Aluminum alloy casting process data filling method based on process association

Abstract

The present invention discloses a method for filling aluminum alloy casting process data based on process association. The method comprises: obtaining and preprocessing aluminum alloy casting process data; dividing the preprocessed aluminum alloy casting process data into different process data sets according to product type; for each process data set, determining the execution order of equipment groups based on the aluminum alloy production process flow, and binding corresponding process parameters to each equipment group; constructing a process parameter prediction model based on an elastic neural network, whose input includes the process parameters of the first to i-th equipment groups and whose output includes the process parameters of the i-th equipment group; and using the process parameter prediction model to fill in the process parameters of each equipment group according to the execution order. The data filling scheme of the present invention inherits the timing constraints of the aluminum alloy casting process and integrates the deep coupling relationship between the process parameters of multiple equipment groups, thereby improving the filling accuracy of aluminum alloy casting process data.

Description

Technical Field

[0001] The present invention relates to the technical field of production data processing, and in particular to a method for filling in aluminum alloy casting process data based on process association. Background Art

[0002] The aluminum processing industry is currently at a critical stage of intelligent transformation, but most companies still face the daunting challenge of managing production data. Their aluminum alloy casting workshop generates over 2,000 process parameter records daily, most of which are manually recorded. These include key indicators such as melting temperature, holding temperature, electrolytic aluminum liquid temperature, degassing box speed, and casting cooling rate. However, oversight in manual data entry resulted in a 15.2% missing key parameter rate.

[0003] Specifically, during the smelting process, operators often omit the resting time and slag removal frequency due to the complex on-site environment (such as low instrument visibility in the high-temperature furnace area). During the casting stage, when multiple strand casting machines are operating in parallel, workers can easily confuse cooling water flow records for different molds. This fragmented data situation leads to three typical problems when companies build quality prediction models: First, traditional interpolation methods often use unified interpolation, failing to account for the unique process characteristics of different product models. Second, traditional interpolation methods (such as mean filling and linear interpolation) ignore the process logic between processes. For example, the peak temperature of the smelting furnace #1 (normal range 680-720°C) is incorrectly imputed to the slag removal process of smelting furnace #2 (process upper limit 650°C), resulting in inaccurate predictions of equipment energy consumption. Third, the isolation of single equipment parameters severs the linkage between multiple processes. For example, the failure to establish a nonlinear mapping between the degassing box rotor speed and the casting defect rate leads to a surge in scrap rates.

[0004] Therefore, it is urgent to develop a data filling solution that conforms to the characteristics of the aluminum processing and casting process. It must inherit the timing constraints of the aluminum alloy casting process and analyze the deep coupling relationship of multiple equipment parameters to provide data support for the company's subsequent establishment of a large industrial model for aluminum alloy casting production. Summary of the Invention

[0005] The present invention provides a method for filling in aluminum alloy melting and casting process data based on process association, aiming to quickly and accurately complete the filling in of aluminum alloy melting and casting process data based on process association.

[0006] In a first aspect, a method for filling aluminum alloy casting process data based on process association is provided, comprising the following steps:

[0007] S1: Acquire aluminum alloy casting process data and perform preprocessing;

[0008] S2: Divide the pre-processed aluminum alloy casting process data into different process data sets according to product types;

[0009] S3: For each process data set, the execution order of the equipment groups is determined according to the aluminum alloy production process flow, and each equipment group is bound to the corresponding process parameters;

[0010] S4: construct a process parameter prediction model based on an elastic neural network, wherein the input includes the process parameters of the first equipment group to the i-th equipment group, and the output includes the process parameters of the i-th equipment group, i ≥ 1;

[0011] S5: Use the process parameter prediction model to fill in the process parameters of each equipment group in the execution order.

[0012] Furthermore, in step S1, the preprocessing process includes:

[0013] Convert categorical variables in process parameters into numerical values;

[0014] Outlier filtering, replacing outliers with NaN and retaining missing labels;

[0015] Unify the data types of all process parameters.

[0016] Furthermore, step S5 specifically includes:

[0017] The process parameters of each equipment group are filled in according to the execution order of the equipment groups. When filling in the process parameters of the i-th equipment group, multiple pieces of filled process parameter data from the 1st equipment group to the i-th equipment group in the current process data set are extracted to construct a sample set, and then the process parameter prediction model corresponding to the i-th equipment group is trained; for each piece of data with missing process parameters of the i-th equipment group, the process parameter prediction model obtained by training is used to predict the process parameters of the i-th equipment group in the data, and the predicted process parameters are used to fill in the missing process parameters of the i-th equipment group in the data.

[0018] Furthermore, before filling in the process parameters of each equipment group according to the execution order using the process parameter prediction model, it also includes:

[0019] If in the current process data set, the process parameters of a certain equipment group are all empty, it is considered that the production process of the current product does not include the process of this equipment group, and the process parameters of this equipment group do not need to be filled; otherwise, the process parameters of this equipment group need to be filled.

[0020] Furthermore, for each piece of data in which the process parameters of the i-th equipment group are missing, the process parameters of the i-th equipment group in the data are predicted by the trained process parameter prediction model, specifically including:

[0021] For each piece of data with missing process parameters for the i-th equipment group, the median of the corresponding process parameters in the current process data set is first used to fill in the missing process parameters in the data;

[0022] Normalize the process parameters of the first to the i-th equipment groups in the data to form the input vector of the process parameter prediction model;

[0023] The input vector is input into the trained process parameter prediction model to predict the process parameters of the i-th equipment group in the data.

[0024] Furthermore, in step S5, when the number of samples in the sample set is less than a threshold, for each piece of data in which the process parameters of the i-th equipment group are missing, the median of the corresponding process parameters in the current process data set is used to fill in the missing process parameters in the data.

[0025] Furthermore, in step S5, after filling in the missing process parameters in the i-th equipment group in a certain data using the process parameters predicted by the process parameter prediction model, if the filled-in process parameters exceed the preset range of the process parameters, the median of the corresponding process parameters in the current process data set is used to fill in the missing process parameters in the data.

[0026] In a second aspect, a system for filling in aluminum alloy casting process data based on process association is provided, comprising:

[0027] Data acquisition module, used to acquire aluminum alloy casting process data and perform preprocessing;

[0028] A data partitioning module is used to divide the pre-processed aluminum alloy casting process data into different process data sets according to product types;

[0029] The equipment group division module is used to determine the equipment group execution order according to the aluminum alloy production process for each process data set, and bind each equipment group to the corresponding process parameters;

[0030] A model building module is used to build a process parameter prediction model based on an elastic neural network, wherein the input includes the process parameters of the first equipment group to the i-th equipment group, and the output includes the process parameters of the i-th equipment group, i≥1;

[0031] The data filling module is used to fill in the process parameters of each equipment group according to the execution order using the process parameter prediction model.

[0032] According to a third aspect, an electronic device is provided, including:

[0033] Memory on which computer programs or instructions are stored;

[0034] The processor is used to load and execute the computer program or instructions to implement the aluminum alloy casting process data filling method based on process association as described above.

[0035] In a fourth aspect, a computer-readable storage medium is provided, on which a computer program or instruction is stored. When the computer program or instruction is executed by a processor, the method for filling in aluminum alloy casting process data based on process association as described above is implemented.

[0036] The present invention proposes a method for filling aluminum alloy casting process data based on process association, which has the following beneficial effects compared with the existing technology:

[0037] (1) The present invention divides the aluminum alloy casting process data into different process data sets according to product type, and then performs independent data filling for each process data set. This takes into account the unique process characteristics of different types of products and avoids the errors caused by mixing process parameters of different types of products for filling.

[0038] (2) The present invention realizes the prediction of missing process parameters through a process parameter prediction model based on an elastic neural network. In addition, the process parameters of the current equipment group and the equipment groups in the previous process are used as inputs of the process parameter prediction model to predict the process parameters of the current equipment group, thereby filling in the missing process parameters. The deep coupling relationship between the equipment groups is fully considered, and a nonlinear mapping between the process parameters is established based on the multi-process linkage relationship, thereby realizing the accurate prediction of the missing process parameters.

[0039] (3) The process parameter prediction model is used to fill in the process parameters of each equipment group according to the execution order, inheriting the timing constraints of the aluminum alloy casting process and further improving the prediction accuracy of the missing process parameters;

[0040] (4) The present invention can achieve high-precision aluminum alloy casting process data filling, providing data support for the company to subsequently establish a large-scale industrial model for aluminum alloy casting production. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0042] Figure 1 This is a flow chart of a method for filling in aluminum alloy casting process data based on process association provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0043] To make the objectives, technical solutions, and advantages of the present invention more apparent, the technical solutions of the present invention will be described in detail below. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other implementations obtained by those of ordinary skill in the art without inventive effort are within the scope of protection of the present invention.

[0044] like Figure 1 As shown, an embodiment of the present invention discloses a method for filling aluminum alloy casting process data based on process association, comprising the following steps:

[0045] S1: Obtain aluminum alloy casting process data and perform preprocessing.

[0046] Specifically, the aluminum alloy melting and casting process data is stored in an Excel file. The Excel file contains process parameters of each process of the aluminum alloy production process, and the process parameters are sorted by column.

[0047] After reading the Excel file containing aluminum alloy casting process data, data preprocessing is required. The preprocessing process includes:

[0048] Numerical conversion of categorical variables: Numerical conversion of categorical variables in process parameters. For example, converting "qualified" fields in process parameters (such as quality inspection results) into binary values ​​("all qualified" → 1, others → 0) to ensure that the model can process them.

[0049] Outlier filtering: replace outliers with NaN and retain missing markers; outliers include missing values ​​and values ​​of current process parameters that exceed the set range of the process parameters;

[0050] Data type unification: Unify the data types of all process parameters, such as forcing them to be converted to float32 type, to optimize memory usage and computing efficiency.

[0051] S2: Divide the pre-processed aluminum alloy casting process data into different process data sets according to product type.

[0052] Because different product types have their own unique process characteristics, to avoid errors caused by mixing process parameters from different product types during infill, the data is divided into different process data sets by product type. Each process data set corresponding to each product is infilled independently without affecting each other, and the infilling of multiple process data sets can be processed in parallel. In this embodiment, the first column of the Excel file containing aluminum alloy casting process data contains the product number. Therefore, during implementation, the process data sets can be quickly divided according to the product number in the first column.

[0053] S3: For each process data set, the execution order of the equipment groups is determined according to the aluminum alloy production process flow, and each equipment group is bound to the corresponding process parameters.

[0054] The equipment group execution sequence enforces adherence to the production process sequence to ensure data fills conform to the time dependencies of actual production. After determining the equipment group execution sequence, you need to bind the process parameter column range corresponding to each equipment group. Each process parameter has a corresponding column number, which can be used to categorize equipment groups.

[0055] In one example, the device groups and their order are divided as follows:

[0056] ("Smelting 1", (1, 9)), ("Smelting 1 Standing", (10, 11)), ("Smelting 1 Slag Removal", (12, 13)), ("Smelting 2 Slag Removal", (14, 15)), ("Smelting 2", (16, 18)), ("Smelting 2 Standing", (19, 20)), ("Smelting 2 Slag Removal", (21, 22)), ("Smelting 2 Pour Down", (23, 24)), ("Housing 1", (25, 28)), ("Housing 1 Slag Removal", (29,30)), ("Housing 2", (31, 35)), ("1# Degassing Box 1# Rotor", (36, 43)), ("2# Degassing Box 1# Rotor", (44,49)), ("1# Degassing Box 2# Rotor", (50, 55)), ("2# Degassing Box 2# Rotor", (56, 61)), ("Casting", (62,80)). Taking ("Smelting 1", (1, 9)) as an example, it means that the process parameters corresponding to the "Smelting 1" equipment group are the data in columns 1-9. Due to the large number of process parameters, some representative data are selected for display, as shown in Table 1:

[0057] ;

[0058] In Table 1, data for equipment groups such as "Smelting 1 Slag Stripping", "Smelting 2 Slag Demolition", "Smelting 2", and "Smelting 2 Standing" are omitted for better presentation.

[0059] S4: Construct a process parameter prediction model based on an elastic neural network, whose input includes the process parameters of the 1st equipment group to the i-th equipment group, and whose output includes the process parameters of the i-th equipment group, i≥1.

[0060] To address the large variations in process parameter scale across equipment groups (e.g., 9 dimensions for a smelting equipment group and 19 dimensions for a casting equipment group), a flexible neural network architecture was designed to dynamically adjust input and output dimensions to accommodate different process scenarios. Dynamic dimensionality adjustment: The input layer dimensions automatically scale with the number of process parameters in the equipment group, while the decoding layer synchronizes to match reconstruction requirements. Overfitting resistance: The Dropout layer (with a probability of 0.2) randomly masks neurons to enhance generalization for sparse data. Gradient clipping (clip_grad_norm_(1.0)) limits parameter updates to prevent training divergence. An early stopping mechanism terminates training after 20 consecutive rounds of loss without improvement, balancing efficiency and accuracy. The loss can be the mean squared error of the input and output.

[0061] S5: Use the process parameter prediction model to fill in the process parameters of each equipment group in the order of execution; when filling in the process parameters of the i-th equipment group, extract multiple pieces of filled process parameter data from the 1st equipment group to the i-th equipment group in the current process data set to construct a sample set, and then train the process parameter prediction model corresponding to the i-th equipment group; for each piece of data with missing process parameters of the i-th equipment group, use the trained process parameter prediction model to predict the process parameters of the i-th equipment group in the data, and use the predicted process parameters to fill in the missing process parameters of the i-th equipment group in the data.

[0062] For example, when filling the first equipment group, namely "Smelting 1": find the rows with complete "Smelting 1" process parameters (all data in columns 1-9 are complete) (as valid data), select the complete "Smelting 1" process data in each row as a sample, build a sample set, and then use the sample set to train the process parameter prediction model; for the remaining "Smelting 1" process parameters, the missing process parameters are temporarily filled with the median of the column, so as to build the model input so that the data can be fully input into the model, and the constructed model input is input into the trained process parameter prediction model to predict the "Smelting 1" process parameters, and then the predicted process parameters are overwritten to the position previously filled with the median, completing the filling process of the first equipment "Smelting 1" process parameters.

[0063] When filling in the second equipment group, "Smelting 1, Standing", we first select rows with complete data for both "Smelting 1" and "Smelting 1, Standing" (with columns 1-11 complete) as valid data. During this filling process, the process parameter prediction model is trained and predicted using all process parameters from "Smelting 1" and "Smelting 1, Standing" as input, completing cross-process parameter transfer. The rest of the processing follows the same principles as for the "Smelting 1" filling process. Finally, data filling for the second equipment group, "Smelting 1, Standing", is complete.

[0064] Similarly, to fill in the third device group, we need to find rows that also contain complete data for device groups 1-3 for training, ultimately completing the fill-in process. This process continues in this order until we reach the final device group. This strategy of "temporary median fill-in → model prediction → overwrite correction" is used to handle missing values. Ultimately, the entire subsequent fill-in process inherits the previously filled-in process parameters, which is consistent with actual engineering practices (subsequent processes are adjusted based on the previous process).

[0065] The above embodiment provides a method for filling in aluminum alloy melting and casting process data based on process association, which has the following beneficial effects: by dividing the aluminum alloy melting and casting process data into different process data sets according to product type, and then performing independent data filling for each process data set, the unique process characteristics of different types of products are taken into account, and the errors caused by filling in the process parameters of mixed different types of products are avoided; the missing process parameters are predicted by a process parameter prediction model based on an elastic neural network, and the process parameters of the current equipment group and the equipment groups in its previous process are used as inputs of the process parameter prediction model to predict the process parameters of the current equipment group, thereby filling in the missing process parameters, fully considering the deep coupling relationship between the equipment groups, establishing a nonlinear mapping between the process parameters based on the multi-process linkage relationship, and realizing the accurate prediction of the missing process parameters; using the process parameter prediction model to fill in the process parameters of each equipment group in the execution order, inheriting the timing constraints of the aluminum alloy melting and casting process, and further improving the prediction accuracy of the missing process parameters; high-precision aluminum alloy melting and casting process data filling can be achieved, providing data support for the company to subsequently establish a large industrial model of aluminum alloy melting and casting production.

[0066] Considering that different products may have different production processes, in some embodiments, before using the process parameter prediction model to fill in the process parameters of each equipment group in the execution order, dynamic process integrity detection is required, specifically including:

[0067] Empty judgment: If all process parameters of a certain equipment group are empty in the current process data set, it is considered that the production process of the current product does not include the process of this equipment group, and the process parameters of this equipment group do not need to be filled. As shown in Table 1, if the data of the "Smelting 2 Slag Stripping" equipment group are all empty, it is considered that this product has not passed through the process of this equipment group, and therefore the process parameters of this equipment group do not need to be filled.

[0068] Execution status decision: The filling process is initiated only when at least one valid data point exists for the equipment group; otherwise, it is skipped to avoid interfering with unexecuted processes. For example, if data is detected for process parameters other than "Smelting 1 Slag Agent Dosage" and "Refining 1 Temperature" in the equipment group "Smelting 1," then it is assumed that the product must undergo all processes in the equipment group "Smelting 1." By default, all processes should have data, and their missing values ​​need to be filled.

[0069] Since process parameters have different dimensions, in order to eliminate the dimension difference, in some embodiments, for each piece of data with missing process parameters of the i-th equipment group, the process of predicting the process parameters of the i-th equipment group in the data using the trained process parameter prediction model includes:

[0070] For each piece of data with missing process parameters for the i-th equipment group, the median of the corresponding process parameters in the current process data set is first used to fill in the missing process parameters in the data;

[0071] The process parameters of the first to the i-th equipment groups in the data are normalized using MinMaxScaler to form the input vector of the process parameter prediction model to eliminate dimensional differences;

[0072] The input vector is input into the trained process parameter prediction model to predict the process parameters of the i-th equipment group in the data. Of course, since the predicted process parameters correspond to normalized values, a denormalization process is required.

[0073] In specific implementation, when predicting the process parameters of the i-th equipment group, the output of the process parameter prediction model can be selected to be the same as the input dimension, that is, the model input is composed of the process parameters of the 1st equipment group to the i-th equipment group, and the model outputs the process parameters of the 1st equipment group to the i-th equipment group; of course, the output of the process parameter prediction model can also be selected to be different from the input dimension, that is, the model input is composed of the process parameters of the 1st equipment group to the i-th equipment group, and the model outputs the process parameters of the i-th equipment group.

[0074] Of course, in some cases, the process parameter prediction model may not be able to accurately fill in the missing process parameters. Therefore, in some embodiments, a mixed filling method is adopted: (1) when the number of samples in the sample set is less than a threshold (such as the number of samples is less than 10) resulting in the failure of the process parameter prediction model training, for each data with missing process parameters of the i-th equipment group, the median of the corresponding process parameters in the current process data set is used to fill in the missing process parameters in the data; (2) after filling in the missing process parameters of the i-th equipment group in a certain data using the process parameters predicted by the process parameter prediction model, if the filled process parameters exceed the preset range of the process parameters, the median of the corresponding process parameters in the current process data set is used to fill in the missing process parameters in the data.

[0075] It should be noted that the process data must be filled in strictly according to the execution order of the equipment groups; every time the process parameters of an equipment group are filled, it is necessary to verify whether the filled process parameter values ​​are within the preset range of the process parameters. Only after the verification is passed can the data be used to fill the process parameters of the subsequent equipment groups to prevent excessive data from contaminating subsequent processes.

[0076] The embodiment of the present invention further discloses a system for filling aluminum alloy casting process data based on process association, comprising:

[0077] Data acquisition module, used to acquire aluminum alloy casting process data and perform preprocessing;

[0078] A data partitioning module is used to divide the pre-processed aluminum alloy casting process data into different process data sets according to product types;

[0079] The equipment group division module is used to determine the equipment group execution order according to the aluminum alloy production process for each process data set, and bind each equipment group to the corresponding process parameters;

[0080] A model building module is used to build a process parameter prediction model based on an elastic neural network, wherein the input includes the process parameters of the first equipment group to the i-th equipment group, and the output includes the process parameters of the i-th equipment group, i≥1;

[0081] The data filling module is used to fill in the process parameters of each equipment group according to the execution order using the process parameter prediction model.

[0082] It should be understood that the functional unit modules in various embodiments of the present invention can be concentrated in one processing unit, or each unit module can exist physically separately, or two or more unit modules can be integrated into one unit module, and can be implemented in the form of hardware or software.

[0083] An embodiment of the present invention further discloses an electronic device, including:

[0084] Memory on which computer programs or instructions are stored;

[0085] The processor is used to load and execute the computer program or instructions to implement the aluminum alloy casting process data filling method based on process association as described above.

[0086] An embodiment of the present invention further discloses a computer-readable storage medium having a computer program or instruction stored thereon. When the computer program or instruction is executed by a processor, the method for filling in aluminum alloy casting process data based on process association as described above is implemented.

[0087] It can be understood that the same or similar parts of the above embodiments can be referenced to each other, and the contents not described in detail in some embodiments can refer to the same or similar contents in other embodiments.

[0088] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0089] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0090] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0091] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0092] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A method for filling aluminum alloy casting process data based on process association, characterized in that: The steps include: S1: Acquire aluminum alloy casting process data and perform preprocessing; S2: Divide the pre-processed aluminum alloy casting process data into different process data sets according to product types; S3: For each process data set, the execution order of the equipment groups is determined according to the aluminum alloy production process flow, and each equipment group is bound to the corresponding process parameters; S4: construct a process parameter prediction model based on an elastic neural network, wherein the input includes the process parameters of the first equipment group to the i-th equipment group, and the output includes the process parameters of the i-th equipment group, i ≥ 1; S5: Fill in the process parameters of each equipment group according to the execution order using the process parameter prediction model; Step S5 specifically includes: Fill in the process parameters of each equipment group in the order in which the equipment groups are executed. When filling in the process parameters of the i-th equipment group, extract multiple pieces of fully filled process parameter data from the first to the i-th equipment groups in the current process data set to construct a sample set, and then train the process parameter prediction model corresponding to the i-th equipment group; for each piece of data with missing process parameters of the i-th equipment group, predict the process parameters of the i-th equipment group in the data using the trained process parameter prediction model, and use the predicted process parameters to fill in the missing process parameters of the i-th equipment group in the data; For each piece of data in which the process parameters of the i-th equipment group are missing, the process parameters of the i-th equipment group in the data are predicted by the trained process parameter prediction model, specifically including: For each piece of data with missing process parameters for the i-th equipment group, the median of the corresponding process parameters in the current process data set is first used to fill in the missing process parameters in the data; Normalize the process parameters of the first to the i-th equipment groups in the data to form the input vector of the process parameter prediction model; The input vector is input into the trained process parameter prediction model to predict the process parameters of the i-th equipment group in the data.

2. The method for filling in aluminum alloy casting process data based on process association according to claim 1, characterized in that: In step S1, the preprocessing process includes: Convert categorical variables in process parameters into numerical values; Outlier filtering, replacing outliers with NaN and retaining missing labels; Unify the data types of all process parameters.

3. The method for filling in aluminum alloy casting process data based on process association according to claim 1, characterized in that: Before filling in the process parameters of each equipment group in the execution order using the process parameter prediction model, it also includes: If in the current process data set, the process parameters of a certain equipment group are all empty, it is considered that the production process of the current product does not include the process of this equipment group, and the process parameters of this equipment group do not need to be filled; otherwise, the process parameters of this equipment group need to be filled.

4. The method for filling in aluminum alloy casting process data based on process association according to claim 1, characterized in that: In step S5, when the number of samples in the sample set is less than the threshold, for each piece of data in which the process parameters of the i-th equipment group are missing, the median of the corresponding process parameters in the current process data set is used to fill in the missing process parameters in the data.

5. The method for filling in aluminum alloy casting process data based on process association according to claim 1, characterized in that: In step S5, after filling in the missing process parameters in the i-th equipment group in a certain data using the process parameters predicted by the process parameter prediction model, if the filled process parameters exceed the preset range of the process parameters, the median of the corresponding process parameters in the current process data set is used to fill in the missing process parameters in the data.

Images