DTW-GaussianNB casting quality prediction method based on mold temperature

Abstract

A DTW-GaussianNB casting quality prediction method based on mold temperature belongs to the field of casting manufacturing technology, including: collecting the temperature of key positions of the casting process mold and obtaining internal defect quality data; defining the process mold temperature corresponding to the good casting as the first mold temperature, and defining the process mold temperature of the casting to be predicted as the second mold temperature; using the Z-Score normalization method to normalize the temperature of each temperature measuring point; calculating the normalized temperature difference value with the DTW algorithm; defining the casting without / with internal defects of the casting as 0 and 1 encoding as output, using the normalized mold temperature difference value as input, establishing and training the GaussianNB model; substituting the training set data into the model training, and evaluating the model accuracy with the test set data. The present invention can efficiently and quickly predict the casting quality by establishing a model between the process mold temperature and the casting quality, and is more synchronous than traditional X-ray detection, improving the timeliness of process parameter adjustment in the casting process.

Description

Technical Field

[0001] The invention relates to the technical field of casting manufacturing, and in particular to a DTW-GaussianNB (Dynamic Time Warping-Gaussian Naive Bayes) casting quality prediction method based on mold temperature. Background Art

[0002] The need for energy conservation and emission reduction in the automotive industry is becoming more urgent. Aluminum automotive parts, as an important part of automotive lightweight technology, have now surpassed iron parts and become key parts for mainstream automotive configurations. Aluminum parts are mainly formed by casting, which is prone to defects during the casting process, so the casting process needs to be constantly adjusted.

[0003] At present, the industry mainly uses X-rays to detect internal defects in castings and adjusts the casting process based on the test results. Obviously, the traditional casting process adjustment route has the problems of lag, poor timeliness of process adjustment, and large equipment demand.

[0004] In addition, as a physical quantity throughout the key casting processes such as aluminum liquid filling and solidification, the mold temperature can directly affect the casting quality, and the mold temperature can be used as a preferred parameter for determining the quality of the casting. However, due to different casting processes, the production time of different castings is different, so it is impossible to directly compare the mold temperature data of different castings one by one, and there is no effective method to establish the relationship between mold temperature and casting quality. Therefore, seeking a method for predicting casting quality based on mold temperature has become one of the technical problems to be solved in this field.

[0005] Therefore, in view of the problems existing in the prior art, the designers of this case actively studied and improved it, and thus came up with the DTW-GaussianNB casting quality prediction method based on mold temperature of the present invention. Summary of the invention

[0006] The purpose of the present invention is to provide a DTW-GaussianNB casting quality prediction method based on mold temperature to address the defects in the prior art, such as the different production times of different castings due to different casting processes, the inability to directly compare the mold temperature data of different castings one by one, and there is no effective method to establish the relationship between mold temperature and casting quality, and the casting quality cannot be predicted by mold temperature.

[0007] To achieve the purpose of the present invention, the present invention provides a DTW-GaussianNB casting quality prediction method based on mold temperature. The DTW-GaussianNB casting quality prediction method based on mold temperature comprises:

[0008] Execute step S1: collect the temperature of key positions of the mold during the casting process through sensors, and obtain the quality data of internal defects of the castings through X-ray detection equipment;

[0009] Execute step S2: based on the casting internal defect quality data obtained by the X-ray detection equipment, select a defect-free good casting as the reference casting, use the temperature of each temperature measuring point of the process mold corresponding to the good casting as the reference mold temperature, and define it as the first mold temperature, and define the temperature of the key position of the process mold of the casting to be predicted as the second mold temperature;

[0010] Execute step S3: use the Z-Score normalization method to normalize the temperature of each temperature measuring point of the process mold of the reference casting and the temperature of each temperature measuring point of the process mold of the casting to be predicted;

[0011] Executing step S4: calculating the difference between the normalized second mold temperature and the first mold temperature using a DTW algorithm;

[0012] Execute step S5: define a casting without internal defects as 0 code, define a casting with internal defects as 1 code, use 0 code and 1 code as output, use the normalized difference between the second mold temperature and the first mold temperature as input, and establish and train a GaussianNB model;

[0013] Execute step S6: divide the test data into a training set and a test set, substitute the training set data into the GaussianNB model for training, and evaluate the model accuracy with the test set data after the training is completed, so as to obtain the DTW-GaussianNB casting quality prediction model based on the mold temperature, thereby realizing the casting quality prediction.

[0014] Optionally, in step S1, collecting the temperature of a key position of the mold during the casting process by using a sensor is collecting the temperature of the key position of the mold during the casting process by using a thermocouple.

[0015] Optionally, the process mold has six temperature measurement positions at key positions, namely the top mold flange position, the middle position of the top mold spoke, the side mold inner rim position, the side mold outer rim position, the bottom mold riser position, and the bottom mold spoke and rim transition position.

[0016] Optionally, the normalization equation is:

[0017] ,

[0018] in, is the normalized data, is the data before normalization, For the Unnormalized data, is the mean of the unnormalized data.

[0019] Optionally, the test data is divided into a training set and a test set at 80%:20%.

[0020] In summary, the DTW-GaussianNB casting quality prediction method based on mold temperature of the present invention can efficiently and quickly predict the casting quality by establishing a GaussianNB model between the mold temperature and the casting quality during the casting process. It is more synchronous than traditional X-ray inspection equipment in obtaining quality data on internal defects of castings, and improves the timeliness of adjustment of process parameters during the casting process. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 Shown is a flow chart of the DTW-GaussianNB casting quality prediction method based on mold temperature of the present invention;

[0022] Figure 2 The figure shows the execution flow chart of the DTW-GaussianNB model structure of the present invention;

[0023] Figure 3 The figure shows the comparison result between the predicted value and the measured value of the quality prediction using the DTW-GaussianNB casting quality prediction method based on the mold temperature of the present invention. DETAILED DESCRIPTION

[0024] In order to explain the technical content, structural features, objectives and effects of the present invention in detail, the following will be described in detail with reference to embodiments and accompanying drawings.

[0025] See also Figure 1 , Figure 2 , Figure 1 Shown is a flow chart of the DTW-GaussianNB casting quality prediction method based on mold temperature of the present invention. Figure 2 The DTW-GaussianNB model structure execution flow chart of the present invention is shown. The DTW-GaussianNB casting quality prediction method based on mold temperature includes:

[0026] Execute step S1: collect the temperature of key positions of the mold during the casting process through sensors, and obtain the quality data of internal defects of the castings through X-ray detection equipment;

[0027] Execute step S2: based on the casting internal defect quality data obtained by the X-ray detection equipment, select a defect-free good casting as the reference casting, use the temperature of each temperature measuring point of the process mold corresponding to the good casting as the reference mold temperature, and define it as the first mold temperature, and define the temperature of the key position of the process mold of the casting to be predicted as the second mold temperature;

[0028] Execute step S3: Use the Z-Score normalization method to normalize the temperature of each temperature measuring point of the process mold of the reference casting and the temperature of each temperature measuring point of the process mold of the casting to be predicted. The normalization equation is:

[0029] ,

[0030] in, is the normalized data, is the data before normalization, For the Unnormalized data, is the mean of the data when it is not normalized;

[0031] Executing step S4: calculating the difference between the normalized second mold temperature and the first mold temperature using a DTW algorithm;

[0032] Execute step S5: define a casting without internal defects as 0 code, define a casting with internal defects as 1 code, use 0 code and 1 code as output, use the normalized difference between the second mold temperature and the first mold temperature as input, and establish and train a GaussianNB model;

[0033] Execute step S6: divide the test data into training set and test set according to 80%:20%, substitute the training set data into the GaussianNB model for training, and evaluate the model accuracy with the test set data after the training is completed, so as to obtain the DTW-GaussianNB casting quality prediction model based on the mold temperature, thereby realizing the casting quality prediction.

[0034] In the present invention, as a specific implementation, the temperature of the key position of the casting process mold collected by the sensor in step S1 is collected by thermocouples. More specifically, the key position of the process mold has six temperature measurement positions, namely, the top mold flange position, the middle position of the top mold spoke, the side mold inner rim position, the side mold outer rim position, the bottom mold riser position, and the bottom mold spoke and rim transition position.

[0035] See also Figure 3 , Figure 3 The figure shows the comparison result of the predicted value and the measured value of the quality prediction using the DTW-GaussianNB casting quality prediction method based on the mold temperature of the present invention. Figure 3 It can be seen that the quality prediction using the DTW-GaussianNB casting quality prediction method based on mold temperature of the present invention fully meets the prediction error requirements and reduces the burden of obtaining casting internal defect quality data through X-ray detection equipment.

[0036] In summary, the DTW-GaussianNB casting quality prediction method based on mold temperature of the present invention can efficiently and quickly predict the casting quality by establishing a GaussianNB model between the mold temperature and the casting quality during the casting process. It is more synchronous than traditional X-ray inspection equipment in obtaining quality data on internal defects of castings, and improves the timeliness of adjustment of process parameters during the casting process.

[0037] Those skilled in the art will appreciate that various modifications and variations may be made to the present invention without departing from the spirit or scope of the present invention. Therefore, if any modification or variation falls within the scope of protection of the appended claims and equivalents, it is considered that the present invention covers these modifications and variations.

Claims

1. A DTW-GaussianNB casting quality prediction method based on mold temperature, characterized in that: The DTW-GaussianNB casting quality prediction method based on mold temperature includes: Execute step S1: collect the temperature of key positions of the mold during the casting process through sensors, and obtain the quality data of internal defects of the castings through X-ray detection equipment; Execute step S2: based on the casting internal defect quality data obtained by the X-ray detection equipment, select a defect-free good casting as the reference casting, use the temperature of each temperature measuring point of the process mold corresponding to the good casting as the reference mold temperature, and define it as the first mold temperature, and define the temperature of the key position of the process mold of the casting to be predicted as the second mold temperature; Execute step S3: use the Z-Score normalization method to normalize the temperature of each temperature measuring point of the process mold of the reference casting and the temperature of each temperature measuring point of the process mold of the casting to be predicted; Executing step S4: calculating the difference between the normalized second mold temperature and the first mold temperature using a DTW algorithm; Execute step S5: define a casting without internal defects as 0 code, define a casting with internal defects as 1 code, use 0 code and 1 code as output, use the normalized difference between the second mold temperature and the first mold temperature as input, and establish and train a GaussianNB model; Execute step S6: divide the test data into a training set and a test set, substitute the training set data into the GaussianNB model for training, and evaluate the model accuracy with the test set data after the training is completed, so as to obtain the DTW-GaussianNB casting quality prediction model based on the mold temperature, thereby realizing the casting quality prediction.

2. The DTW-GaussianNB casting quality prediction method based on mold temperature as claimed in claim 1, characterized in that: In step S1, collecting the temperature of the key position of the mold in the casting process by using the sensor is collecting the temperature of the key position of the mold in the casting process by using a thermocouple.

3. The DTW-GaussianNB casting quality prediction method based on mold temperature as claimed in claim 2, characterized in that: There are six temperature measuring positions at the key positions of the process mold, namely the top mold flange position, the middle position of the top mold spoke, the inner rim position of the side mold, the outer rim position of the side mold, the bottom mold riser position, and the transition position between the bottom mold spoke and the rim.

4. The DTW-GaussianNB casting quality prediction method based on mold temperature as claimed in claim 1, characterized in that: The normalized processing equation is: ， in, is the normalized data, is the data before normalization, For the Unnormalized data, is the mean of the unnormalized data.

5. The DTW-GaussianNB casting quality prediction method based on mold temperature as claimed in claim 1, characterized in that: The test data is divided into training set and test set according to 80%:20%.

Images