Product production whole-process management method and system based on multi-source data

Abstract

The application discloses a product production whole-process management method and system based on multi-source data, and relates to the field of product production process management.The method comprises the following steps: receiving a product model of a target product, and acquiring product performance requirements of the target product; acquiring a same product production record set according to the product model and a production line number, each same product production record comprising process parameter data of multiple production nodes on a target product production line; analyzing process parameter fluctuation ranges of the production nodes based on the process parameter data of the production nodes in the multiple same product production records, and determining multi-parameter fluctuation constraint intervals of the production nodes; and within the multi-parameter fluctuation constraint intervals, acquiring optimal production parameters by optimization with the product performance requirements as optimization targets, and configuring the target product production line to produce the target product.The application solves the problems of low product production efficiency and high product scrap rate of the production line through the product production whole-process management method and system based on multi-source data.

Description

Technical Field

[0001] This invention relates to the field of product manufacturing process management, specifically to a method and system for managing the entire product manufacturing process based on multi-source data. Background Technology

[0002] Traditional product manufacturing process management relies on human experience, which can easily lead to product quality defects, affect production process management, and result in low production efficiency and high scrap rates. Therefore, process management methods are needed to reduce human experience differences or judgment errors and optimize product quality. Summary of the Invention

[0003] This application provides a product manufacturing process management method and system based on multi-source data, aiming to solve the problems of outdated product manufacturing process management methods in the prior art, which cause equipment risks and production quality defects, resulting in low production efficiency and high product scrap rate on the production line.

[0004] In view of the above problems, this application provides a method and system for managing the entire product manufacturing process based on multi-source data.

[0005] Firstly, this application provides a method for managing the entire product manufacturing process based on multi-source data, including:

[0006] Receive the product model of the target product, determine the target product production line and production line number of the product model, and obtain the product performance requirements of the target product based on the product model;

[0007] Based on the product model and the production line number, retrieve the same product production record set. The same product production record set includes multiple same product production records. Each same product production record includes process parameter data of multiple production nodes on the target product production line.

[0008] Based on the process parameter data of each production node in the multiple production records of the same product, analyze the fluctuation range of process parameters of each production node, and determine the multi-parameter fluctuation constraint range of each production node.

[0009] Within the multi-parameter fluctuation constraint range of each production node, the optimal production parameters of each production node are obtained with the product performance requirements as the optimization objective. The target product production line is then configured to produce the target product. The product performance requirements are split into the optimization objective splitter, which is constructed using a BP neural network. The optimal production parameters are tested using a joint node twin model and configured when the product performance requirements are met.

[0010] Secondly, this application provides a product manufacturing process management system based on multi-source data, including:

[0011] The product model acquisition module is used to receive the product model of the target product, determine the target product production line and production line number of the product model, and obtain the product performance requirements of the target product based on the product model.

[0012] The product production record retrieval module is used to retrieve the same product production record set according to the product model and the production line number. The same product production record set includes multiple same product production records, and each same product production record includes process parameter data of multiple production nodes on the target product production line.

[0013] The process parameter fluctuation range analysis module is used to analyze the process parameter fluctuation range of each production node based on the process parameter data of each production node in the multiple production records of the same product, and to determine the multi-parameter fluctuation constraint range of each production node.

[0014] The optimal production parameter acquisition module is used to find the optimal production parameters for each production node within the multi-parameter fluctuation constraint range of each production node, with the product performance requirements as the optimization objective, and configure the target product production line to produce the target product. The optimization objective splitter is used to split the product performance requirements for optimization. The optimization objective splitter is constructed using a BP neural network. The optimal production parameters are tested using a joint node twin model and configured when the product performance requirements are met.

[0015] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0016] This application provides a method and system for managing the entire product manufacturing process based on multi-source data, fundamentally reducing batch product quality problems caused by differences in human experience or judgment errors. Simultaneously, by employing an optimization architecture using an optimized target splitter and a joint twin model, parallel optimization is performed using their respective digital twin models and fluctuation constraint ranges, solving the problem of parameter mismatch in product manufacturing and ensuring consistent product performance. Furthermore, thorough simulation and verification using the joint twin model avoids potential quality defects and equipment conflicts, addressing the problems of outdated product manufacturing process management methods in existing technologies, which lead to equipment risks and production quality defects, resulting in low production efficiency and high scrap rates on the production line. This enables rapid response to production tasks, resulting in higher and more stable product quality, reduced production resource consumption, fewer product quality problems caused by judgment errors, and full assurance of product performance. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating a product manufacturing process management method based on multi-source data;

[0019] Figure 2 This is a schematic diagram of the structure of a product manufacturing process management system based on multi-source data.

[0020] The components represented by each number in the attached diagram are explained below:

[0021] Product model acquisition module 11; Product production record retrieval module 12; Process parameter fluctuation range analysis module 13; Optimal production parameter acquisition module 14. Detailed Implementation

[0022] This application provides a product manufacturing process management method and system based on multi-source data, which is used to address the problems of outdated product manufacturing process management methods in the prior art, which cause equipment risks and production quality defects, resulting in low production efficiency and high product scrap rate on the production line.

[0023] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0024] It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to these processes, methods, products, or devices.

[0025] Example 1, as Figure 1 As shown, this application provides a product manufacturing process management method based on multi-source data, the method comprising:

[0026] S10: Receive the product model of the target product, determine the target product production line and production line number of the product model, and obtain the product performance requirements of the target product based on the product model;

[0027] In this embodiment of the application, the production line and production number of the target product are retrieved from the production record database. At the same time, based on the product model, the product performance requirements such as physical attributes, functional indicators or quality parameters are obtained from customer orders, design specifications, industry standards or historical data, providing data support for the management process.

[0028] Step S10 in the method provided in this application embodiment includes:

[0029] A first search condition is constructed based on the product model, and a second search condition is constructed based on the production line number;

[0030] Connect to the production record database, and match production records in the production record database according to the first search condition and the second search condition to obtain multiple production records of the same product;

[0031] The production records of the same product are summarized to obtain the production record set of the same product.

[0032] Specifically, the product model can include a version number and a variant code. A first search condition is constructed based on the obtained product model, and a second search condition is constructed based on the production line number. The first search condition, based on the product model, ensures that only production records of products with the same model are retrieved; the second search condition, based on the production line number, ensures that the retrieved production records come from the same production line.

[0033] For example, in the antenna array unit production line, the antenna array unit model is HR-X03K13-231. The first search condition is constructed using the product model HR-X03K13-231, and the second search condition is constructed using the production line number A4-B3-76C.

[0034] Furthermore, by connecting to the production record database and utilizing historical production record data, the database is matched against the first and second search criteria to obtain multiple production records for the same product. These production records reflect the production status of the product.

[0035] Furthermore, multiple production records for the same product are aggregated to obtain a set of production records for the same product. This set refers to the data collection recorded when the same or similar products were produced on the same production line in the past. By integrating multi-source data on production records for the same product, the production effect of the same production line producing the same or similar products can be obtained, facilitating subsequent production process management. The size of the set of production records for the same product affects the reliability of the analysis. If the number of records obtained from the search is too small, the database is expanded to obtain a set of records for the same product model.

[0036] For example, in the antenna array unit production line, the antenna array unit model is HR-X03K13-231. Using the product model HR-X03K13-231 and the production line number A4-B3-76C as the first and second search criteria, the records of the most recent 100 productions are retrieved.

[0037] In this embodiment of the application, step S10 involves matching the first and second search conditions in the production record database to obtain multiple production records of the same product, including:

[0038] Extract the first production record from the production record database, and retrieve the product model and production line number of the first production record to obtain the first product model and the first production line number;

[0039] When the first product model meets the first search condition and the first production line number meets the second search condition, the first production record is added to the plurality of production records of the same product.

[0040] Specifically, the system utilizes an API to access the production record database, extracts the first production record from it, and retrieves the product model and production line number from that record to obtain the first product model and production line number. Multiple batch extractions can be performed when extracting the first production record to improve efficiency while ensuring data accuracy.

[0041] For example, if the database retrieves 100 records, the system will start from the first record and extract the product model and production line number of the first 100 production records in batches.

[0042] Furthermore, when the first product model meets the first search condition and the first production line number meets the second search condition, the first production record is added to multiple production records of the same product. If the conditions are not met, the first production record will be removed, and the search condition judgment for the next first production record will continue.

[0043] For example, 100 records are retrieved from the production record database. After judging by the first and second search conditions, 80 first production records that meet the conditions are obtained and used as production records for the same product.

[0044] In this embodiment, the first and second search criteria are constructed using the product model and production line number, respectively. Then, production records are matched against the production record database using these criteria to obtain multiple production records for the same product, which are then aggregated to form a set of production records for the same product. This establishes the data foundation for the entire management process, ensuring that all subsequent data retrieval, parameter analysis, and optimization calculations are based on accurate product information.

[0045] S20: Retrieve a set of production records for the same product based on the product model and the production line number. The set of production records for the same product includes multiple production records for the same product, and each production record for the same product includes process parameter data for multiple production nodes on the target product production line.

[0046] In this embodiment, relevant historical product production records are filtered from a massive amount of production records using a dual matching condition of product model and production line number. The production records for the same product include process parameter data from multiple production nodes on the target product's production line. These production nodes include processing and assembly, and the process parameter data includes temperature, pressure, and time. This reflects the multi-source nature of the data in the production record set for the same product, allowing for subsequent multi-source data integration to provide a high-quality, representative data foundation.

[0047] For example, in the antenna array production line, the antenna array model is HR-X03K13-231. Using the product model HR-X03K13-231 and the production line number A4-B3-76C as the first and second search criteria, the records of the most recent 100 production times are retrieved. Each record contains parameters of multiple production nodes. Among them, the spindle speed of the CNC milling node of the antenna array is 8000-15000 rpm, and the peak value of the furnace temperature curve of the welding node is 235-245°C.

[0048] In this embodiment, the obtained process parameter data facilitates parameter analysis and provides a data foundation for subsequent analysis. Compared to traditional technologies, this application ensures data reliability and avoids data interference from different product types or production lines, providing a high-quality data foundation for subsequent parameter analysis.

[0049] S30: Based on the process parameter data of each production node in the multiple production records of the same product, analyze the fluctuation range of the process parameters of each production node, and determine the multi-parameter fluctuation constraint range of each production node.

[0050] In this embodiment, since the process parameter data of each production node fluctuates, the box plot analysis method is used to statistically model the process parameters of each production node. The reasonable fluctuation range of the parameters is determined by the quartiles and interquartile ranges to prevent the parameters from exceeding the safe range and reduce parameter variability and product defects.

[0051] Step S30 in the method provided in this application embodiment includes:

[0052] Traverse multiple production nodes and determine the first production node;

[0053] Based on the first production node, process parameter data is extracted from multiple production records of the same product to obtain the first process parameter dataset.

[0054] Box plot analysis is performed on each process parameter in the first process parameter dataset to determine the fluctuation constraint range of each process parameter and construct the first multi-parameter fluctuation constraint range.

[0055] Following the method of constructing the first multi-parameter fluctuation constraint interval of the first production node, the multi-parameter fluctuation constraint intervals of the remaining production nodes are constructed to obtain the multi-parameter fluctuation constraint intervals of each production node.

[0056] Specifically, production nodes are run sequentially or simultaneously, multiple production nodes are traversed, and the starting node in the production process, i.e., the first production node, is determined.

[0057] For example, the antenna array production nodes include CNC milling, welding, electroplating, etc. After traversing the production nodes, the CNC milling node is determined as the first production node.

[0058] Furthermore, based on the first production node, all relevant parameter values, i.e., process parameter data, are extracted from multiple production records of the same product to obtain a first process parameter dataset. The first production node may contain multiple process parameter datasets. The first parameter dataset extracts multiple process parameter data from multiple records of the same product at the first production node; these process parameter data may be temperature parameter data or time parameter data.

[0059] For example, for CNC milling nodes, spindle speed parameter data is extracted from historical records: [12000, 11000, 12500, 11800, 13000] rpm, forming a speed parameter dataset; for welding nodes, furnace temperature peak parameter data is extracted from historical records: [238, 240, 239, 242, 241] °C.

[0060] Furthermore, box plot analysis is performed on each process parameter in the first process parameter dataset to determine the fluctuation constraint interval for each process parameter, thus constructing the first multi-parameter fluctuation constraint interval. The fluctuation constraint interval refers to the reasonable range of variation for each process parameter at each production node. This interval defines the parameter values ​​and ensures production stability. Historical production records of the same product are analyzed using box plot analysis, a statistical method. Based on quartile resistance to outliers, the upper quartile (Q3), lower quartile (Q1), and interquartile range (IQR) are calculated for each parameter, and the constraint interval [Q1-k×IQR, Q3+k×IQR] is defined, where k is a constant, typically 1.5. Domain knowledge can be incorporated to adjust the interval, preventing excessive errors that could produce unreliable values, thus obtaining the first multi-parameter fluctuation constraint interval.

[0061] Similarly, following the method of constructing the first multi-parameter fluctuation constraint interval of the first production node, the multi-parameter fluctuation constraint intervals of the remaining production nodes are constructed according to the order of one or more production nodes, thus obtaining the multi-parameter fluctuation constraint intervals of each production node.

[0062] For example, the first multi-parameter fluctuation constraint ranges for spindle speed and furnace temperature peak are [10000, 14000] rpm and [235, 245] °C, respectively.

[0063] In this embodiment of the application, step S30, which involves constructing the first multi-parameter fluctuation constraint interval, includes:

[0064] Based on the first process parameter dataset, determine multiple process parameter types for the first production node;

[0065] A first process parameter type is determined from the plurality of process parameter types, and a first parameter dataset is extracted from the first process parameter dataset according to the first process parameter type;

[0066] Calculate the upper quartile, lower quartile, and interquartile range of the first parameter dataset, and construct a fluctuation constraint interval for the first process parameter type based on the upper quartile, lower quartile, and interquartile range;

[0067] Following the method of constructing the first process parameter type fluctuation constraint interval, fluctuation constraint intervals for other process parameter types are constructed, and the fluctuation constraint intervals for multiple process parameter types are combined to form the first multi-parameter fluctuation constraint interval.

[0068] Specifically, the types of multiple process parameters for the first production node are determined based on the first process parameter dataset.

[0069] The parameter type corresponds to an adjustable variable within the production node, such as temperature or time. The parameter type of all nodes in the production line can be identified through historical production records of the same product.

[0070] For example, by using historical production records of the same product, a total of 15 parameter types can be identified for all nodes in the production line.

[0071] Furthermore, in the production records of the same product, all process parameter types of the first production node are grouped and filtered to determine the first process parameter type from multiple process parameter types of the production line, and the first parameter dataset is extracted from the first process parameter dataset based on the first process parameter type.

[0072] For example, the first process parameter type includes spindle speed, feed rate, and depth of cut. The first parameter dataset [12000, 11000, 12500, 11800, 13000] rpm is extracted from the first parameter dataset.

[0073] Furthermore, the upper quartile, lower quartile, and interquartile range of the first parameter dataset are calculated, and the fluctuation constraint interval of the first process parameter type is constructed based on the upper quartile, lower quartile, and interquartile range.

[0074] For example, box plot analysis shows that the spindle speeds are Q1=11250rpm, Q3=12750rpm, and IQR=1500rpm. 11250-1.5×1500=9000, 12750+1.5×1500=15000, so the constraint range is [9000, 15000]rpm.

[0075] Furthermore, following the method for constructing the first process parameter type fluctuation constraint interval, the construction of the first process parameter type fluctuation constraint interval is repeated until fluctuation constraint intervals for all process parameter types are completed. Then, the fluctuation constraint intervals of multiple process parameter types are combined to finally form the first multi-parameter fluctuation constraint interval. The combination of fluctuation constraint intervals involves merging intervals of all parameter types into the first multi-parameter fluctuation constraint interval, representing the overall constraint of the first process node. However, because there are correlations between process parameter types, parameter interactions need to be considered when combining fluctuation constraint intervals. If temperature and pressure both affect quality, and there is a correlation between them, this can be determined through multivariate statistics to ensure the correlation between them. This improves production predictability, provides a safety boundary for optimization, and enables multi-parameter fusion calculation.

[0076] For example, the first parameter fluctuation constraint ranges for spindle speed and feed rate are [9000, 15000] rpm and [800, 1200] mm / min, respectively. After parameter interaction adjustment, the multi-parameter constraint range for the CNC milling node is {spindle speed: [10000, 14000] rpm, feed rate: [900, 1100] mm / min}. In this embodiment, based on the process parameter data of each production node in multiple production records of the same product, a box plot analysis method is used to statistically model the process parameters of each production node, eliminating outliers in historical data and ensuring that the constraint range covers the parameter variation range of normal production, providing reliable constraint conditions for the optimization algorithm. This improves the predictability of production, provides a safety boundary for optimization, realizes multi-parameter fusion calculation, and provides sufficient persuasiveness for subsequent multi-source data analysis.

[0077] S40: Within the multi-parameter fluctuation constraint range of each production node, with the product performance requirements as the optimization objective, the optimal production parameters of each production node are obtained, and the target product production line is configured to produce the target product. The product performance requirements are split and optimized using an optimization target splitter constructed with a BP neural network. The optimal production parameters are tested using a joint node twin model and configured when the product performance requirements are met.

[0078] In this embodiment, with the goal of optimizing product performance, production parameters are optimized within the multi-parameter fluctuation constraint range of each production node to obtain the optimal production parameters for each production node, and the target product line is configured according to the optimal production parameters to produce the target product.

[0079] Step S40 in the method provided in this application embodiment includes:

[0080] Retrieve the optimized target splitter bound to the production line number;

[0081] The product performance requirements are used as optimization targets and input into the optimization target splitter to obtain the optimization sub-targets for each production node.

[0082] Based on the optimization sub-objectives of each production node, the optimal production parameters of each production node are obtained within the multi-parameter fluctuation constraint range of each production node.

[0083] Specifically, this is achieved by retrieving the optimization target splitter associated with the production line number. Based on the production line number, multiple historical product performance requirements from the target product production line are collected to construct a sample optimization target set. This sample optimization target set is then split into optimization target sub-sets, and a backpropagation neural network is used to construct the optimization target splitter.

[0084] Furthermore, using the sample optimization target set as input features and the sample optimization sub-target set as supervision labels, an optimization target splitter is trained and generated.

[0085] Furthermore, the product performance requirements are taken as the optimization target and input into the optimization target splitter to obtain the optimization sub-targets for each production node.

[0086] Furthermore, based on the optimization sub-objectives of each production node, the optimal production parameters of each production node are obtained within the multi-parameter fluctuation constraint range of each production node.

[0087] In this embodiment of the application, the construction step of optimizing the target splitter in step S40 includes:

[0088] Based on the production line number, collect multiple historical product performance requirements of the target product production line, and construct a sample optimization target set based on the multiple historical product performance requirements;

[0089] The optimization objectives of each sample in the sample optimization objective set are split into optimization objectives according to multiple production nodes of the target product production line to construct a sample optimization sub-objective set;

[0090] The optimization target splitter is trained and generated using the sample optimization target set as input features and the sample optimization sub-target set as supervision labels.

[0091] Specifically, based on the production line number, multiple historical product performance requirements of the target product production line are collected, and a sample optimization target set is constructed based on these historical product performance requirements. Performance requirements for the production line are collected from historical production records of the same product over the past three months, and the collected historical product performance requirements are cleaned and normalized to form training samples. The sample optimization target set should cover multiple scenarios to ensure the generalization ability of the optimization target splitter.

[0092] Furthermore, the optimization objectives of each sample in the sample optimization objective set are broken down into optimization objectives according to multiple production nodes of the target product production line, resulting in sample optimization sub-objectives. After integrating multiple production nodes, a sample optimization sub-objective set is obtained. By dividing the sample optimization sub-objective set, the optimal production parameters for each production node can be accurately output.

[0093] Furthermore, using the sample optimization target set as input features and the sample optimization sub-target set as supervision labels, a backpropagation (BP) neural network is used to train and generate an optimization target splitter. The BP neural network model is a feedforward neural network trained through error backpropagation and is commonly used to predict continuous values.

[0094] For example, the construction steps for optimizing the target splitter model are as follows:

[0095] Model structure: It mainly consists of an input layer, a hidden layer, and an output layer, containing 5 input nodes, 5 hidden layer nodes, and 1 output layer node. The input layer sample optimization target set is weighted and summed using weights and biases. The hidden layer uses an activation function to perform a nonlinear transformation on the sample optimization target set using the sample optimization sub-target set as the supervision label. The output layer again uses weights and biases to perform a weighted sum and then uses an activation function to obtain the output sample optimization sub-target set.

[0096] Model training: The sample optimization sub-target set is used as the supervision label. An initial learning rate and weights are set and weights are assigned. The mean squared error function (MSE) is used to calculate the error between the prediction result and the sample optimization sub-target set. Weight adjustments are made and calculations are repeated iteratively until the error is minimized. Images are generated through forward propagation and parameters are updated through backpropagation. Performance is evaluated using a validation set after each training epoch to avoid overfitting. The model is considered successful when the MSE loss decreases by less than 1% over five consecutive training epochs. e-5 When the MSE loss on the validation set stabilizes below 0.01, the model is considered to have converged, and the optimized target splitter is obtained.

[0097] In this embodiment of the application, step S40, which involves optimizing each production node according to its sub-objectives and within the multi-parameter fluctuation constraint range of each production node, to obtain the optimal production parameters for each production node, includes:

[0098] Determine the first production node from multiple production nodes, and construct a twin model for the first production node;

[0099] Randomly generate first random multi-parameter data within the multi-parameter fluctuation constraint range of the first production node;

[0100] The first production result is obtained by running the first random multi-parameter data through the first node twin model;

[0101] Determine whether the first production result satisfies the optimization sub-objective of the first production node;

[0102] When the first production result satisfies the optimization sub-objective of the first production node, the first random multi-parameter data is used as the optimal production parameter of the first production node.

[0103] When the first production result does not meet the optimization sub-objective of the first production node, a second random multi-parameter data is randomly generated in the multi-parameter fluctuation constraint range of the first production node, and the second random multi-parameter data is run through the twin model of the first node to obtain the second production result, and the second production result is judged according to the optimization sub-objective of the first production node.

[0104] Iterate until the optimal production parameters of the first production node are obtained;

[0105] Following the method of obtaining the optimal production parameters of the first production node, the optimal production parameters of the remaining production nodes are obtained synchronously, thus obtaining the optimal production parameters of each production node.

[0106] Specifically, a first production node is determined from multiple production nodes, and a first-node twin model is constructed for it. Similar to constructing an optimization target splitter, a backpropagation neural network is used to build the first-node twin model, where the first-node parameter dataset is used as sample data, and the sample optimization sub-target set is used as supervision labels. The corresponding optimal production parameters are obtained using the output of the first-node twin model.

[0107] For example, the construction steps of the first node twin model are as follows:

[0108] Model structure: It mainly consists of an input layer, a hidden layer, and an output layer, containing 5 input nodes, 5 hidden layer nodes, and 1 output layer node. The parameter dataset of the first node of the input layer is weighted and summed using weights and biases. The hidden layer uses an activation function to perform a nonlinear transformation on the sample optimization sub-objective set as the supervision label. The output layer again uses weights and biases to perform a weighted sum and then uses an activation function to obtain the output sample optimization sub-objective set.

[0109] Model training: The sample optimization sub-target set is used as the supervision label. An initial learning rate and weights are set and weights are assigned. The mean squared error function (MSE) is used to calculate the error between the prediction result and the sample optimization sub-target set. Weight adjustments are made and calculations are repeated iteratively until the error is minimized. Images are generated through forward propagation and parameters are updated through backpropagation. Performance is evaluated using a validation set after each training epoch to avoid overfitting. The model is considered successful when the MSE loss decreases by less than 1% over five consecutive training epochs. e-5 When the MSE loss on the validation set stabilizes below 0.01, the model is considered converged, and the first node twin model is obtained.

[0110] Furthermore, using a random data generator, first random multi-parameter data is randomly generated at the first production node, and the randomly generated first random multi-parameter data can be used as input to the twin model of the first node.

[0111] For example, in an antenna array production line, a random data generator is used to randomly generate the first random multi-parameter data 12500rpm of the CNC milling node, which can be used as the input of the first node twin model.

[0112] Furthermore, by running the first random multi-parameter data through the first node twin model, a first production result is obtained. The randomly generated first random multi-parameter data is run in the first node twin model, and the first production result is output. Based on the output first production result, it can be determined whether the production result satisfies the optimization sub-objective.

[0113] Furthermore, by comparing the output of the first production result with the optimization sub-objective, it can be determined whether the first production result satisfies the optimization sub-objective of the first production node. Here, the optimization sub-objective is an interval range; production results falling within this interval range are considered to satisfy the optimization sub-objective.

[0114] For example, in the antenna array production line, the target for CNC milling nodes is a surface roughness Ra≤1.6μm, and the target for welding nodes is a furnace temperature peak ≥235°C.

[0115] Furthermore, when the first production result satisfies the optimization sub-objective of the first production node, this first random multi-parameter data is used as the optimal production parameter of the first production node; when the first production result does not satisfy the optimization sub-objective of the first production node, random data is generated again within the multi-parameter fluctuation constraint range of the first production node to obtain the second random multi-parameter data, and the second random multi-parameter data is run through the twin model of the first node to obtain the second production result, and the second production result is judged according to the optimization sub-objective of the first production node.

[0116] Furthermore, the process is iterated until the optimal production parameters for the first production node are obtained. This iteration is repeated multiple times until the optimal production parameters are obtained.

[0117] Furthermore, following the method of obtaining the optimal production parameters of the first production node, synchronous optimization is performed, and the optimal production parameters of all nodes are obtained simultaneously to obtain the optimal production parameters of the remaining production nodes, thus obtaining the optimal production parameters of each production node.

[0118] In this embodiment of the application, before configuring the target product production line to produce the target product in step S40, the method further includes:

[0119] Retrieve the node twin models of each production node and combine them according to the production sequence to obtain a combined node twin model.

[0120] By running the optimal production parameters of each production node through the joint node twin model, the joint production results are obtained;

[0121] When the joint production results meet the product performance requirements, the target product production line is configured according to the optimal production parameters of each production node.

[0122] Specifically, the node twin models of each production node are retrieved, and these node twin models are combined according to the production sequence to obtain a joint node twin model. After retrieving the node twin models of each production node, they are combined according to the production sequence to obtain a concatenated joint node model.

[0123] For example, a joint model of the antenna array production line is created. This results in a joint node twin model of the CNC milling model, welding model, and electroplating model.

[0124] Furthermore, the optimal production parameters of each production node are run through the joint node twin model to obtain the joint production results. Using the joint node twin model, the optimal parameters of all nodes are input, and the joint node twin model is run. The obtained joint production results may include multiple indicators, such as the axial ratio of the circularly polarized antenna and the in-band standing wave ratio.

[0125] For example, in an antenna array production line, with the optimal input node parameters of a spindle speed of 12,500 rpm and a welding furnace temperature of 240°C, the joint production results are: in-band standing wave ratio <1.5 and circularly polarized antenna axial ratio <3.0 dB.

[0126] Furthermore, the quality of the produced products is assessed based on the results of joint production and product performance. When the joint production results meet the product performance requirements, the target product production line is configured according to the optimal production parameters of each production node; otherwise, the parameters need to be adjusted or re-optimized.

[0127] For example, if the input node's optimal parameters of spindle speed 12500 rpm and welding furnace temperature 240°C meet the requirements, then the optimal production parameters will be configured for the target product production line.

[0128] In this embodiment, with the goal of optimizing product performance, production parameters are optimized within the multi-parameter fluctuation constraint range of each production node to obtain the optimal production parameters for each node. The target product line is then configured according to these optimal parameters for target product production. Simultaneously, the optimization target set is split using an optimization target splitter model to obtain accurate optimal production parameters. Furthermore, a joint node twin model is constructed by building node twin models and retrieving the node twin models of each production node. The optimal production parameters are then verified through the joint production results. Compared to existing technologies, this approach ensures both the manufacturability of parameter configuration and the achievement of product performance, avoiding the blindness of traditional trial-and-error methods. Systematic optimization calculations achieve scientific and precise parameter configuration.

[0129] The embodiments of this application, through the specific implementation methods described above, achieve the following technical effects:

[0130] In this embodiment, a first search condition and a second search condition are first constructed using the product model and production line number, respectively. Then, production records are matched against the production record database using the first and second search conditions to obtain multiple production records of the same product, which are then aggregated to form a set of production records for the same product. This establishes the data foundation for the entire management process, ensuring that all subsequent data retrieval, parameter analysis, and optimization calculations are based on accurate product information. Secondly, the obtained process parameter data facilitates parameter analysis, providing a data foundation for subsequent analysis. Compared to traditional technologies, this application guarantees data reliability, avoids data interference from different product types or different production lines, and provides a high-quality data foundation for subsequent parameter analysis.

[0131] Secondly, based on the process parameter data of each production node in multiple production records of the same product, box plot analysis was used to statistically model the process parameters of each production node. This eliminated outliers in historical data and ensured that the constraint interval covered the parameter variation range of normal production, providing reliable constraints for the optimization algorithm. This improved the predictability of production, provided a safety boundary for optimization, and enabled multi-parameter fusion calculation, providing sufficient persuasiveness for subsequent multi-source data analysis. Finally, with the goal of optimizing product performance, production parameters were optimized within the multi-parameter fluctuation constraint interval of each production node to obtain the optimal production parameters for each production node. The target product line was then configured according to the optimal production parameters, and the target product was produced. Simultaneously, the optimization target set was split by an optimization target splitter model to obtain accurate optimal production parameters. Furthermore, by constructing node twin models and retrieving the node twin models of each production node, a joint node twin model was built, and the optimal production parameters were verified through joint production results.

[0132] Compared to existing technologies, this approach ensures both the manufacturability of parameter configuration and the achievement of product performance. It addresses the problems of outdated product manufacturing process management methods in existing technologies, which lead to equipment risks, production quality defects, low production efficiency, and high scrap rates. It improves production predictability, provides safety boundaries for optimization, and enables multi-parameter fusion calculations. It avoids the blindness of traditional methods, achieving scientific and precise parameter configuration and enhancing self-learning and adaptive capabilities.

[0133] Example 2, as Figure 2 As shown, this application provides a product manufacturing process management system based on multi-source data, including:

[0134] Product model acquisition module 11 is used to receive the product model of the target product, determine the target product production line and production line number of the product model, and obtain the product performance requirements of the target product based on the product model.

[0135] Product production record retrieval module 12 is used to retrieve a set of production records for the same product based on the product model and the production line number. The set of production records for the same product includes multiple production records for the same product, and each production record for the same product includes process parameter data of multiple production nodes on the target product production line.

[0136] The process parameter fluctuation range analysis module 13 is used to analyze the process parameter fluctuation range of each production node based on the process parameter data of each production node in the multiple production records of the same product, and to determine the multi-parameter fluctuation constraint range of each production node.

[0137] The optimal production parameter acquisition module 14 is used to obtain the optimal production parameters of each production node within the multi-parameter fluctuation constraint range of each production node, with the product performance requirements as the optimization objective, and configure the target product production line to produce the target product. The optimization objective splitter is used to split the product performance requirements for optimization. The optimization objective splitter is constructed using a BP neural network. The optimal production parameters are tested using a joint node twin model and configured when the product performance requirements are met.

[0138] In one embodiment, the product model acquisition module 11 is used for:

[0139] A first search condition is constructed based on the product model, and a second search condition is constructed based on the production line number;

[0140] Connect to the production record database, and match production records in the production record database according to the first search condition and the second search condition to obtain multiple production records of the same product;

[0141] The production records of the same product are summarized to obtain the production record set of the same product.

[0142] The first and second search criteria are matched against production records in the production record database to obtain multiple production records for the same product, including:

[0143] Extract the first production record from the production record database, and retrieve the product model and production line number of the first production record to obtain the first product model and the first production line number;

[0144] When the first product model meets the first search condition and the first production line number meets the second search condition, the first production record is added to the plurality of production records of the same product.

[0145] In one embodiment, the process parameter fluctuation range analysis module 13 is used for:

[0146] Traverse multiple production nodes and determine the first production node;

[0147] Based on the first production node, process parameter data is extracted from multiple production records of the same product to obtain the first process parameter dataset.

[0148] Box plot analysis is performed on each process parameter in the first process parameter dataset to determine the fluctuation constraint range of each process parameter and construct the first multi-parameter fluctuation constraint range.

[0149] Following the method of constructing the first multi-parameter fluctuation constraint interval of the first production node, the multi-parameter fluctuation constraint intervals of the remaining production nodes are constructed to obtain the multi-parameter fluctuation constraint intervals of each production node.

[0150] The step of performing box plot analysis on each process parameter in the first process parameter dataset to determine the fluctuation constraint range of each process parameter and constructing the first multi-parameter fluctuation constraint range includes:

[0151] Based on the first process parameter dataset, determine multiple process parameter types for the first production node;

[0152] A first process parameter type is determined from the plurality of process parameter types, and a first parameter dataset is extracted from the first process parameter dataset according to the first process parameter type;

[0153] Calculate the upper quartile, lower quartile, and interquartile range of the first parameter dataset, and construct a fluctuation constraint interval for the first process parameter type based on the upper quartile, lower quartile, and interquartile range;

[0154] Following the method of constructing the first process parameter type fluctuation constraint interval, fluctuation constraint intervals for other process parameter types are constructed, and the fluctuation constraint intervals for multiple process parameter types are combined to form the first multi-parameter fluctuation constraint interval.

[0155] In one embodiment, the optimal production parameter acquisition module 14 is used for:

[0156] Retrieve the optimized target splitter bound to the production line number;

[0157] The product performance requirements are used as optimization targets and input into the optimization target splitter to obtain the optimization sub-targets for each production node.

[0158] Based on the optimization sub-objectives of each production node, the optimal production parameters of each production node are obtained within the multi-parameter fluctuation constraint range of each production node.

[0159] The construction steps of the optimized target splitter include:

[0160] Based on the production line number, collect multiple historical product performance requirements of the target product production line, and construct a sample optimization target set based on the multiple historical product performance requirements;

[0161] The optimization objectives of each sample in the sample optimization objective set are split into optimization objectives according to multiple production nodes of the target product production line to construct a sample optimization sub-objective set;

[0162] The optimization target splitter is trained and generated using the sample optimization target set as input features and the sample optimization sub-target set as supervision labels.

[0163] In one embodiment, the optimal production parameter acquisition module 14 is further configured to:

[0164] Determine the first production node from multiple production nodes, and construct a twin model for the first production node;

[0165] Randomly generate first random multi-parameter data within the multi-parameter fluctuation constraint range of the first production node;

[0166] The first production result is obtained by running the first random multi-parameter data through the first node twin model;

[0167] Determine whether the first production result satisfies the optimization sub-objective of the first production node;

[0168] When the first production result satisfies the optimization sub-objective of the first production node, the first random multi-parameter data is used as the optimal production parameter of the first production node.

[0169] When the first production result does not meet the optimization sub-objective of the first production node, a second random multi-parameter data is randomly generated in the multi-parameter fluctuation constraint range of the first production node, and the second random multi-parameter data is run through the twin model of the first node to obtain the second production result, and the second production result is judged according to the optimization sub-objective of the first production node.

[0170] Iterate until the optimal production parameters of the first production node are obtained;

[0171] Following the method of obtaining the optimal production parameters of the first production node, the optimal production parameters of the remaining production nodes are obtained synchronously, thus obtaining the optimal production parameters of each production node.

[0172] In one embodiment, a parameter validation model is also included, which is used for:

[0173] Retrieve the node twin models of each production node and combine them according to the production sequence to obtain a combined node twin model.

[0174] By running the optimal production parameters of each production node through the joint node twin model, the joint production results are obtained;

[0175] When the joint production results meet the product performance requirements, the target product production line is configured according to the optimal production parameters of each production node.

[0176] The embodiments of this application, through the specific implementation methods described above, achieve the following technical effects:

[0177] In this embodiment, firstly, the product model acquisition module 11 determines the target product production line and production line number, and obtains the product performance requirements of the target product, providing a high-quality, highly relevant multi-source data foundation for subsequent analysis. Secondly, the product production record retrieval module 12 retrieves the same product production record set, which includes process parameter data for multiple production nodes. Thirdly, the process parameter fluctuation range analysis module 13 analyzes the process parameter fluctuation range of each production node using the process parameter data, performs box plot analysis to determine the multi-parameter fluctuation constraint range of each production node, and obtains a safe and feasible operating range. Finally, the optimal production parameter acquisition module 14 splits the optimization target set using an optimization target splitter model to obtain accurate optimal production parameters. A node twin model is constructed, and the node twin models of each production node are retrieved to construct a joint node twin model. The optimal production parameters are verified through the joint production results. By utilizing the multi-parameter fluctuation constraint range of each production node, the optimal production parameters of each production node are obtained, and the target product production line is configured to produce the target product.

[0178] Compared to existing technologies, this application constructs a product production process management system based on multi-source data by receiving product models and determining production lines and performance requirements, retrieving production record sets of the same product, analyzing the fluctuation range of process parameters to determine constraint intervals, and finding the optimal production parameters and configuring the production line within the constraint intervals. Through a data-driven approach, it solves the problems of outdated product production process management methods in existing technologies, which lead to equipment risks and production quality defects, resulting in low production efficiency and high scrap rates. It achieves refined modeling of the entire production process, quantitative inheritance of historical experience, and global collaborative optimization, constructing a closed loop for production control and significantly reducing trial-and-error costs and production risks.

[0179] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0180] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A product manufacturing process management method based on multi-source data, characterized in that, The method includes: Receive the product model of the target product, determine the target product production line and production line number of the product model, and obtain the product performance requirements of the target product based on the product model; Based on the product model and the production line number, retrieve the same product production record set. The same product production record set includes multiple same product production records. Each same product production record includes process parameter data of multiple production nodes on the target product production line. Based on the process parameter data of each production node in the multiple production records of the same product, analyze the fluctuation range of process parameters of each production node, and determine the multi-parameter fluctuation constraint range of each production node. Within the multi-parameter fluctuation constraint range of each production node, with the product performance requirements as the optimization objective, the optimal production parameters of each production node are obtained, and the target product production line is configured to produce the target product. The product performance requirements are split into the optimization target splitter and optimized. The optimization target splitter is constructed using a BP neural network, and the optimal production parameters are tested using a joint node twin model. The configuration is performed when the product performance requirements are met. Within the multi-parameter fluctuation constraint range of each production node, the optimal production parameters for each production node are obtained by taking the product performance requirements as the optimization objective, including: Retrieve the optimized target splitter bound to the production line number; The product performance requirements are used as optimization targets and input into the optimization target splitter to obtain the optimization sub-targets for each production node. Based on the optimization sub-objectives of each production node, the optimal production parameters of each production node are obtained within the multi-parameter fluctuation constraint range of each production node. The construction steps of the optimized target splitter include: Based on the production line number, collect multiple historical product performance requirements of the target product production line, and construct a sample optimization target set based on the multiple historical product performance requirements; The optimization objectives of each sample in the sample optimization objective set are split into optimization objectives according to multiple production nodes of the target product production line to construct a sample optimization sub-objective set; Using the sample optimization target set as input features and the sample optimization sub-target set as supervision labels, the optimization target splitter is trained and generated. Specifically, based on the optimization sub-objectives of each production node, the optimal production parameters for each production node are obtained within the multi-parameter fluctuation constraint range, including: Determine the first production node from multiple production nodes, and construct a twin model for the first production node; Randomly generate first random multi-parameter data within the multi-parameter fluctuation constraint range of the first production node; The first production result is obtained by running the first random multi-parameter data through the first node twin model; Determine whether the first production result satisfies the optimization sub-objective of the first production node; When the first production result satisfies the optimization sub-objective of the first production node, the first random multi-parameter data is used as the optimal production parameter of the first production node. When the first production result does not meet the optimization sub-objective of the first production node, a second random multi-parameter data is randomly generated in the multi-parameter fluctuation constraint range of the first production node, and the second random multi-parameter data is run through the twin model of the first node to obtain the second production result, and the second production result is judged according to the optimization sub-objective of the first production node. Iterate until the optimal production parameters of the first production node are obtained; Following the method of obtaining the optimal production parameters of the first production node, the optimal production parameters of the remaining production nodes are obtained synchronously, thus obtaining the optimal production parameters of each production node.

2. The method according to claim 1, characterized in that, Based on the product model and the production line number, a set of production records for the same product is retrieved, including: A first search condition is constructed based on the product model, and a second search condition is constructed based on the production line number; Connect to the production record database, and match production records in the production record database according to the first search condition and the second search condition to obtain multiple production records of the same product; The production records of the same product are summarized to obtain the production record set of the same product.

3. The method according to claim 2, characterized in that, Based on the first and second search criteria, production records are matched in the production record database to obtain multiple production records of the same product, including: Extract the first production record from the production record database, and retrieve the product model and production line number of the first production record to obtain the first product model and the first production line number; When the first product model meets the first search condition and the first production line number meets the second search condition, the first production record is added to the plurality of production records of the same product.

4. The method according to claim 1, characterized in that, Based on the process parameter data of each production node in the multiple production records of the same product, the fluctuation range of process parameters at each production node is analyzed, and the multi-parameter fluctuation constraint interval of each production node is determined, including: Traverse multiple production nodes and determine the first production node; Based on the first production node, process parameter data is extracted from multiple production records of the same product to obtain the first process parameter dataset. Box plot analysis is performed on each process parameter in the first process parameter dataset to determine the fluctuation constraint range of each process parameter and construct the first multi-parameter fluctuation constraint range. Following the method of constructing the first multi-parameter fluctuation constraint interval of the first production node, the multi-parameter fluctuation constraint intervals of the remaining production nodes are constructed to obtain the multi-parameter fluctuation constraint intervals of each production node.

5. The method according to claim 4, characterized in that, Box plot analysis is performed on each process parameter in the first process parameter dataset to determine the fluctuation constraint interval of each process parameter, thus constructing the first multi-parameter fluctuation constraint interval, including: Based on the first process parameter dataset, determine multiple process parameter types for the first production node; A first process parameter type is determined from the plurality of process parameter types, and a first parameter dataset is extracted from the first process parameter dataset according to the first process parameter type; Calculate the upper quartile, lower quartile, and interquartile range of the first parameter dataset, and construct a fluctuation constraint interval for the first process parameter type based on the upper quartile, lower quartile, and interquartile range; Following the method of constructing the first process parameter type fluctuation constraint interval, fluctuation constraint intervals for other process parameter types are constructed, and the fluctuation constraint intervals for multiple process parameter types are combined to form the first multi-parameter fluctuation constraint interval.

6. The method according to claim 1, characterized in that, Before configuring the target product production line to produce the target product, the method further includes: Retrieve the node twin models of each production node and combine them according to the production sequence to obtain a combined node twin model. By running the optimal production parameters of each production node through the joint node twin model, the joint production results are obtained; When the joint production results meet the product performance requirements, the target product production line is configured according to the optimal production parameters of each production node.

7. A product manufacturing process management system based on multi-source data, characterized in that: The system for implementing the method according to any one of claims 1-6 comprises: The product model acquisition module is used to receive the product model of the target product, determine the target product production line and production line number of the product model, and obtain the product performance requirements of the target product based on the product model. The product production record retrieval module is used to retrieve the same product production record set according to the product model and the production line number. The same product production record set includes multiple same product production records, and each same product production record includes process parameter data of multiple production nodes on the target product production line. The process parameter fluctuation range analysis module is used to analyze the process parameter fluctuation range of each production node based on the process parameter data of each production node in the multiple production records of the same product, and to determine the multi-parameter fluctuation constraint range of each production node. The optimal production parameter acquisition module is used to find the optimal production parameters for each production node within the multi-parameter fluctuation constraint range of each production node, with the product performance requirements as the optimization objective, and configure the target product production line to produce the target product. The optimization objective splitter is used to split the product performance requirements for optimization. The optimization objective splitter is constructed using a BP neural network. The optimal production parameters are tested using a joint node twin model and configured when the product performance requirements are met.

Images