Intelligent temperature control method for casting system

Abstract

The invention provides a casting system temperature intelligent control method which comprises the following steps: acquiring a casting mold characteristic area, and setting a thermocouple in the mold characteristic area based on casting structure characteristics; a random forest model is established according to thermocouple temperature measurement results and casting quality detection results, recursive feature elimination is carried out, correlation between thermocouple temperature measurement data and casting quality is discriminated, the number of thermocouples is optimized, and the thermocouples are screened; and analyzing according to the temperature data of the thermocouple temperature measuring points, the cooling process parameters and the corresponding casting quality, constructing the relationship among the cooling process parameters, the initial temperature of each mold thermocouple and the casting quality through a gradient lifting decision tree model, and controlling the temperature of the casting system based on the relationship. By arranging the thermocouple and based on the random forest model and the gradient boosting decision tree model, the temperature control accuracy is improved, the temperature is adjusted in time, and the purposes of optimizing the casting process, reducing defects and ensuring high-quality manufacturing of the aluminum alloy hub are achieved.

Description

Technical Field

[0001] The invention belongs to the technical field of low-pressure casting process, and in particular relates to a casting system temperature intelligent control method. Background Art

[0002] Lightweighting is one of the important strategies for energy conservation and emission reduction of fuel vehicles and efficiency improvement of new energy vehicles. Among them, the use of aluminum alloy materials to replace traditional steel materials is a key means to achieve lightweighting of automobiles. The low-pressure casting process has become the mainstream process for producing automotive aluminum alloy castings due to its stable filling, good shrinkage compensation effect and high degree of automation. In the low-pressure casting process, the influence of mold temperature on the quality of castings is particularly important. The alloy melt is in direct contact with the mold during the filling and solidification process, and is cooled and solidified through heat exchange. Therefore, the temperature distribution of the mold directly determines the solidification sequence and cooling rate of the alloy melt, which has a key influence on the solidification structure and performance of the casting. By controlling the mold temperature within a reasonable range through a suitable cooling process, it can ensure that the alloy solidifies rapidly in the expected order, thereby significantly reducing casting defects such as shrinkage cavities and shrinkage, and shortening the production cycle, achieving a dual improvement in casting quality and production efficiency.

[0003] At present, in the low-pressure casting production of aluminum alloy castings, the temperature control of the mold mainly adopts an open-loop control method. Usually, the mold temperature is not detected and monitored in real time, but relies on on-site technicians to directly adjust the cooling process parameters according to the casting defects. Although this method is easy to operate, it has obvious defects: the adjustment of the cooling process is greatly affected by the experience and subjective judgment of the technicians, lacks scientific basis, resulting in unstable process improvement effects and difficulty in achieving automatic and intelligent control of mold temperature. With the advancement of the lightweight process of automobiles and the increasing complexity of casting structures, it is more challenging to achieve reasonable control of mold temperature. Traditional temperature control methods can no longer meet modern production needs.

[0004] Temperature measurement and control of low-pressure die-casting aluminum alloy wheel hub molds is a key task in the manufacturing field. It involves obtaining and analyzing temperature data from the mold operation status to achieve precise temperature control. Effective temperature measurement and control methods can help manufacturing companies reduce production costs, improve product quality and increase production efficiency. In the low-pressure casting process, the temperature of the mold has a direct impact on the molding quality and production stability of the aluminum alloy wheel hub. Therefore, accurate temperature measurement and timely temperature adjustment are particularly important. Existing temperature control methods have the problems of poor accuracy and untimely temperature adjustment. Summary of the invention

[0005] In view of the problems existing in the prior art, the present invention provides a casting system temperature intelligent control method, which at least partially solves the problems of poor accuracy and untimely temperature adjustment existing in the prior art.

[0006] An embodiment of the present invention provides a casting system temperature intelligent control method, comprising:

[0007] Acquire the characteristic area of ​​the casting mold, and set a thermocouple in the characteristic area of ​​the mold based on the structural characteristics of the casting;

[0008] Based on the thermocouple temperature measurement results and casting quality inspection results, a random forest model is established and recursive feature elimination is performed to determine the correlation between each thermocouple temperature measurement data and casting quality, thereby optimizing the number of thermocouples and obtaining the thermocouple temperature measurement positions corresponding to various casting defects, thereby screening the thermocouples.

[0009] According to the screened temperature data of thermocouple measuring points, cooling process parameters and corresponding casting quality, the relationship between cooling process parameters, the starting temperature of each mold thermocouple and casting quality is constructed through the gradient boosting decision tree model, and the casting system temperature is controlled based on the constructed relationship.

[0010] Optionally, obtaining the casting mold feature area includes:

[0011] Collect historical process data of castings and casting molds, and determine the characteristic areas of casting molds based on X-ray detection and simulated casting shrinkage volume results of historical process data.

[0012] Optionally, the historical process data of the casting and the casting mold are collected, including: cooling process plan data and casting quality data;

[0013] The cooling process plan data includes cooling channel configuration, switching time and / or coolant flow rate at each production stage;

[0014] The casting quality data includes casting quality indicators, and the casting quality indicators include defect type, defect location, characteristic area yield strength and / or tensile strength.

[0015] Optionally, a random forest model is established based on the thermocouple temperature measurement results and the casting quality inspection results, and recursive feature elimination is performed to determine the correlation between each thermocouple temperature measurement data and the casting quality, thereby optimizing the number of thermocouples, obtaining thermocouple temperature measurement positions corresponding to various defects of the casting, and screening the thermocouples, including:

[0016] A sensitivity test was conducted on the mold to select the temperature data that meets the mold closing time as the independent variable and the casting quality as the dependent variable for calculation. A random forest model was established and the importance of each feature was calculated. According to the feature importance, thermocouples that reflect the casting quality were selected as standard thermocouples in the top mold, bottom mold and side mold of the mold.

[0017] Optionally, establishing a random forest model based on the thermocouple temperature measurement results and the casting quality inspection results and performing recursive feature elimination includes:

[0018] Step 1: Use the acquired features to train the random forest model and calculate the weight or coefficient of each feature of the random forest model;

[0019] Step 2: Sort the features according to their weights or coefficients;

[0020] Step 3: Delete one or more features with the smallest weight or coefficient in the ranking and retrain the random forest model with the remaining features;

[0021] Step 4: Repeat steps 2 and 3 until the desired number of features is reached or no further features can be removed.

[0022] Optional, thermocouple temperature characteristic importance calculation, including:

[0023] Calculate the node Gini impurity;

[0024] Calculate the contribution of each feature;

[0025] Cumulative feature contribution;

[0026] Calculate the feature importance of all trees in the random forest and average the feature importance values ​​across all trees to get the final importance of the feature.

[0027] Optionally, the temperature data of the selected thermocouple temperature measuring points, cooling process parameters and corresponding casting quality are analyzed, and the relationship between the cooling process parameters, the starting temperature of each mold thermocouple and the casting quality is constructed through a gradient boosting decision tree model, and the casting system temperature is controlled based on the constructed relationship, including:

[0028] Obtain the selected temperature data of the thermocouple temperature measuring points, cooling process parameters and parameter data sets corresponding to the casting quality;

[0029] Under the condition of ensuring the quality of castings, the relationship model between the initial temperature of the mold feature area and the cooling process is output based on the parameter data set to predict the performance strength of the wheel hub;

[0030] The relationship model is optimized and iterated according to the difference between the casting quality and performance strength and the actual value;

[0031] The relationship model is evaluated based on the mean square error and the coefficient of determination. When the prediction success rate is greater than the set value, the optimization iteration is stopped.

[0032] Optionally, the cooling process scheme data and casting quality data are used to train a gradient boosting decision tree model and dynamically update the mold cooling process scheme and control rules;

[0033] A gradient boosted decision tree model was used to predict cooling process solutions that met the inspection criteria.

[0034] Optionally, the learning rate of the gradient boosting decision tree model is 0.01 to 0.3.

[0035] Optionally, the tree depth of the gradient boosting decision tree model is 3 to 10.

[0036] The intelligent temperature control method for a casting system provided by the present invention improves the accuracy of temperature control and adjusts the temperature in time by setting thermocouples and based on a random forest model and a gradient boosting decision tree model, so as to achieve the purpose of optimizing the casting process, reducing defects and ensuring high-quality manufacturing of aluminum alloy wheels. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The above and other objects, features and advantages of the present disclosure will become more apparent through a more detailed description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, wherein like reference numerals generally represent like components throughout the exemplary embodiments of the present disclosure.

[0038] Figure 1 A flow chart of the casting system temperature intelligent control technology provided by the embodiment of the present disclosure;

[0039] Figure 2 A schematic diagram of the specific positions of the mold thermocouple temperature measurement position, cooling channel and heating channel on the mold provided in the embodiment of the present disclosure;

[0040] Figure 3 A schematic diagram of thermocouple weight coefficients and feature importance ranking based on random forest provided in an embodiment of the present disclosure;

[0041] Figure 4 A schematic diagram of a method for constructing a gradient boosting decision tree model provided in an embodiment of the present disclosure;

[0042] Figure 5 A schematic diagram of the mold temperature measurement results of the actual production of castings after optimizing the process parameters provided in the embodiment of the present disclosure;

[0043] Figure 6a and Figure 6b The present invention provides an embodiment of the present invention that uses optimized process parameters to obtain an X-ray inspection image of the wheel core of the actual casting. DETAILED DESCRIPTION

[0044] The embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.

[0045] It should be clear that the following embodiments of the present disclosure are described by specific specific examples, and those skilled in the art can easily understand other advantages and effects of the present disclosure from the contents disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, rather than all of the embodiments. The present disclosure can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present disclosure. It should be noted that the following embodiments and features in the embodiments can be combined with each other in the absence of conflict. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present disclosure.

[0046] It should be noted that various aspects of the embodiments within the scope of the appended claims are described below. It should be apparent that the aspects described herein may be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on the present disclosure, it should be understood by those skilled in the art that an aspect described herein may be implemented independently of any other aspect, and two or more of these aspects may be combined in various ways. For example, any number of aspects described herein may be used to implement the device and / or practice the method. In addition, other structures and / or functionalities other than one or more of the aspects described herein may be used to implement this device and / or practice this method.

[0047] It should also be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present disclosure. The illustrations only show components related to the present disclosure rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0048] Additionally, in the following description, specific details are provided to facilitate a thorough understanding of the examples. However, it will be understood by those skilled in the art that the aspects described may be practiced without these specific details.

[0049] For ease of understanding, this embodiment discloses a casting system temperature intelligent control method, including:

[0050] Acquire the characteristic area of ​​the casting mold, and set a thermocouple in the characteristic area of ​​the mold based on the structural characteristics of the casting;

[0051] Based on the thermocouple temperature measurement results and casting quality inspection results, a random forest model is established and recursive feature elimination is performed to determine the correlation between each thermocouple temperature measurement data and casting quality, thereby optimizing the number of thermocouples and obtaining the thermocouple temperature measurement positions corresponding to various casting defects, thereby screening the thermocouples.

[0052] According to the screened temperature data of thermocouple measuring points, cooling process parameters and corresponding casting quality, the relationship between cooling process parameters, the starting temperature of each mold thermocouple and casting quality is constructed through the gradient boosting decision tree model, and the casting system temperature is controlled based on the constructed relationship.

[0053] Optionally, obtaining the casting mold feature area includes:

[0054] Collect historical process data of castings and casting molds, and determine the characteristic areas of casting molds based on X-ray detection and simulated casting shrinkage volume results of historical process data.

[0055] Optionally, the historical process data of the casting and the casting mold are collected, including: cooling process plan data and casting quality data;

[0056] The cooling process plan data includes cooling channel configuration, switching time and / or coolant flow rate at each production stage;

[0057] The casting quality data includes casting quality indicators, and the casting quality indicators include defect type, defect location, characteristic area yield strength and / or tensile strength.

[0058] Optionally, a random forest model is established based on the thermocouple temperature measurement results and the casting quality inspection results, and recursive feature elimination is performed to determine the correlation between each thermocouple temperature measurement data and the casting quality, thereby optimizing the number of thermocouples, obtaining thermocouple temperature measurement positions corresponding to various defects of the casting, and screening the thermocouples, including:

[0059] A sensitivity test was conducted on the mold to select the temperature data that meets the mold closing time as the independent variable and the casting quality as the dependent variable for calculation. A random forest model was established and the importance of each feature was calculated. According to the feature importance, thermocouples that reflect the casting quality were selected as standard thermocouples in the top mold, bottom mold and side mold of the mold.

[0060] Optionally, establishing a random forest model based on the thermocouple temperature measurement results and the casting quality inspection results and performing recursive feature elimination includes:

[0061] Step 1: Use the acquired features to train the random forest model and calculate the weight or coefficient of each feature of the random forest model;

[0062] Step 2: Sort the features according to their weights or coefficients;

[0063] Step 3: Delete one or more features with the smallest weight or coefficient in the ranking and retrain the random forest model with the remaining features;

[0064] Step 4: Repeat steps 2 and 3 until the desired number of features is reached or no further features can be removed.

[0065] Optional, thermocouple temperature characteristic importance calculation, including:

[0066] Calculate the node Gini impurity;

[0067] Calculate the contribution of each feature;

[0068] Cumulative feature contribution;

[0069] Calculate the feature importance of all trees in the random forest and average the feature importance values ​​across all trees to get the final importance of the feature.

[0070] Optionally, the temperature data of the selected thermocouple temperature measuring points, cooling process parameters and corresponding casting quality are analyzed, and the relationship between the cooling process parameters, the starting temperature of each mold thermocouple and the casting quality is constructed through a gradient boosting decision tree model, and the casting system temperature is controlled based on the constructed relationship, including:

[0071] Obtain the selected temperature data of the thermocouple temperature measuring points, cooling process parameters and parameter data sets corresponding to the casting quality;

[0072] Under the condition of ensuring the quality of castings, the relationship model between the initial temperature of the mold feature area and the cooling process is output based on the parameter data set to predict the performance strength of the wheel hub;

[0073] The relationship model is optimized and iterated according to the difference between the casting quality and performance strength and the actual value;

[0074] The relationship model is optimized and iterated according to the difference between the casting quality and performance strength and the actual value, including firstly calculating the maximum deviation Emax and the average deviation Eavg between the target value of casting quality inspection and the actual inspection value under the current cooling process conditions. Then, based on these deviations, the residual value is determined, and the opening time adjustment value t of the cooling channel is calculated by the residual value. openm and closing time adjustment value t closem The cooling process plan is modified according to these adjustment values ​​to obtain new cooling process settings. This optimization process is repeated until the actual quality detection value of the casting falls completely within the target range and the deviation from the target value is reduced to the preset range, and then the optimization is stopped.

[0075] The relationship model is evaluated based on the mean square error and the coefficient of determination. When the prediction success rate is greater than the set value, the optimization iteration is stopped. The set value is 90%.

[0076] Optionally, the cooling process scheme data and casting quality data are used to train a gradient boosting decision tree model and dynamically update the mold cooling process scheme and control rules;

[0077] A gradient boosted decision tree model was used to predict cooling process solutions that met the inspection criteria.

[0078] Optionally, the learning rate of the gradient boosting decision tree model is 0.01 to 0.3.

[0079] Optionally, the tree depth of the gradient boosting decision tree model is 3 to 10.

[0080] In a specific application scenario, take the wheel hub as an example:

[0081] Step 1: Determine the mold feature area and arrange the thermocouples.

[0082] First, it is necessary to collect the historical process data of the target product and mold. These data include numerical simulation results and trial production results. The characteristic area of ​​the casting mold is determined based on the X-ray detection and simulated wheel shrinkage volume results.

[0083] Next, a detailed database is established based on the collected process data. The database should include the entire production process of each product, covering the following key contents:

[0084] Cooling process plan data: record detailed information such as cooling channel configuration, switching time, coolant flow rate, etc. at each production stage.

[0085] Casting quality data: record the corresponding casting quality indicators, such as defect type, defect location, characteristic area yield strength, tensile strength, etc.

[0086] Each database record corresponds to the production process of a specific product. By systematically organizing and storing this data, it can provide a reliable foundation for subsequent process optimization and model training.

[0087] Step 2: Establish the thermocouple weight coefficient model and optimize the quantity.

[0088] Perform sensitivity test on mold: select a set of standard cooling process as reference. In the continuous production process, the first three hubs use standard cooling process to ensure that the mold temperature is at normal temperature and the casting quality is good. The last five hubs choose to close one cooling channel for testing in the standard cooling process. Repeat this process to complete the test of the closure of each cooling channel. Process the data collected by thermocouples, select the temperature data of the 1st second and the 100th second of the mold closing time as independent variables, and the wheel quality as the dependent variable for calculation. Establish a random forest model in python and calculate the importance of each feature. According to the importance of the feature, select a thermocouple that reflects the casting quality in each of the three major mold structures of the mold top mold, bottom mold and side mold as the standard thermocouple for subsequent calculation and analysis. The process is as follows. .

[0089] Step 1. Initialization: First, a model is trained using all available features and the weights or coefficients for each feature are calculated.

[0090] Step 2: Feature Sorting: Then, the features are sorted according to their weights or coefficients, and usually the features with the smallest weights or coefficients are considered the least important.

[0091] Step 3: Feature elimination: Next, remove one or more features with the smallest weight or coefficient in the ranking and retrain the model with the remaining features.

[0092] Step 4: Iterative process: Repeat steps 2 and 3 until the required number of features is reached or no further features can be removed.

[0093] The importance of each thermocouple temperature characteristic is calculated as follows:

[0094] 1. Training a Random Forest Model

[0095] Constructing random forest: Random forest is an integrated model composed of multiple decision trees. Each decision tree randomly selects a subset (sample and feature subset) from the original data during training, and then trains on these subsets.

[0096] 2. Calculate Gini Importance

[0097] Gini importance measures the contribution of each feature in reducing the impurity of the model (Gini impurity).

[0098] Calculate node Gini impurity: For each node j, calculate the Gini impurity Gj:

[0099]

[0100] Among them, p jk is the probability of the kth class in node j, where k is the number of classes.

[0101] Calculate the contribution of each feature: The contribution of feature i on node j is:

[0102]

[0103] Among them, G parent is the Gini impurity of the parent node, G left and G right are the Gini impurities of the left and right child nodes, N parent is the number of samples of the parent node, N left and N right are the number of samples of the left child node and the right child node respectively.

[0104] Cumulative feature contribution: For each tree t, the importance of feature i Ii(t) is the sum of the contributions of all nodes:

[0105]

[0106] Calculate the feature importance of all trees in the random forest: average the feature importance values ​​in all trees to get the final importance of feature i:

[0107]

[0108] Where T is the number of trees.

[0109] Step 3: Establish cooling process, mold initial temperature and casting quality model.

[0110] After eliminating the thermocouple positions, the temperature data of the selected thermocouple temperature measurement points, the cooling process parameters and the corresponding casting quality are analyzed, and the relationship between the cooling process parameters and the starting temperature of each mold thermocouple and the casting quality is constructed through the gradient boosting decision tree model. The specific operations are as follows:

[0111] The temperature data of each thermocouple temperature measuring point is extracted and the 1st second data of mold closing is T starti , where i represents different positions, arranged in order according to the importance of thermocouple characteristics. The air flow rate of the cooling channel is 80m 3 / h, water cooling channel is 5L / min. The cooling channel opening time is t openj , the closing time is t closej , where j represents different cooling channels. The parameter encoding for each casting is:

[0112] x=(T start1 , T peak1 , …, T startn , T peakn , t open1 , t close1 ,…,t openm , t closem ,)

[0113] y=(q1,q2,q3,…,q n , Rm1,…,Rm n , Q1,…,Q n ,)

[0114] y is the casting quality inspection, q iIt represents whether the i-th type 3 defect occurs, represented by 0 and 1, 0 represents no defect, and 1 represents defect. Rm1 represents the test result of the tensile strength performance of the hub in the first characteristic area, and Q1 represents the test result of the yield strength performance of the hub in the first characteristic area. The characteristic area includes the outer rim, inner rim, spokes, wheel core and rim of the hub and the finite element simulation results.

[0115] Under the condition of ensuring the quality of castings, the relationship model between the initial temperature of the mold feature area and the cooling process is output, and the performance strength of the wheel hub is predicted.

[0116] The model is optimized and iterated according to the difference between the casting quality and performance strength and the actual value: First, the maximum deviation Emax and the average deviation Eavg between the target value of the casting quality inspection and the actual inspection value under the current cooling process conditions are calculated. Then, based on these deviations, the residual value is determined, and the opening time adjustment value t of the cooling channel is calculated by the residual. openm and closing time adjustment value t closem The cooling process plan is modified according to these adjustment values ​​to obtain new cooling process settings. This optimization process is repeated until the actual quality detection value of the casting falls completely within the target range and the deviation from the target value is reduced to the preset range, and then the optimization is stopped.

[0117] The model is iterated based on the residuals and evaluated based on the mean square error and the coefficient of determination. When the prediction success rate is greater than 90%, the model can be used.

[0118] In step one, the database used for model training contains at least one set of trial production data; at least 10 sets of cooling process solutions are output through the prediction of the gradient boosting decision tree model for on-site production verification.

[0119] In step 1, the GBDT model is used to predict the cooling process plan that meets the inspection standards and is verified on site. For example, according to the ASTM E155 standard, the cooling process plan corresponding to the casting with a defect level not exceeding level 2 is selected for on-site production testing.

[0120] The newly acquired field verification data (including cooling process plan and casting quality) in step 1 are added to the database for training the GBDT model and dynamically updating the mold cooling process plan and control rules.

[0121] In step 1, the upper die, lower die and side die in the mold must all be equipped with thermocouples for measurement and arranged at different positions.

[0122] In step one, at least 90% of the data in the established database is used as a training set for model training, and the remaining data is used as a test set for evaluating the model prediction performance.

[0123] In step 3, the learning rate (learning_rate) of the gradient boosted decision tree (GBDT) model is between 0.01 and 0.3. The tree depth (max_depth) is between 3 and 10.

[0124] In step three, after the casting quality defect prediction model is established, when the accuracy of the model on the test set reaches or exceeds the set value, the model is used to predict the cooling process plan and conduct on-site production verification.

[0125] The feature importance ranking in step 2 is based on the criteria of different areas of the mold, and the thermocouples arranged at different positions of the upper mold, lower mold and side mold are ranked respectively.

[0126] Specific as Figure 1 As shown, the process is as follows:

[0127] Step 101: Determine the mold feature area.

[0128] Taking the low-pressure casting mold of a certain wheel hub as an example, among the 100 trial-produced wheel hubs, the production wheel hubs were subjected to X-ray inspection. Shrinkage defects were found in the outer rim, the junction between the spoke and the outer rim, and the wheel center. The corresponding mold location according to the wheel hub defect location is the mold feature area.

[0129] like Figure 2 As shown in Table 1, the thermocouples of the mold are arranged according to the above mold characteristic areas: ten thermocouples are arranged on the top mold, respectively at the mold ejector port, beside the T4 cooling channel, near the outer side of the inner rim of the top mold, inside the inner rim of the top mold, the edge of the top mold, on the inner side of the top mold rim, on the inner side of the top mold rim, on the outer side of the top mold rim, on the outer side of the top mold rim and the inner rim ejector port. The side mold consists of four side molds, marked with 1, 2, 3, and 4. Thermocouples are arranged on each side mold close to the inner rim, close to the outer rim and on the rim, for a total of 12 thermocouples, and one thermocouple is arranged on the inner and outer sides of the 1# side mold. Six thermocouples are arranged on the bottom mold, respectively beside the bottom mold B1, between B2 and B3, between B2 and B3, beside the bottom mold B3, between the bottom mold B4 and B5, and between the bottom mold B4 and B5. There are 16 cooling channels and two heating channels on the mold, among which B1, B2, B3, B4, and B5 cooling channels are arranged at the bottom mold. The cooling medium is air cooling, and the flow rate is set to 80m 3 / h, according to the number size, from B1 closest to the wheel center to B5 on the outside of the bottom mold; the top mold is arranged with T1, T2, T3, T4, T5, T6, T7 cooling channels, the cooling medium is air cooling, and the flow rate is set to 80m 3 / h, according to the number size, from T1 closest to the ejector port to T7 close to the outside of the top mold. Each of the four side molds is equipped with a cooling channel, the cooling medium is water cooling, the flow rate is set to 5L / mi, the numbers S1, S2, S3, S4 correspond to the side mold numbers, and the side mold cooling channel is arranged near the outer rim of the side mold. Two heating channels are arranged near the wheel core of the top mold, numbered H1 and H2, the heating resistance wire is nickel-chromium alloy, and the diameter of the electric heating wire resistance wire is 0.5mm.

[0130] Table 1. Thermocouple layout

[0131]

[0132]

[0133] Step 102: Establish a thermocouple weight coefficient model and optimize the quantity.

[0134] To conduct a sensitivity test on the casting, a standard cooling process was first selected as a benchmark, and the standard process was used for the first three hubs to ensure normal mold temperature and good casting quality. Subsequently, the last five hubs were tested, each time closing one cooling channel. The temperature data at the 1st and 100th seconds of the mold closing time were recorded, and these data were used as independent variables, and the hub quality was used as the dependent variable. The importance of each feature was analyzed through the random forest model in Python. Based on the analysis results, one key thermocouple was selected in each of the top mold, bottom mold, and side mold to reflect the casting quality and conduct subsequent analysis. Thermocouple temperature measurement points with an importance greater than 90% were selected for subsequent data analysis.

[0135] As shown in Table 2, Figure 3 As shown in Table 3,

[0136] Table 2. Specific parameters of random forest model

[0137] Random Forest Model Conditions parameter Number of trees 100 to 1000 The number of features considered by each tree 'sqrt' Minimum number of samples required for internal node repartitioning 2 to 10 The number of features to remove in each recursion 1 Split Quality 'gini' Minimum number of samples required for a leaf node 1-10 Metrics for evaluating model performance 'precision' .

[0138] Table 3. Codes and specific locations of thermocouple temperature measurement positions after optimization

[0139]

[0140]

[0141] Step 103: Establish cooling process, initial mold temperature and casting quality model

[0142] like Figure 4 As shown, the construction method of the gradient boosting decision tree model is demonstrated.

[0143] Collect process simulation and actual production trial data to establish a database, including cooling process data of 16 cooling channels, temperature data of 10 mold temperature measurement points, and mechanical properties of characteristic areas of corresponding castings. Each data record covers the entire production process of a product. The cooling process and mold temperature data are collected in steps of 1 second, and the state of the cooling channel is represented by 1 (on) or 0 (off). The database contains at least 100 data.

[0144] Based on the database established in step one, the gradient boosting decision tree (GBDT) method is used to establish a prediction model for the cooling process scheme. First, 90% of the data in the database is used for model training, and the remaining 10% of the data is used for performance evaluation. A gradient boosting decision tree model is established based on 10 thermocouple temperature measurement data, cooling process and casting quality. The initial model training settings include a learning rate of 0.1, a number of trees of 100, a maximum depth of 3, and GBDT is used for training. During the training process, we continuously adjust the learning rate, the number of trees, and the maximum depth until the average prediction accuracy of the model on the test set reaches or exceeds 90%. In this embodiment, the average prediction accuracy of the final model on the test set is 90.3%. Output the cooling process parameters at different mold temperatures according to the model. The process optimization is shown in Table 4: This process was used for trial production, and no scrap was generated during the 2-hour uninterrupted trial production. The temperature data of some wheel hub molds are as follows Figure 5 As shown, the temperature remains stable, and the difference between the maximum temperature and the minimum temperature at the same time does not exceed 5°C. Figure 6a and Figure 6b As shown in Table 5, the X-ray detection of the wheel core before and after the process optimization shows that there are certain shrinkage and shrinkage holes before the optimization. The mechanical properties of the characteristic area of ​​the hub before and after the process optimization are compared, and the performance of each position of the hub is improved as shown in Table 5. As shown in Table 6, the defects in the characteristic area of ​​the actual production castings are compared before and after the optimized process parameters are used. After the optimized process parameters are used, the defects in the characteristic area of ​​the hub are significantly improved.

[0145] Table 4. Process optimization table

[0146]

[0147] Table 5. Comparison of mechanical properties of the hub characteristic area before and after process optimization, performance of each position of the hub

[0148]

[0149]

[0150] Table 6. Comparison of characteristic regional defects in actual production castings before and after using optimized process parameters

[0151] Characteristic area defects Initial process example Optimization process example Spoke shrinkage many Very rare Rim shrinkage many Very rare Wheel core air hole More none Pinholes Exceeding Standard few none Insufficient watering few none .

[0152] The control method disclosed in this embodiment has the following advantages:

[0153] 1. The on-site production data is combined with simulation to determine the mold feature area according to the X-ray detection of wheel hub defects. Thermocouples are set in the mold feature area for temperature measurement according to the structural characteristics of the wheel hub. A random forest model is established based on the temperature measurement results and casting quality inspection, and recursive feature elimination is performed to determine the correlation between each thermocouple temperature measurement data and the wheel hub quality, optimize the number of thermocouples, and find the thermocouple temperature measurement positions corresponding to various casting defects.

[0154] 2. A gradient boosting decision tree model is established based on cooling process parameters, mold temperature, casting quality and characteristic area performance, and the casting quality is judged based on the cooling process and mold temperature. Compared with manual control, it is possible to find out whether the casting has defects during the production process and then adjust it, change the cooling parameters, control the mold temperature, and achieve automation and intelligence.

[0155] 3. Predict casting quality based on mold temperature range, abandon the time delay of traditional cooling process. When defects occur in castings, on-site workers change cooling parameters based on experience, which cannot improve mold temperature in a short time. Control according to mold temperature range. In each casting production process, the opening and closing time of cooling process can be determined according to thermocouple temperature, and controlled within a reasonable temperature range to ensure casting quality.

[0156] The basic principles of the present disclosure are described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, effects, etc. mentioned in the present disclosure are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. are required by each embodiment of the present disclosure. In addition, the specific details disclosed above are only for the purpose of illustration and ease of understanding, and are not limitations. The above details do not limit the present disclosure to the necessity of adopting the above specific details to be implemented.

[0157] In the present disclosure, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations. The block diagrams of the devices, devices, equipment, and systems involved in the present disclosure are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagram. As will be appreciated by those skilled in the art, these devices, devices, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including", "comprising", "having", etc. are open words, referring to "including but not limited to", and can be used interchangeably with them. The words "or" and "and" used here refer to the words "and / or" and can be used interchangeably with them, unless the context clearly indicates otherwise. The words "such as" used here refer to the phrase "such as but not limited to", and can be used interchangeably with them.

[0158] Additionally, as used herein, "or" used in a list of items beginning with "at least one" indicates a separate list, so that, for example, a list of "at least one of A, B, or C" means A or B or C, or AB or AC or BC, or ABC (i.e., A and B and C). Furthermore, the word "exemplary" does not mean that the example described is preferred or better than other examples.

[0159] It should also be noted that in the system and method of the present disclosure, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present disclosure.

[0160] Various changes, substitutions, and modifications of the techniques described herein may be made without departing from the teachings defined by the appended claims. Furthermore, the scope of the claims of the present disclosure is not limited to the specific aspects of the processes, machines, manufactures, compositions of events, means, methods, and actions described above. Currently existing or later to be developed processes, machines, manufactures, compositions of events, means, methods, or actions that perform substantially the same functions or achieve substantially the same results as the corresponding aspects described herein may be utilized. Thus, the appended claims include such processes, machines, manufactures, compositions of events, means, methods, or actions within their scope.

[0161] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present disclosure. Therefore, the present disclosure is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

[0162] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present disclosure to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.

Claims

1. A casting system temperature intelligent control method, characterized in that: include: Acquire the characteristic area of ​​the casting mold, and set a thermocouple in the characteristic area of ​​the mold based on the structural characteristics of the casting; Based on the thermocouple temperature measurement results and casting quality inspection results, a random forest model is established and recursive feature elimination is performed to determine the correlation between each thermocouple temperature measurement data and casting quality, thereby optimizing the number of thermocouples and obtaining the thermocouple temperature measurement positions corresponding to various casting defects, thereby screening the thermocouples. According to the screened temperature data of thermocouple measuring points, cooling process parameters and corresponding casting quality, the relationship between cooling process parameters, the starting temperature of each mold thermocouple and casting quality is constructed through the gradient boosting decision tree model, and the casting system temperature is controlled based on the constructed relationship.

2. The casting system temperature intelligent control method according to claim 1 is characterized in that: The step of obtaining the characteristic area of ​​the casting mold comprises: Collect historical process data of castings and casting molds, and determine the characteristic areas of casting molds based on X-ray detection and simulated casting shrinkage volume results of historical process data.

3. The casting system temperature intelligent control method according to claim 2 is characterized in that: The historical process data of the casting and the casting mold are collected, including: cooling process plan data and casting quality data; The cooling process plan data includes cooling channel configuration, switching time and / or coolant flow rate at each production stage; The casting quality data includes casting quality indicators, and the casting quality indicators include defect type, defect location, characteristic area yield strength and / or tensile strength.

4. The casting system temperature intelligent control method according to claim 1, characterized in that: The random forest model is established based on the thermocouple temperature measurement results and the casting quality inspection results, and recursive feature elimination is performed to determine the correlation between each thermocouple temperature measurement data and the casting quality, thereby optimizing the number of thermocouples, obtaining the thermocouple temperature measurement positions corresponding to various defects of the casting, and screening the thermocouples, including: A sensitivity test was conducted on the mold to select the temperature data that meets the mold closing time as the independent variable and the casting quality as the dependent variable for calculation. A random forest model was established and the importance of each feature was calculated. According to the feature importance, thermocouples that reflect the casting quality were selected as standard thermocouples in the top mold, bottom mold and side mold of the mold.

5. The casting system temperature intelligent control method according to claim 4, characterized in that: The method of establishing a random forest model based on the thermocouple temperature measurement results and the casting quality inspection results and performing recursive feature elimination includes: Step 1: Use the acquired features to train the random forest model and calculate the weight or coefficient of each feature of the random forest model; Step 2: Sort the features according to their weights or coefficients; Step 3: Delete one or more features with the smallest weight or coefficient in the ranking and retrain the random forest model with the remaining features; Step 4: Repeat steps 2 and 3 until the desired number of features is reached or no further features can be removed.

6. The casting system temperature intelligent control method according to claim 5, characterized in that: Thermocouple temperature characteristic importance calculations, including: Calculate the node Gini impurity; Calculate the contribution of each feature; Cumulative feature contribution; Calculate the feature importance of all trees in the random forest and average the feature importance values ​​across all trees to get the final importance of the feature.

7. The casting system temperature intelligent control method according to claim 1, characterized in that: The temperature data of the selected thermocouple temperature measurement points, cooling process parameters and corresponding casting quality are analyzed, and the relationship between cooling process parameters, the starting temperature of each mold thermocouple and casting quality is constructed through the gradient boosting decision tree model. The casting system temperature is controlled based on the constructed relationship, including: Obtain the selected temperature data of the thermocouple temperature measuring points, cooling process parameters and parameter data sets corresponding to the casting quality; Under the condition of ensuring the quality of castings, the relationship model between the initial temperature of the mold feature area and the cooling process is output based on the parameter data set to predict the performance strength of the wheel hub; The relationship model is optimized and iterated according to the difference between the casting quality and performance strength and the actual value; The relationship model is evaluated based on the mean square error and the coefficient of determination. When the prediction success rate is greater than the set value, the optimization iteration is stopped.

8. The casting system temperature intelligent control method according to claim 3, characterized in that: The cooling process scheme data and casting quality data are used to train the gradient boosting decision tree model and dynamically update the mold cooling process scheme and control rules; Use the gradient boosting decision tree model to predict cooling process solutions that meet the inspection standards.

9. The casting system temperature intelligent control method according to claim 8, characterized in that: The learning rate of the gradient boosting decision tree model is 0.01~0.

3.

10. The casting system temperature intelligent control method according to claim 9, characterized in that: The tree depth of the gradient boosting decision tree model is 3 to 10.

Images