scope.

[0049] In this embodiment, the compressed data set is segmented, for example, by performing business segmentation and time segmentation, thereby generating a classification set including the first demand quantity and the best-selling category.

[0050] By performing data cleaning operations on the original data set in the above steps, a candidate data set is obtained, and the data cleaning includes one of the following: missing value interpolation, outlier removal, and deduplication; dimensionality reduction operations and feature selection processing are performed on the candidate data set in turn to generate a compressed data set, and the dimensionality reduction operation includes one of the following: principal component analysis and local linear embedding, and the feature selection processing includes one of the following: recursive feature elimination and chi-square test; the compressed data set is segmented according to a preset segmentation dimension to generate a first data set, and the segmentation dimension includes at least one of the following: business category, time, and service scope, thereby achieving the generation of a high-quality data set.

[0051] In some embodiments, the first data set further includes a first demand corresponding to current product production information. The current product production information of the determined product type is input into a preset demand time series forecasting model to obtain first product demand forecast information, which is achieved by the following steps:

[0052] Step 51: Obtain the first demand quantity corresponding to the product of each product type, and use the first demand quantity as the product demand forecast set.

[0053] In this embodiment, the first demand quantity is the production quantity corresponding to each product produced in the current period or the production quantity to be produced. The corresponding production quantity serves as the input data corresponding to the product demand forecast in the future planning period, and can also be understood as the predicted sample data.

[0054] Step 52: Input the product demand forecast set into the demand time series forecast model to obtain first product demand forecast information.

[0055] Through the above steps, the first demand quantity corresponding to the product of each product type is obtained, and the first demand quantity is used as the product demand forecast set; the product demand forecast set is input into the demand time series forecast model to obtain the first product demand forecast information, so as to realize the prediction of product demand in the corresponding planning period in the future based on the current time demand.

[0056] In some embodiments, the training of the demand time series forecasting model includes the following steps:

[0057] Step 61: pre-processing a preset demand time series data sample set to obtain a sample set, wherein the demand time series data sample set is used to represent historical demand data of the product;

[0058] Step 62: Split the sample set according to a preset ratio to generate a training sample set and a test sample set.

[0059] In this embodiment, the preset ratio is that the training sample set accounts for 80% and the test sample set accounts for 20%, and in this process, the data set needs to be randomly disrupted to ensure the balance and randomness of the training sample set and the test sample set.

[0060] Step 63: Based on the training sample set, the constructed LSTM network is trained, and the trained LSTM network is tested using the test sample set until the LSTM network converges to obtain a demand time series prediction model.

[0061] In this embodiment, LSTM is a recursive neural network structure in artificial neural networks that can memorize and process long-term dependencies of input sequences. In this embodiment, the advantage of using the LSTM network is that it can process time series data with long-term dependencies, and compared with traditional time series analysis methods, LSTM can adaptively learn the patterns of time series data. The calculation formula of the LSTM network is as follows:

[0062] F t =σ(X t W f +H t-1 W f +b f ) (1)

[0063] I t =σ(X t W i +H t-1 W i +b i ) (2)

[0064]

[0065]

[0066] O t =σ(X t W o +H t-1 W o +b o ) (5)

[0067] H t =O t ⊙tanh(C t ) (6)

[0068] Among them, formula (1) represents the forget gate; formula (2) represents the input gate; formula (3) represents the candidate memory cell; formula (4) represents the update of the memory cell; formula (5) represents the output gate; formula (6) represents the hidden state.

[0069] In addition, to accurately evaluate the performance and effect of the model, model evaluation indicators such as Mean Absolute Percentage Error (MAPE), Mean Absolute Error (MAE), and Mean Square Error (MSE) are introduced. The calculation formula is as follows:

[0070]

[0071]

[0072]

[0073] Among them, E i Indicates the actual value Y i and predicted value The error is n, and n represents the number of sample data. The smaller the evaluation value, the better the model fitting effect.

[0074] In some embodiments, the preset demand time series data sample set is preprocessed by the following steps:

[0075] Step 71 : normalize the demand time series data sample set to obtain a normalized data set.

[0076] Step 72 : converting the normalized data set into a standard three-dimensional array format to obtain a three-dimensional data set, wherein the sample set includes the three-dimensional data set.

[0077] In this embodiment, the standard three-dimensional array format includes the number of samples, the number of time steps, and the number of features. At the same time, it can be understood that by normalizing the demand time series data sample set (corresponding to the product historical demand data) and then converting the normalized data set into the standard three-dimensional array format, the demand time series forecasting model can process data more accurately in the time and feature dimensions.

[0078] In some embodiments, after the first product demand forecast information is imported into the economic production batch decision model corresponding to the product type, the economic production batch result is calculated using a preset genetic algorithm and iterative rules, which is achieved by the following steps:

[0079] Step 81: Combine the economic production batch decision models corresponding to the product types to construct a comprehensive economic production batch decision model. Import the first product demand forecast information into the comprehensive economic production batch decision model for processing to obtain economic production batch data, wherein the economic production batch result includes the target economic production batch data. The comprehensive economic production batch decision model is constructed as follows:

[0080] Min Cost = C p +C h +C s +C l (10)

[0081]

[0082]

[0083]

[0084]

[0085]

[0086]

[0087]

[0088]

[0089]

[0090]

[0091]

[0092]

[0093] Among them, model A, model B and model C represent three economic batch models corresponding to product types; T represents the number of time periods in the planning cycle, C p Production cost during the planning period, C h represents the storage cost during the planning period, C s represents the equipment adjustment cost during the planning period, C l represents the penalty cost of delay within the planning period, P ij represents the unit production cost of the j-th type i equipment, K ij represents the unit adjustment cost of the j-th type i equipment, Q ij represents the production volume of the j-th type i equipment, h t It represents the unit storage cost of the product in period t, I t represents the inventory at the end of period t, X t represents the planned production batch for period t, d t represents the actual demand or product forecast quantity during period t, C ij represents the capacity of the jth type i device, l t represents the unit delay penalty cost in period t, d t - Indicates that the quantity produced during period t is less than the quantity demanded, d t + Indicates that the quantity produced during period t exceeds the quantity demanded. represents a 0-1 variable, indicating whether the jth device of type i is selected within period t; T represents the set of planning periods, |T| is the number of periods in the period in set T; I represents the set of equipment types, |I| is the number of equipment types in set I; J represents the set of equipment quantities, |J| is the number of equipment in set J; Formula (10) represents the minimum sum of equipment production cost, equipment adjustment cost, storage cost and delay penalty cost within the planning period; Formula (11) represents production cost; Formula (12) represents storage cost; Formula (13) represents equipment adjustment cost; Formula (14) represents the penalty cost for delay; formula (15) represents the quantitative relationship between the production batch of different best-selling models and the existing inventory and current demand; formula (16) represents whether the different best-selling models are fully equipped for production; formula (17) represents the inventory balance relationship of different best-selling models; formula (18) represents that the batch size is equal to the sum of the production volumes of the selected equipment; formula (19) represents the production-demand balance relationship within the planning cycle; formula (20) represents the size of the backlog of orders; formula (21) represents that the inventory is not negative at any time; formula (22) represents that the batch size is not negative at any time.

[0094] Step 82: Based on the preset genetic algorithm and iteration rules, the economic production batch data is genetically evolved and iteratively processed to generate target economic production batch data.

[0095] In this embodiment, it should be understood that the use of genetic algorithms for solving problems is well known and clear to those skilled in the art. By adopting genetic algorithms and encoding economic production batch data, genetic evolution operations (for example, crossover and mutation) are performed until a preset number of iterations are performed, and then target economic production batch data are generated; in this embodiment, the specific genetic operations are not limited.

[0096] The following uses a certain intermittent production enterprise as an example to illustrate the feasibility and rationality of the economic batch decision-making method for intermittent production systems based on industrial big data proposed in this application.

[0097] Planners receive current customer orders and product demand forecasts. Table 1 shows this demand information. The planning cycle is divided into three time periods: Time periods T1 and T2 represent currently placed customer orders, while T3 represents the demand forecast. Regarding equipment information, adjustment fees, unit production costs, and equipment quantity vary depending on equipment capacity, as shown in Table 2. Related cost items include unit warehousing costs and unit tardiness costs, as shown in Table 3.

[0098] Table 1 Customer demand in each period of the planning cycle

[0099]

[0100] Table 2 Equipment information

[0101]

[0102] Table 3 Related costs

[0103]

[0104] After necessary desensitization processing, more than 3,000 real sales data of a certain type of product in the enterprise over the past 10 years were selected as experimental data to verify the feasibility of the method.

[0105] To verify the superiority of the demand forecasting method proposed in this paper, the Gradient Boosting Decision Tree (GBDT), Support Vector Regression (SVR), and Multi-Layer Perceptron (MLP) algorithms were selected for performance comparison. Since the estimated value of the proposed method is closer to the actual value, it is proved that the method can effectively predict the future trend of the product. The performance of the four algorithms in terms of evaluation indicators was further compared. According to the comparison results, it can be seen that the economic batch decision method for intermittent production systems based on industrial big data proposed in this application is lower than the GBDT, SVR, and MLP algorithms in all indicators, further proving the effectiveness and reliability of the method, as shown in Table 3. Finally, the trained LSTM model was used to predict the product demand in the T3 stage of the planning cycle. The prediction results are shown in Table 5.

[0106] Table 4 Algorithm error comparison

[0107]

[0108] Table 5 Product demand during period T3

[0109]

[0110] A genetic algorithm was used to solve the three economic batch models, with the results shown in Table 6. Subsequently, the total cost of rule-based production planning decisions was compared with that of big data-driven economic batch decision-making. Manual decision-making primarily relies on fixed rules, such as the requirement that actual production volume during each period of the planning cycle must meet customer demand, with no delays allowed and all products produced to stock. Table 7 shows a cost comparison between the big data and rule-based decision-making solutions. As can be seen, for batch planning problems involving products of varying marketability, the proposed solution achieved total cost savings of 7.55%, 1.76%, and 2.59%, respectively, compared to rule-based production planning decisions.

[0111] Table 5 Comparison between economic production batch size driven by big data and actual demand

[0112]

[0113] Table 7 Cost comparison between big data solution and rule-based solution

[0114]

[0115] First, a comparison of the various costs for the best-selling product plans reveals that, because the big data-driven approach can accurately predict demand for this type of product, most customer orders are produced in large batches. This results in lower unit production costs due to the high-volume equipment used. Furthermore, the smaller number of machines selected results in lower equipment adjustment costs. A portion of the finished product in each period is used directly to meet current customer demand, while the remaining portion is temporarily stored in the warehouse and will be sold soon in the next period, eliminating the need for ongoing high storage costs.

[0116] Secondly, a comparison of various costs across planning schemes for common products reveals that for products with low demand and moderate order frequency, planners often rely on manual production to minimize inventory. While this strategy can reduce warehousing costs, it doesn't scale, leading to a sharp increase in production costs. Based on demand forecasts, the big data-driven approach avoids the pursuit of full capacity and adopts a flexible batch configuration. This approach maximizes the benefits of batch planning by dynamically balancing production and warehousing costs within the planning period.

[0117] Finally, a comparison of decision-making options for slow-moving products reveals that the big data-driven approach, by predicting future product demand and consolidating orders through delayed delivery to achieve batch production, significantly reduces the number of times ultra-small equipment is used, as this type of equipment has a low level of automation and requires a large number of people to complete production. Therefore, under the premise of accurately predicting product demand, batch production can be achieved by paying a certain penalty cost for delayed delivery, thereby achieving overall cost savings.

[0118] It should be noted that this application proposes an economic batch intelligent decision-making method for intermittent production systems driven by industrial big data. Based on the digital twin image system, it expands the research on big data-driven intelligent decision-making service systems and explores decision-making optimization methods driven by data model fusion. The case is based on the planning and decision-making process of intermittent production enterprises, and discusses the implementation methods, steps and application effects of economic production batch decision-making methods under uncertain demand, including: 1. Economic batch decision-making process design under digital-analog fusion; 2. LSTM method predicts product demand, and horizontally compares the application effects of three algorithms: GBDT, SVR and MLP; 3. Constructs and solves economic production batch decision models for three types of products with different sales volume. The results show that the big data-driven solution proposed in the embodiment of this application can accurately grasp customer demand trends, reduce system resource waste and operating costs in a proactive decision-making manner, and improve the effect and efficiency of economic production batch decision-making.

[0119] This embodiment also provides a production and storage collaborative planning device under an intermittent production mode, which is used to implement the above-mentioned embodiments and preferred implementation methods, and the details that have been described will not be repeated. As used below, the terms "module", "unit", "sub-unit", etc. can be a combination of software and / or hardware that can implement a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation by hardware, or a combination of software and hardware, is also possible and conceivable.

[0120] Figure 3 This is a structural block diagram of an economic batch decision-making device for an intermittent production system based on industrial big data according to an embodiment of the present application. Figure 3 As shown, the device includes an acquisition module 31, a classification module 32, a prediction module 33 and a processing module 34, wherein:

[0121] An acquisition module 31 is configured to acquire currently received demand order information, wherein the demand order information includes current product production information;

[0122] The classification module 32 is coupled to the acquisition module 31 and is configured to process the current product production information according to a preset support vector machine (SVM) multi-classification model to determine a product type, wherein the product type is used to characterize the marketability of the product corresponding to the current product production information;

[0123] The prediction module 33 is coupled to the classification module 32 and is configured to input the current product production information of the determined product type into a preset demand time series prediction model to obtain first product demand forecast information, wherein the demand time series prediction model is a neural network trained based on a long short-term memory network (LSTM) and a preset product demand information dataset, and is trained to predict and output second product demand forecast information corresponding to the product production information based on the corresponding product production information, wherein the first product demand forecast information is used to represent the demand for the product corresponding to the product type within a preset planning period;

[0124] The processing module 34 is coupled to the forecasting module 33 and is used to calculate the economic production batch results by using a preset genetic algorithm and iteration rules after importing the first product demand forecast information into the economic production batch decision model corresponding to the product type, wherein the economic production batch decision model is constructed based on the product's sales popularity and a preset production and inventory cost objective function, and the economic production batch results at least include the production volume of the product corresponding to the product type within the planning period.

[0125] In some embodiments, the classification module 32 further includes:

[0126] A generating unit, configured to generate an original data set based on current product production information, wherein the original data set is production data including at least order data, product data, and equipment data;

[0127] a processing unit coupled to the generating unit and configured to clean, compress, and segment the original data set to obtain a first data set, wherein the first data set includes at least one product classification set based on best-selling degree;

[0128] The classification unit is coupled to the processing unit and is used to select a product classification set and a corresponding sales popularity from the first data set according to a preset selection rule to obtain a product type, wherein the selection rule is generated according to the economic production batch decision requirement.

[0129] In some embodiments, the generating unit is further configured to flatten the current product production information to obtain a flattened data set; and perform time-series integration on the flattened data set according to timestamps to obtain an original data set.

[0130] In some embodiments, the processing unit is also used to perform data cleaning operations on the original data set to obtain a candidate data set, wherein the data cleaning includes one of the following: missing value interpolation, outlier removal, and deduplication; performing dimensionality reduction operations and feature selection processing on the candidate data set in sequence to generate a compressed data set, wherein the dimensionality reduction operation includes one of the following: principal component analysis, local linear embedding, and the feature selection processing includes one of the following: recursive feature elimination, chi-square test; the compressed data set is segmented according to a preset segmentation dimension to generate a first data set, wherein the segmentation dimension includes at least one of the following: business category, time, and service scope.

[0131] In some embodiments, the first data set also includes a first demand corresponding to current product production information, and the prediction module 33 further includes:

[0132] an acquiring unit, configured to acquire a first demand quantity corresponding to a product of each product type, and use the first demand quantity as a product demand forecast set;

[0133] The prediction unit is coupled to the acquisition unit and is used to input the product demand prediction set into the demand time series prediction model to obtain the first product demand prediction information.

[0134] In some embodiments, the planning device is also used to preprocess a preset demand time series data sample set to obtain a sample set, wherein the demand time series data sample set is used to characterize the historical demand data of the product; the sample set is divided according to a preset ratio to generate a training sample set and a test sample set; based on the training sample set, the constructed LSTM network is trained, and the trained LSTM network is tested using the test sample set until the LSTM network converges to obtain a demand time series prediction model.

[0135] In some embodiments, the planning device is further used to normalize the demand time series data sample set to obtain a normalized data set; convert the normalized data set into a standard three-dimensional array format to obtain a three-dimensional data set, wherein the sample set includes the three-dimensional data set.

[0136] In some embodiments, the planning device is also used to, after importing the first product demand forecast information into the economic production batch decision model corresponding to the product type, merge the economic production batch decision models corresponding to the product type to construct a comprehensive economic production batch decision model, import the first product demand forecast information into the comprehensive economic production batch decision model for processing to obtain economic production batch data; based on the preset genetic algorithm and iterative rules, perform genetic evolution and iterative processing on the economic production batch data to generate target economic production batch data.

[0137] This embodiment further provides a service platform, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0138] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0139] Optionally, in this embodiment, the processor may be configured to execute the following steps through a computer program:

[0140] S1, obtaining currently received demand order information, wherein the demand order information includes current product production information.

[0141] S2. Process the current product production information according to a preset support vector machine (SVM) multi-classification model to determine a product type, wherein the product type is used to characterize the sales popularity of the product corresponding to the current product production information.

[0142] S3. Input the current product production information of the determined product type into a preset demand time series forecasting model to obtain first product demand forecast information, wherein the demand time series forecasting model is a neural network trained based on a long short-term memory network LSTM and a preset product demand information data set, and is trained to predict and output second product demand forecast information corresponding to the product production information based on the corresponding product production information. The first product demand forecast information is used to characterize the demand for the product corresponding to the product type within a preset planning period.

[0143] S4. After importing the first product demand forecast information into the economic production batch decision model corresponding to the product type, the economic production batch result is calculated using the preset genetic algorithm and iterative rules, wherein the economic production batch decision model is constructed based on the product's sales popularity and the preset production and inventory cost objective function, and the economic production batch result at least includes the production volume of the product corresponding to the product type within the planning period.

[0144] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be repeated here.

[0145] In addition, in conjunction with the material distribution method for a mixed-flow assembly workshop in the above-mentioned embodiments, embodiments of the present application may provide a storage medium for implementation. The storage medium stores a computer program; when the computer program is executed by a processor, it implements any of the material distribution methods for a mixed-flow assembly workshop in the above-mentioned embodiments.

[0146] Those skilled in the art should understand that the technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0147] The above embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. An economic batch decision method for intermittent production system based on industrial big data, characterized in that: include: Acquire the currently received demand order information, wherein the demand order information includes current product production information; According to a preset support vector machine (SVM) multi-classification model, the current product production information is processed to determine the product type, wherein the product type is used to characterize the popularity of the product corresponding to the current product production information; Inputting the current product production information of the determined product type into a preset demand time series forecasting model to obtain first product demand forecasting information, wherein the demand time series forecasting model is a neural network trained based on a long short-term memory network LSTM and a preset product demand information data set, and is trained to predict and output second product demand forecasting information corresponding to the product production information according to the corresponding product production information, wherein the first product demand forecasting information is used to characterize the demand for the product corresponding to the product type within a preset planning period; After the first product demand forecast information is imported into the economic production batch decision model corresponding to the product type, the economic production batch result is calculated using a preset genetic algorithm and iteration rules, wherein the economic production batch decision model is constructed based on product sales and a preset production and inventory cost objective function, and the economic production batch result at least includes the production volume of the product corresponding to the product type within the planning period; The current product production information is processed according to a preset support vector machine (SVM) multi-classification model to determine the product type, including: Based on the current product production information, an original data set is generated, wherein the original data set is production data including at least order data, product data and equipment data; Performing data cleaning, data compression and data segmentation on the original data set to obtain a first data set, wherein the first data set includes at least one product classification set with best-selling degree; According to a preset selection rule, the product classification set and the corresponding best-selling degree are selected from the first data set to obtain the product type, wherein the selection rule is generated according to the economic production batch decision requirement; Wherein, based on the current product production information, an original data set is generated, including: Planarizing the current product production information to obtain a planarized data set; Performing time-series integration on the flattened data set according to timestamps to obtain the original data set; The raw data is cleaned, compressed and segmented to obtain a first data set, including: Performing a data cleaning operation on the original data set to obtain a candidate data set, wherein the data cleaning includes one of the following: missing value interpolation, outlier removal, and duplication removal; Performing dimensionality reduction and feature selection processing on the candidate data set in sequence to generate a compressed data set, wherein the dimensionality reduction operation includes one of the following: principal component analysis and local linear embedding, and the feature selection processing includes one of the following: recursive feature elimination and chi-square test; The compressed data set is segmented according to a preset segmentation dimension to generate the first data set, wherein the segmentation dimension includes at least one of the following: business category, time, and service scope; After the first product demand forecast information is imported into the economic production batch decision model corresponding to the product type, the economic production batch result is calculated using a preset genetic algorithm and iteration rules, including: The economic production batch decision models corresponding to the product types are combined to form a comprehensive economic production batch decision model, which is constructed as follows: , , , , , , , , , , , , , Among them: Model A, Model B and Model C represent three economic batch models corresponding to product types; T represents the number of time periods in the planning cycle, C p Production cost during the planning period, C h represents the storage cost during the planning period, C s represents the equipment adjustment cost during the planning period, C l represents the penalty cost of delay within the planning period, P ij represents the unit production cost of the j-th type i equipment, K ij represents the unit adjustment cost of the j-th type i equipment, Q ij The production volume of the jth type i equipment, h t It represents the unit storage cost of the product in period t, I t represents the inventory at the end of period t, X t represents the planned production batch for period t, d t represents the actual demand or product forecast quantity during period t, C ij represents the capacity of the j-th type i device, l t represents the unit delay penalty cost in period t, It means that the quantity produced in period t is less than the quantity demanded. It means that the quantity produced in period t is more than the quantity demanded. represents a 0-1 variable, indicating whether the jth device of type i is selected in period t; |T| is the number of periods in the cycle in set T; I represents the set of device types, |I| is the number of device types in set I; J represents the set of device numbers, |J| is the number of devices in set J; Importing the first product demand forecast information into the comprehensive economic production batch decision model for processing to obtain economic production batch data; Based on a preset genetic algorithm and iteration rules, the economic production batch data is genetically evolved and iteratively processed to generate target economic production batch data, wherein the economic production batch result includes the target economic production batch data.

2. The method according to claim 1, characterized in that: The first data set also includes a first demand corresponding to the current product production information. The current product production information of the determined product type is input into a preset demand time series forecasting model to obtain the first product demand forecasting information, including: Obtaining the first demand quantity corresponding to the product corresponding to each of the product types, and using the first demand quantity as a product demand forecast set; The product demand forecast set is input into the demand time series forecast model to obtain the first product demand forecast information.

3. The method according to claim 1, characterized in that The training of the demand time series forecasting model includes: Preprocessing a preset demand time series data sample set to obtain a sample set, wherein the demand time series data sample set is used to characterize product historical demand data; Splitting the sample set according to a preset ratio to generate a training sample set and a test sample set; Based on the training sample set, the constructed LSTM network is trained, and the trained LSTM network is tested using the test sample set until the LSTM network converges to obtain the demand timing prediction model.

4. The method according to claim 3, characterized in that Preprocess the preset demand time series data sample set, including: Normalizing the demand time series data sample set to obtain a normalized data set; The normalized data set is converted into a standard three-dimensional array format to obtain a three-dimensional data set, wherein the sample set includes the three-dimensional data set.

5. A service platform, comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps of the economic batch decision-making method for an intermittent production system based on industrial big data as described in any one of claims 1 to 4.

6. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the economic batch decision method for an intermittent production system based on industrial big data described in any one of claims 1 to 4 are implemented.