System and method for generating process route

Abstract

This invention provides a process route generation system and a process route generation method. The process route generation system includes a processor and a storage device. The processor executes multiple modules. A regression analysis module is used to execute a regression algorithm based on demand data to obtain process parameters and material information. A correlation analysis module is used to perform correlation analysis based on process parameters and material information to obtain first process information and second process information, respectively. The correlation analysis module includes a process-to-process correlation module. The process-to-process correlation module is used to obtain a related process route based on one of the first process information and the second process information. A deep learning module is used to generate candidate process routes based on demand data. The processor compares the related process routes and candidate process routes to generate a recommended process route.

Description

Process route generation system and process route generation method The present invention relates to a process route recommendation and management technology, and in particular to a process route generation system and a process route generation method. In the manufacturing industry, the selection and design of accurate and efficient process routes is crucial for product quality, production efficiency, and cost control. Current methods for generating production process routes and product manufacturing information often rely on extensive practical experience and repeated trials to generate a process route that meets requirements. This results in experienced personnel having to adjust or redesign the process route for each different finished product, leading to inefficiencies and high costs in process route generation. Furthermore, with increasing market demand and the diversification of product cycles, manufacturers are demanding faster and more accurate methods for generating process routes. The present invention is directed to a process route generation system and a process route generation method, which can provide a suitable process route according to user demand data, and provide a recommended process route based on historical data and a mining algorithm. According to an embodiment of the present invention, the process route generation system of the present invention includes a processor and a storage device. The processor is coupled to the storage device and is used to execute multiple modules in the storage device. The multiple modules include a regression analysis module, a correlation analysis module and a deep learning module. The regression analysis module is used to execute a regression algorithm based on the demand data to obtain process parameters and material information. The correlation analysis module is used to execute correlation analysis based on the process parameters to obtain first process information. In addition, the correlation analysis module obtains second process information based on the material information. The correlation analysis module includes a process and process correlation module. The process and process correlation module is used to obtain an associated process route based on one of the first process information and the second process information. The deep learning module is used to generate candidate process routes based on the demand data. The processor compares the associated process route and the candidate process route to generate a recommended process route. According to an embodiment of the present invention, a process route generation method is applicable to a process route generation system. The process route generation method includes the following steps: executing a regression algorithm based on demand data using a regression analysis module to obtain process parameters and material information; executing a correlation analysis based on the process parameters using a correlation analysis module to obtain first process information, and obtaining second process information based on the material information using the correlation analysis module; obtaining a correlated process route based on one of the first process information and the second process information using a process and process correlation module; generating candidate process routes based on the demand data using a deep learning module; and comparing the correlated process route with the candidate process routes using a processor to generate a recommended process route. Based on the above, the process route generation system and process route generation method of the present invention can effectively and automatically generate a process route corresponding to demand data and historical data as well as process parameters and a bill of materials corresponding to the process route. In order to make the above features and advantages of the present invention more clearly understood, embodiments are given below and described in detail with reference to the accompanying drawings. Reference will now be made in detail to exemplary embodiments of the present invention, examples of which are illustrated in the accompanying drawings. Whenever possible, the same reference numerals are used in the drawings and the description to refer to the same or like parts. FIG1 is a schematic diagram of a process route generation system according to one embodiment of the present invention. Referring to FIG1 , the process route generation system 100 includes a processor 110 and a storage device 120. The processor 110 is coupled to the storage device 120. The storage device 120 stores a regression analysis module 121, a correlation analysis module 122, a deep learning module 123, and a procurement calculation module 124. In this embodiment, the process route generation system 100 may be, for example, located on a cloud server, allowing users to connect and execute business service functions related to various application programming interfaces (APIs) also located on the cloud server. The cloud server may be, for example, a Software as a Service (SaaS) server, and the APIs may correspond to SaaS applications, but the present invention is not limited thereto. In one embodiment, the assembly sequence-based material calculation system 100 is located on the cloud server, and users can connect to log into their respective system accounts to receive data and instructions such as input data, demand data, selection instructions, and input operation instructions. In this embodiment, the processor 110 may include, for example, a central processing unit (CPU), or other programmable general-purpose or special-purpose microprocessor, digital signal processor (DSP), application-specific integrated circuit (ASIC), programmable logic device (PLD), other similar processing circuits, or a combination of these devices. The storage device 120 may include memory and / or a database, wherein the memory may be, for example, non-volatile memory (NVM). The storage device 120 may store relevant programs, modules, systems, or algorithms for implementing various embodiments of the present invention, for access and execution by the processor 110 to implement the relevant functions and operations described in various embodiments of the present invention. In this embodiment, the regression analysis module 121, the association analysis module 122, the deep learning module 123, and the procurement calculation module 124 can be implemented in a programming language such as JSON (JavaScript Object Notation), Extensible Markup Language (XML), or YAML, but the present invention is not limited thereto. In this embodiment, a user can execute the process route generation system 100, for example, through a personal computer device, and input initial data, process data, process history data, demand data, and instructions into the process route generation system to execute the corresponding data content. The assembly sequence-based material calculation system 100 can automatically execute the regression analysis module 121, the notification module 112, and the abnormality analysis module 113 based on the request data to automatically generate corresponding delivery plan data and analysis results (such as production gap analysis data). In this embodiment, a user can connect to the material calculation system 100 located on a cloud server and log in to the corresponding system account to input and set the detection path of the material calculation system. Therefore, the assembly sequence-based material calculation system 100 can automatically detect the production plan data 101 stored in a database or the user's electronic device. FIG2 is a flow chart of a process route generation method according to one embodiment of the present invention. Referring to FIG1 and FIG2 , in this embodiment, the process route generation system 100 may execute steps S210-S250 to generate a process route and / or process parameters and a bill of materials corresponding to the required data. In step S210, the processor 110, via the regression analysis module 121, executes a regression algorithm based on the required data to obtain process parameters and material information. The required data includes at least one of expected finished product performance, finished product specifications, and a manufacturing budget. For example, a user operates a terminal device to communicate with the process route generation system 100 and then sends a process route generation instruction to the processor 110. Based on the process route generation instruction, the processor 110 retrieves the required data corresponding to the process route instruction from a database, or retrieves the required data within the process route generation instruction. The processor 110 then executes the regression analysis module 121, causing the regression analysis module 121 to execute a regression algorithm based on the required data to obtain corresponding process parameters and material information. In one embodiment, the terminal device is a user's personal computer, smartphone, tablet, or other device. In one embodiment, regression analysis module 121 matches corresponding process parameters and material information from process history data stored in storage device 120 or an external database based on performance requirements in the demand data. The performance requirements are multiple product performance information, such as durability, speed, scalability, handling, and stability. For example, performance information such as insulation of 2 meters or withstand voltage of 0.3 MPa may be used, but the present invention is not limited to this. The process history information includes data and parameters related to the finished product, materials, and process operations, such as material inventory, material and process costs, supplier information, process information, and historical performance information. In step S220, the processor 110 performs correlation analysis based on the process parameters through the correlation analysis module 122 to obtain first process information, and the correlation analysis module 122 obtains second process information based on the material information (i.e., material information). The process information includes process parameters and material information. The process parameters include the type, sequence, and instruction arguments of each process operation in a plurality of process operations. For example, the process parameters include splitting and cutting (i.e., slitting) operations, core and core group testing operations, winding operations, gold spraying operations, welding and assembly operations, dipping operations and assembly operations, welding temperature values, test voltage volt values ​​and current ampere values, dipping environment temperature values, and the like. The present invention should not be limited to this. The material information includes the type of material and the amount of each material used. The material types include, for example, wires, gold, copper foil, insulating materials, impregnating agents, and the like. The present invention should not be limited to this. In one embodiment, the association analysis module 122 uses association rule learning based on historical process information to identify multiple association rules and stores them in the storage device 120 or a database. These association rules include process-to-process association rules and material-to-process association rules. Thus, the association analysis module 122 performs association analysis based on process parameters to obtain first process information. The association analysis module 122 then generates second process information based on the material information generated in step S210 and the association rules between materials and public welfare. In one embodiment, the correlation analysis module 122 includes a process and process correlation module. In step S230, the processor 110 executes the process and process correlation module to obtain a correlated process route based on one of the first process information and the second process information. For example, the process and process correlation module verifies the first process information and the second process information according to the verification rules stored in the storage device 120, and then obtains the correlated process route. The verification rules include matching of process information. When the matching results of the first process information and the second process information are the same, the process and process correlation module converts the first process information into a process route based on the process history data in the database or the storage device 120, and uses it as the correlated process route. When the matching results of the first process information and the second process information are different, the process and process correlation module outputs the differences to the user, and then converts the corresponding process information into a process route according to the user's selection instruction, and uses it as the correlated process route. In step S240, the processor 110 generates candidate process routes based on the demand data through the deep learning module 123. Specifically, the deep learning module 123 establishes a deep learning model based on the process history data, and then generates candidate process routes based on the expected performance information in the demand data. The candidate process routes include multiple process parameters, process operation sequences, and material information. In step S250, the processor 110 compares the associated process routes and the candidate process routes to generate a recommended process route. The recommended process route includes the process route (i.e., process operation sequence), process parameters, bill of materials, and supplier information. In one embodiment, the deep learning module 123 establishes a deep learning model based on the process history data through the long short-term memory algorithm. In this way, the deep learning module 123 can use the long short-term memory algorithm to process the process data with time series and predict the optimal continuous process steps (i.e., process route), and then generate candidate process routes based on the demand data. It is worth noting that the procurement calculation module 124 generates recommended supplier information based on the material inventory, cost data and recommended process routes, and generates recommended procurement information based on the material inventory and recommended supplier information. Specifically, the procurement calculation module 124 calculates the material information that should be purchased based on the material information in the recommended process route and the material inventory in the database or storage device 120, and calculates the most cost-effective recommended supplier information based on the material type and amount in the supplier information. In other words, the procurement calculation module 124 can compare the material inventory quantity based on the material type and quantity to be used in the recommended process route to obtain the type and data of missing materials. Then, the procurement calculation module 124 calculates the type and quantity of missing materials from the supplier information, and uses the supplier information with the lowest required cost as the recommended supplier information. For example, the recommended supplier information includes the type of material in short supply (e.g., copper foil, gold, wire, etc.), the quantity of each material in short supply, the recommended supplier information, and a suggested purchase list and purchase amount (i.e., suggested purchase information). Thus, the processor 110 of the process route generation system 100 outputs a recommended process route, process parameters corresponding to the recommended process route, suggested purchase information, and a list of materials required for the recommended process route. Figures 3A and 3B are data analysis flow charts of a process route generation method according to one embodiment of the present invention. Referring to Figures 1, 3A, and 3B, the process route generation system 100 may execute steps S310-S350, S321-S324, S331-S333, S341-S344, and S351-S353 to establish and train a deep learning model and a regression model based on the input data. In this embodiment, the storage device 120 also stores a data acquisition module and a data preprocessing module. The data acquisition module is coupled to at least one of a server, a database, and an electronic device to receive input data and demand data. The input data includes process history data 301, customer demand data, and material inventory cost data. The process history data 301 includes historical process routes, process parameters, required material information, and product performance test results. In this embodiment, the data acquisition module receives the process history data 301 and material inventory and cost data 302. In step S310, the data preprocessing module performs data preprocessing on the input data received by the data acquisition module (i.e., process history data 301 and material inventory and cost data 302). Data preprocessing includes at least one of data cleaning, missing value processing, and data standardization, thereby ensuring data quality and consistency, making the input data suitable for subsequent data analysis and model building steps. In step S320, the association analysis module 122 performs association rule analysis on the normalized data, where the association rule analysis can include steps S321 through S324. In steps S321 and S322, a frequent pattern growth algorithm (FP-GROWTH) is used to build a frequent pattern tree structure to identify frequent itemsets. Specifically, the association analysis module 122 uses the FP-GROWTH algorithm to build a frequent pattern tree structure based on the preprocessed input data to identify frequent itemsets. In step S323, the association analysis module 122 selects association rules based on the confidence level greater than the threshold as appropriate association rules. The association analysis module 122 then obtains association rules between multiple process parameters in the process route based on the frequent item sets. In this embodiment, the threshold is 50%. In other words, the association analysis module 122 selects rules with a confidence level greater than 50% as association rules, specifically the material-process association 303 and the process-process association 304. In this way, the association analysis module 122 can perform association analysis based on the established frequent pattern tree structure. In step S324, the association analysis module 122 stores the material-process association 303 and the process-process association 304 in the storage device 120 or a database. Specifically, the association analysis module 122 stores the material-process association 303 in the process-process association database 308 of the storage device 120, and stores the process-process association 304 in the material-process association database 309 of the storage device 120. In step S330, the deep learning module 123 trains a deep learning model based on the preprocessed input data. Step S330 includes the following steps: S331 to S333. In step S331, the deep learning module 123 divides the input data into training data and test data. In one embodiment, the deep learning module 123 divides the input data into training data, test data, and validation data in a ratio of 8:1:1. In step S332, the deep learning module 123 trains a long short-term memory (LSTM) model using the training set data, the test set data, and the validation set data based on a neural network. In other words, the LSTM model serves as a deep learning framework. In another embodiment, the deep learning module 123 divides the input data into training set data and test set data, and trains the LSTM model on the training set data and the test set data based on a deep neural network. In step S333, the deep learning module 123 calculates the accuracy of the trained LSTM model using mean squared error (MSE), thereby determining the effectiveness of the LSTM model training. In this way, the deep learning module 123 processes the time-series process data using the LSTM model and predicts the optimal process route 305 corresponding to the demand data (i.e., the candidate process route corresponding to the demand data). The deep learning module 123 stores the optimal process route 305 in the deep learning model database 3051. In one embodiment, the storage device 120 further stores a correlation analysis module. In step S340, before the processor 110 trains the regression model using input data, the processor 110 executes the correlation analysis module to perform correlation analysis. Step S340 includes the following steps S341 to S344. In step S341, the correlation analysis module performs correlation analysis on the variable parameters and the performance categories in the input data. In step S342, the correlation analysis module performs correlation analysis on the variable parameters and the performance categories in the input data. In step S343 and step S344, the correlation analysis module selects variables with a confidence correlation coefficient greater than 0.5, and the correlation analysis module organizes and classifies the influencing factors of each performance. In other words, the correlation analysis module uses variables with a confidence greater than a threshold (for example, 0.5) as influencing factors to obtain the influencing factors (i.e., influencing factors) of each performance category. In one embodiment, the correlation analysis module performs correlation analysis on different performance data (i.e., performance classifications) to obtain correlation analysis results, and the correlation analysis module can use pandas and numpy modules to perform specific correlation analysis. The correlation analysis results in this embodiment can be measured by a correlation coefficient. The correlation coefficient is, for example, the Pearson coefficient, the Spearman coefficient, or the Kendall coefficient. In one embodiment, since Pearson is applicable to continuous data, the correlation analysis module selects the Pearson correlation coefficient as the correlation coefficient, and selects the correlation coefficient standard as 0.5. In step S350, processor 110 executes regression analysis module 121 to build and train a regression model. In other words, processor 110 inputs the relationship between process parameters, material data, and product performance into regression analysis module 121 for modeling. Specifically, regression analysis module 121 is a regression model. Processor 110 trains the regression model based on process parameters, material data, and product performance using a deep neural network. In one embodiment, step S350 includes steps S351 to S353. In step S351, the processor 110 divides the performance categories and the influencing factors (ie, influencing factors) generated in step S344 into test set data and training set data in a ratio of 3:7. In step S352, the processor 110 executes the regression analysis module 121 to train a regression model based on the test set data and the training set data using a deep neural network. In this embodiment, the deep neural network is a neural network based on the Support Vector Regression (SVR) algorithm. In other words, the processor 110 trains the regression model based on the SVR algorithm. In another embodiment, the processor 110 uses a genetic algorithm to continuously optimize the parameters of the SVR algorithm. In step S353, the processor 110 determines the training effectiveness of the regression model using the R-squared value of the regression model. Specifically, the coefficient of determination (R-squared) of the regression model is used. The regression analysis module 121 determines the training effectiveness of the model based on the R-squared value and selects the model with the highest R-squared value as the baseline model. In this way, the trained regression model performs regression analysis on the input data using the baseline model to generate a relationship 306 between the process parameters, material type and amount, and performance. The processor 110 then stores the relationship 306 between the process parameters, material type and amount, and performance in the regression model database 307 of the storage device 120. Figures 4A and 4B are schematic diagrams of the architecture of a process route generation system according to an embodiment of the present invention. Referring to Figures 1, 4A, and 4B, the process route generation system 100 can execute the following steps S410 to S470: data input, model building, and generating a recommended process route based on customer requirements 402. In step S410, the data preprocessing module preprocesses the input data 401. The input data 401 includes process history data 301, material inventory, cost data, supplier information, and other data related to the process and material costs. In this embodiment, the data preprocessing module summarizes and organizes the preprocessed input data 401 into an application data table 411. The application data table 411 is a data set organized by collecting and consolidating the input data 401 to obtain multiple parameter factors that affect product performance in the process. Next, as shown in FIG3 and the process described above regarding establishing multiple models based on input data 401, in this embodiment, processor 110 stores the regression process parameters, material data, and performance result data 403 generated by the trained regression model in regression model database 307. Furthermore, processor 110 stores the material-process parameter association rules 404 and process-process association rules 405 generated by performing association analysis on the trained frequent pattern tree structure in material-process association database 309 and process-process association database 308, respectively. In step S420, processor 110 processes the time-series process data based on the trained long short-term memory model and stores the results in deep learning model database 3051. In step S430, the regression analysis module 121 inputs the customer requirements 402 into the regression model database 307. Using the trained regression model, the module generates process parameters 4062 and material information based on the expected performance in the customer requirements 402. Customer requirements 402 include data such as the expected performance of the finished product, the specifications for the finished product, and the finished product budget. Furthermore, the regression analysis module 121 stores the process parameters 4062 and material information in the material-process association database 309. In step S440, the association analysis module 122 generates corresponding process information (i.e., first process information) based on the process parameters 4062 generated in step S430 and the association relationships between multiple processes in the material-process parameter association rules 404. Furthermore, the association analysis module 122 generates corresponding process information (i.e., second process information) based on the material information generated in step S430 and the material-process parameter association rules 404. In step S450, the process-process association module of the association analysis module 122 verifies the first process information and the second process information to determine whether the first process information matches the second process information. If the first process information matches the second process information, the process-process association module obtains the corresponding associated process route from the process-process association rules 405 based on the first process information. In another embodiment, when the first process information does not match the second process information, the process and process association module outputs a prompt message to the corresponding terminal device (i.e., the user's terminal device), and the process and process association module selects the process information corresponding to the instruction from the first process information and the second process information as the associated process route based on the received selection instruction. In step S460, the processor 110 compares the verified associated process routes with the optimal process route 305 generated by the deep learning model. When the process parameters 4062 between the associated process route and the optimal process route 305 are consistent, the processor 110 selects the associated process route or the optimal process route 305 as a recommended process route (i.e., final process route recommendation 406). The processor 110 also presents a bill of materials 4061 and process parameters 4062 with timing characteristics for the recommended process route. In another embodiment, when the process parameters 4062 between the associated process route and the optimal process route 305 are inconsistent, the processor 110 prioritizes matching based on the process-process association rules 405 and generates a recommended process route that complies with the process-process association rules 405. In other words, the processor 110 matches the associated process routes or the optimal process route 305 with identical process parameters 4062 based on the process-process association rules 405 to generate a recommended process route. In step S470, the procurement calculation module 124 performs procurement calculations on the bill of materials 4061 corresponding to the recommended process route based on the inventory data in inventory information maintenance 407. The module then determines whether the material inventory is sufficient to meet the bill of materials 4061 and quantity in the recommended process route, thereby determining whether procurement is necessary. If the material quantity in the inventory data is less than the material list and data in the recommended process route, the procurement calculation module 124 generates a supplier recommendation 408 with the lowest procurement cost based on supplier information (including material prices and material types from multiple suppliers). In summary, the process route generation system and method of the present invention can establish long-short-term memory models, regression models, and association rules between process data based on historical process data to generate a corresponding process route based on user-entered demand data. Notably, the process route generation system generates a process route based on the expected performance in the demand data and compares the process route generated based on the association rules with the process route generated by the deep learning model to generate a process route that meets customer needs with high accuracy. Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention. 100: Process route generation system 110: Processor 120: Storage device 121: Regression analysis module 122: Correlation analysis module 123: Deep learning module 124: Procurement calculation module 301: Process history data 302: Material inventory and cost data 303: Correlation between materials and processes 304: Correlation between processes 305: Optimal process route 3051: Deep learning model database 306: Relationship between process parameters and material type, dosage and performance 307: Regression model database 308: Process and process correlation database 309: Material and process correlation Database 401: Input data 411: Application data sheet 402: Customer requirements 403: Regression process parameters, material data and performance result data 404: Material and process parameter association rules 405: Process and process association rules 406: Final process route recommendation 4061: Bill of materials 4062: Process parameters 407: Inventory information maintenance 408: Supplier recommendation S210~S250, S310~S350, S410~S470: Steps S321~S324, S331~S333, S341~S344, S351~S353: Steps FIG1 is a schematic diagram of a process route generation system according to an embodiment of the present invention. FIG2 is a flow chart of a process route generation method according to an embodiment of the present invention. FIG3A and FIG3B are flow charts of data analysis of the process route generation method according to an embodiment of the present invention. FIG4A and FIG4B are schematic diagrams of the architecture of a process route generation system according to an embodiment of the present invention. S210~S250: Steps

Claims

1. A process route generation system, comprising: Storage device, storing multiple modules; and a processor, coupled to the storage device, for executing the plurality of modules, the plurality of modules including: a regression analysis module, which executes a regression algorithm based on demand data to derive process parameters and material information; A correlation analysis module performs correlation analysis based on the process parameters to obtain first process information, and the correlation analysis module obtains second process information based on the material information. The correlation analysis module includes a process-to-process correlation module, which obtains a related process route based on either the first process information or the second process information. A deep learning module generates candidate process routes based on the demand data. The processor compares the related process routes and the candidate process routes to generate a recommended process route. The correlation analysis module uses a frequent pattern growth algorithm to build a frequent pattern tree structure based on the input data, and then obtains the correlation rules between multiple process parameters in the process route based on frequent itemsets. The correlation analysis module performs the correlation analysis based on the frequent pattern tree structure.

2. The process route generation system as described in claim 1, wherein the plurality of modules further includes: The procurement calculation module generates recommended supplier information based on material inventory, cost data, and the recommended process route, and generates suggested procurement messages based on material inventory and the recommended supplier information. The processor outputs the recommended process route, the process parameters corresponding to the recommended process route, the suggested procurement messages, and the required material list for the recommended process route.

3. The process route generation system as described in claim 1, wherein the demand data includes at least one of expected finished product performance, finished product specifications, and manufacturing budget, and the recommended process route includes process route, process parameters, bill of materials, and supplier information.

4. The process route generation system as described in claim 1, wherein the process and process association module performs a verification process on the first process information and the second process information to determine whether the first process information matches the second process information; when the first process information matches the second process information, the process and process association module obtains the associated process route based on the first process information; when the first process information does not match the second process information, the process and process association module outputs a prompt message to the corresponding terminal device, and the process and process association module receives a selection instruction, and then uses the corresponding process information from the first process information and the second process information as the associated process route according to the selection instruction.

5. The process route generation system as described in claim 1, further comprising a data acquisition module and a data preprocessing module, wherein the data acquisition module is coupled to at least one of a server, a database, and an electronic device to receive the input data and the demand data, wherein the input data includes historical process data, customer demand data, and material inventory cost data, wherein the historical process data includes historical process routes, process parameters, required material information, and product performance test results; wherein the data preprocessing module performs data preprocessing on the input data received by the data acquisition module, wherein the data preprocessing includes at least one of data cleaning, missing value handling, and data standardization.

6. The process route generation system as described in claim 1, wherein the deep learning module divides the input data into training set data, test set data, and validation set data, and the deep learning module trains a long short-term memory (LSTM) model on the training set data, the test set data, and the validation set data based on a deep neural network, wherein the deep learning module generates the candidate process route based on the demand data by executing the LSTM model.

7. The process route generation system as described in claim 1 further includes a correlation analysis module, wherein the correlation analysis module performs correlation analysis based on the performance categories in the input data and the variable parameters in the input data, and then uses variables with a confidence level greater than a threshold as influencing factors to obtain the influencing factor for each performance category.

8. The process route generation system as described in claim 7, wherein the regression analysis module is a regression model, and wherein the processor trains the regression model based on process parameters, material data, and product performance using a deep neural network.

9. The process route generation system as described in claim 8, wherein the deep neural network is a Support Vector Regression (SVR) algorithm, wherein the regression analysis module divides the performance category and the influencing factor into test set data and training set data, wherein the regression analysis module trains the regression model on the test set data and the training set data based on the deep neural network, and then optimizes the parameter selection of the support vector regression through a genetic algorithm, wherein the coefficient of determination of the regression model is R-squared, and the regression analysis module judges the model training effect based on the coefficient of determination, and then takes the model with the highest R-squared value as the benchmark model.

10. A method for generating a process route, comprising: The regression analysis module executes regression algorithms based on demand data to derive process parameters and material information; The steps of obtaining first process information by performing correlation analysis based on the process parameters using the correlation analysis module, and obtaining second process information based on the material information using the correlation analysis module; obtaining associated process routes based on either the first process information or the second process information using the process-to-process correlation module; generating candidate process routes based on the demand data using the deep learning module; and generating recommended process routes by comparing the associated process routes and the candidate process routes using the processor, wherein the steps of obtaining first process information by performing correlation analysis based on the process parameters using the correlation analysis module and obtaining second process information based on the material information using the correlation analysis module include: establishing a frequent pattern tree structure based on the input data using the frequent pattern growth algorithm using the correlation analysis module, and then obtaining association rules between multiple process parameters in the process route using the correlation analysis module based on frequent itemsets, wherein the correlation analysis module performs the correlation analysis based on the frequent pattern tree structure.

11. The process route generation method as described in claim 10, wherein the method further comprises: The procurement calculation module generates recommended supplier information based on material inventory, cost data, and the recommended process route, and generates suggested procurement messages based on material inventory and the recommended supplier information; and the processor outputs the recommended process route, the process parameters corresponding to the recommended process route, the suggested procurement messages, and the required material list for the recommended process route.

12. The process route generation method as described in claim 10, wherein the demand data includes at least one of expected finished product performance, finished product specifications, and manufacturing budget, and the recommended process route includes process route, process parameters, bill of materials, and supplier information.

13. The process route generation method as described in claim 10, wherein the step of the process and process association module obtaining the associated process route based on one of the first process information and the second process information includes: When the first process information matches the second process information, the associated process route is obtained through the process and process association module based on the first process information; When the first process information does not match the second process information, a prompt message is output to the corresponding terminal device through the process and process association module, and a selection instruction is received through the process and process association module. Then, the process information corresponding to the first process information and the second process information is selected as the associated process route according to the selection instruction.

14. The process route generation method as described in claim 10, wherein the method further comprises: The input data and the demand data are received by the data acquisition module. The input data includes historical process data, customer demand data, and material inventory cost data. The historical process data includes historical process routes, process parameters, required material information, and product performance test results. The input data received by the data acquisition module is preprocessed by the data preprocessing module. The data preprocessing includes at least one of the following: data cleaning, missing value handling, and data standardization.

15. The process route generation method as described in claim 10, further comprising: The input data is divided into training set data, test set data, and validation set data by the deep learning module, and a long short-term memory (LSTM) model is trained on the training set data, the test set data, and the validation set data by the deep learning module based on a deep neural network. The step of generating the candidate process route based on the requirement data by the deep learning module includes: generating the candidate process route based on the requirement data by executing the long short-term memory model by the deep learning module.

16. The process route generation method as described in claim 10, wherein the process route generation method further comprises: The correlation analysis module performs correlation analysis based on the performance categories and variable parameters in the input data, and then uses variables with a confidence level greater than a threshold as influencing factors to obtain the influencing factor for each performance category.

17. The process route generation method as described in claim 16, wherein the regression analysis module is a regression model, and wherein the process route generation method further includes: The processor trains the regression model based on process parameters, material data, and product performance using a deep neural network.

18. The process route generation method as described in claim 17, wherein the deep neural network is a Support Vector Regression (SVR) algorithm, and wherein the process route generation method further includes: The regression analysis module divides the performance category and the influencing factors into test set data and training set data. The regression analysis module trains the regression model on the test set data and the training set data based on a deep neural network. Then, the parameter selection of the support vector regression is optimized through a genetic algorithm. The regression analysis module judges the model training effect based on the coefficient of determination, wherein the coefficient of determination of the regression model is R-squared. The regression analysis module then uses the model with the highest R-squared value as the benchmark model.

Images