A method and system for determining casting solidification process similarity
By using a multi-factor comprehensive evaluation and a three-layer artificial neural network model, the problem of low accuracy in determining the similarity of casting solidification processes was solved, and higher accuracy in similarity judgment was achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HARBIN INST OF TECH
- Filing Date
- 2023-09-11
- Publication Date
- 2026-05-12
AI Technical Summary
Existing methods for determining the similarity of casting solidification processes have low accuracy and cannot effectively assess the similarity between two solidification processes.
A multi-factor comprehensive evaluation method based on casting modulus, pouring temperature, sand mold temperature and heat transfer coefficient was adopted. An intelligent casting analysis model with a three-layer artificial neural network was used to determine the similarity of the casting solidification process through the Niyama criterion value, and casting solidification similarity analysis software was established.
It improves the accuracy of the similarity evaluation of the solidification process of castings, and can directly assess whether the solidification processes of two castings are similar, avoiding the judgment error caused by a single criterion and improving the accuracy of the judgment.
Smart Images
Figure CN117034039B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of casting solidification technology, and in particular to a method and system for determining the similarity of casting solidification processes. Background Technology
[0002] To analyze the formation patterns of defects in castings, it is necessary to start from the problematic areas of the solidified casting. The solidification process at these locations is then scaled down or enlarged and transferred to laboratory conditions to study the defect patterns during the solidification process. The results are then transferred back to the original casting to guide production. Sometimes, it is also necessary to transfer the casting technology obtained under laboratory conditions to actual production, either for a specific part or the entire casting. All of this hinges on the issue of solidification similarity. Poor solidification similarity makes it impossible to bridge the gap between two solidification phenomena. Therefore, the study of solidification similarity is of significant theoretical and practical value for exploring the evolutionary laws of two solidification processes and obtaining control methods for both.
[0003] Currently, temperature similarity criteria or flow similarity criteria are mainly used to determine the similarity between two solidification processes. At present, a single criterion is mainly used to determine the similarity of the solidification process of castings, such as solidification temperature gradient, solidification time, solid fraction or cooling rate. However, since there are many factors affecting the solidification process, using only a single criterion to determine the similarity of the solidification process of castings will lead to low accuracy. Summary of the Invention
[0004] The purpose of this invention is to solve the problem of low accuracy in existing methods for determining the similarity of casting solidification processes, and to propose a method and system for determining the similarity of casting solidification processes.
[0005] The specific process of a method for determining the similarity of casting solidification processes is as follows:
[0006] S1. Obtain the modulus, pouring temperature, sand mold temperature and heat transfer coefficient of the two castings to be evaluated for solidification similarity. Input the modulus, pouring temperature, sand mold temperature and heat transfer coefficient of the two castings to be evaluated for solidification similarity into the trained intelligent casting analysis model to obtain the Niyama criterion value of the two castings to be evaluated for solidification similarity.
[0007] S2. Using the niyama criterion value obtained in S1, obtain the solidification similarity conclusion of the two castings to be evaluated.
[0008] The trained intelligent casting analysis model is obtained through the following method:
[0009] Step 1: Establish a solidification model of the casting and simulate the solidification process of the casting to obtain the simulation results of the solidification process of the casting.
[0010] Step 2: Obtain the casting modulus, pouring temperature, sand mold temperature and heat transfer coefficient from the simulation results of the casting solidification process. Also, obtain the temperature gradient G and cooling rate L of points P1, P2 and P3 on the casting solidification model under the current casting modulus, pouring temperature, sand mold temperature and heat transfer coefficient, so as to obtain the Niyama criteria for points P1, P2 and P3.
[0011] Points P1, P2, and P3 in the solidification model of the casting are as follows:
[0012] Get a vertex P1 of the solidified model of the casting, and set the current vertex P1 as the origin of the spatial rectangular coordinate system, and set P2(0,L / 2,L / 2) and P3(L / 2,L / 2,L / 2);
[0013] Step 3: Combine the casting modulus, pouring temperature, sand mold temperature, heat transfer coefficient, and Niyama criteria for points P1, P2, and P3 obtained in Step 2 into a training set. Use the training set to train the intelligent casting analysis model. The trained intelligent casting analysis model is then obtained.
[0014] Furthermore, the Niyama criteria for points P1, P2, and P3 are obtained through the following formula:
[0015]
[0016] Where G is the temperature gradient and L is the cooling rate.
[0017] Furthermore, the intelligent casting analysis model is a three-layer artificial neural network, including: an input layer, a hidden layer, and an output layer;
[0018] The hidden layer consists of two layers.
[0019] Furthermore, the loss function of the intelligent casting analysis model is as follows:
[0020]
[0021] Where O is the target output value, a is the model prediction value, λ, θ i This is the regularization parameter.
[0022] Furthermore, in S2, using the Niyama criterion value obtained in S1, the solidification similarity conclusion of the two castings to be evaluated is obtained, specifically as follows:
[0023] If the Niyama criteria for points P1, P2, and P3 on two castings to be evaluated for solidification similarity are all equal, it indicates that the solidification processes of the two castings to be evaluated for solidification similarity are similar. If the Niyama criteria for any point are not equal, it indicates that the solidification processes of the two castings to be evaluated for solidification similarity are not similar.
[0024] The standard equality is defined as follows: the error between the two Niyama criteria is within the preset error range.
[0025] A casting solidification process similarity determination system is provided for use in a casting solidification process similarity determination method. The system includes: a casting parameter acquisition module, a Niyama criterion value acquisition module, and a solidification similarity determination module.
[0026] The casting parameter acquisition module is used to obtain the modulus, pouring temperature, sand mold temperature and heat transfer coefficient of the two castings to be evaluated for solidification similarity, and send the modulus, pouring temperature, sand mold temperature and heat transfer coefficient of the two castings to be evaluated for solidification similarity to the niyama criterion value acquisition module.
[0027] The Niyama criterion value acquisition module is used to input the modulus, pouring temperature, sand mold temperature and heat transfer coefficient of each casting whose solidification similarity is to be evaluated into the trained intelligent casting analysis model to obtain the Niyama criterion value of each casting whose solidification similarity is to be evaluated, and input the Niyama criterion value of each casting whose solidification similarity is to be evaluated into the solidification similarity determination module.
[0028] The solidification similarity determination module is used to obtain the solidification similarity conclusion of two castings to be evaluated by using the niyama score of each casting to be evaluated for solidification similarity.
[0029] Furthermore, the solidification similarity analysis software obtains the results through the following methods:
[0030] Step 1: Establish a solidification model of the casting and simulate the solidification process of the casting to obtain the simulation results of the solidification process of the casting.
[0031] Step 2: In the simulation results of the casting solidification process, obtain the casting modulus, pouring temperature, sand mold temperature and heat transfer coefficient, and obtain the temperature gradient G and cooling rate L of points P1, P2 and P3 on the casting solidification model under the current casting modulus, pouring temperature, sand mold temperature and heat transfer coefficient, so as to obtain the Niyama criterion value of points P1, P2 and P3.
[0032] Points P1, P2, and P3 in the solidification model of the casting are as follows:
[0033] Obtain a vertex P1 of the solidified casting model, and use the current vertex P1 as the origin to construct the x-axis and y-axis along the side length of the casting; set P2(0,L / 2,L / 2) and P3(L / 2,L / 2,L / 2);
[0034] The Niyama criterion value is obtained using the following formula:
[0035]
[0036] Step 3: Combine the casting modulus, pouring temperature, sand mold temperature, heat transfer coefficient, and Niyama criteria for points P1, P2, and P3 obtained in Step 2 into a training set. Use the training set to train the intelligent casting analysis model and obtain the trained intelligent casting analysis model.
[0037] Furthermore, the intelligent casting analysis model is a three-layer artificial neural network, including: an input layer, a hidden layer, and an output layer;
[0038] The hidden layer consists of two layers.
[0039] Furthermore, the loss function of the intelligent casting analysis model is as follows:
[0040]
[0041] Where O is the target output value, a is the model prediction value, λ, θ i This is the regularization parameter.
[0042] Furthermore, the step of using the Niyama score of each casting to be evaluated for solidification similarity to obtain the solidification similarity conclusion between the two castings to be evaluated is specifically as follows:
[0043] If the Niyama criteria for points P1, P2, and P3 on two castings to be evaluated for solidification similarity are all equal, it indicates that the solidification processes of the two castings to be evaluated for solidification similarity are similar. If the Niyama criteria for any point are not equal, it indicates that the solidification processes of the two castings to be evaluated for solidification similarity are not similar.
[0044] The standard equality is defined as follows: the error between the two Niyama criteria is within the preset error range.
[0045] The beneficial effects of this invention are as follows:
[0046] This invention proposes a method for evaluating the similarity of casting solidification processes. Based on the relationship between casting modulus, pouring temperature, sand mold temperature, heat transfer coefficient, and the Niyama criterion value, this invention establishes solidification similarity analysis software. This software can directly determine whether the solidification processes of two castings are similar without complex calculations. Furthermore, this invention considers multiple factors influencing the solidification process and integrates these factors into a single Niyama criterion value. This avoids the low accuracy problem caused by using only a single criterion to judge the similarity of casting solidification processes, thus improving the accuracy of casting solidification process similarity evaluation. Attached Figure Description
[0047] Figure 1 A three-dimensional model of the solidified casting;
[0048] Figure 2(a) shows the influence curves of the modulus on the temperature gradient, cooling rate, and Niyama criterion at point P1.
[0049] Figure 2(b) shows the influence curves of the modulus on the temperature gradient, cooling rate, and Niyama criterion at point P2.
[0050] Figure 2(c) shows the influence curves of the modulus on the temperature gradient, cooling rate, and Niyama criterion at point P3.
[0051] Figure 3(a) shows the influence curves of casting temperature on temperature gradient, cooling rate and Niyama criterion at point P1.
[0052] Figure 3(b) shows the influence curves of pouring temperature on the temperature gradient, cooling rate, and Niyama criterion at point P2.
[0053] Figure 3(c) shows the influence curves of pouring temperature on the temperature gradient, cooling rate, and Niyama criterion at point P3.
[0054] Figure 4(a) shows the influence curves of sand mold temperature on temperature gradient, cooling rate, and Niyama criterion at point P1.
[0055] Figure 4(b) shows the influence curves of sand mold temperature on the temperature gradient, cooling rate, and Niyama criterion at point P2.
[0056] Figure 4(c) shows the influence curves of sand mold temperature on the temperature gradient, cooling rate, and Niyama criterion at point P3.
[0057] Figure 5(a) shows the influence curves of heat transfer coefficient on temperature gradient, cooling rate and Niyama criterion at point P1;
[0058] Figure 5(b) shows the influence curves of heat transfer coefficient on temperature gradient, cooling rate and Niyama criterion at point P2;
[0059] Figure 5(c) shows the influence curves of heat transfer coefficient on temperature gradient, cooling rate and Niyama criterion at point P3;
[0060] Figure 6(a) is a comparison chart of the actual value of point P1 and the predicted value obtained by the analysis software;
[0061] Figure 6(b) is a comparison chart of the actual value of point P2 and the predicted value obtained by the analysis software;
[0062] Figure 6(c) is a comparison chart of the actual value of point P3 and the predicted value obtained by the analysis software;
[0063] Figure 7 Cross-sectional views of actual castings under different moduli. Detailed Implementation
[0064] Specific Implementation Method 1: The specific process of this implementation method for determining the similarity of casting solidification processes is as follows:
[0065] S1. Obtain the modulus, pouring temperature, sand mold temperature and heat transfer coefficient of the two castings to be evaluated for solidification similarity. Input the modulus, pouring temperature, sand mold temperature and heat transfer coefficient of each casting to be evaluated for solidification similarity into the trained intelligent casting analysis model to obtain the Niyama criterion value of each casting to be evaluated for solidification similarity.
[0066] S2. Using the niyama score of each casting to be evaluated for solidification similarity, obtain the solidification similarity conclusion between the two castings to be evaluated for solidification similarity.
[0067] If the Niyama criteria for solidification similarity of two castings are equal at all points, it indicates that the solidification processes of the two castings are similar. If the Niyama criteria for solidification similarity of any point are not equal, it indicates that the solidification processes of the two castings are not similar.
[0068] The meaning of standard equality is: the error between the two Niyama criteria is within 3%;
[0069] The trained intelligent casting analysis model is obtained through the following method:
[0070] Step 1: Establish a solidification model of the casting and simulate the solidification process of the casting to obtain the simulation results of the solidification process of the casting.
[0071] The solidification model of the casting is a cube with a modulus of M;
[0072] Step 2: In the simulation results of the casting solidification process, obtain the modulus M, pouring temperature, sand mold temperature and heat transfer coefficient, and obtain the temperature gradient G and cooling rate L of points P1, P2 and P3 on the casting solidification model under the current casting modulus, pouring temperature, sand mold temperature and heat transfer coefficient. Then, use the temperature gradient G and cooling rate L to obtain the Niyama criterion.
[0073] P1 is a vertex of the solidification model of the casting, and P1 is used as the origin of the spatial coordinate system. P2(0,L / 2,L / 2) and P3(L / 2,L / 2,L / 2) are set.
[0074] The Niyama criterion is as follows:
[0075] Step 3: Input the Niyama criteria of points P1, P2, and P3 on the solidification model of the casting after changing the modulus M, pouring temperature, sand mold temperature, and heat transfer coefficient obtained in Step 2 into the data-driven intelligent casting analysis model to train the intelligent casting analysis model and obtain the trained intelligent casting analysis model.
[0076] The intelligent casting analysis model is a three-layer artificial neural network (ANN), including: an input layer, a hidden layer, and an output layer;
[0077] If there is only one hidden layer, it is called a shallow neural network; if there are multiple hidden layers, it is called a deep neural network.
[0078] This invention employs two hidden layers;
[0079] In order to reduce variance and avoid overfitting, L2 regularization was used in the intelligent casting analysis model.
[0080] The loss function of the intelligent casting analysis model is:
[0081]
[0082] Where O is the target output value, a is the model prediction value, λ, θ i This is the regularization parameter.
[0083] Specific Implementation Method 2: A casting solidification process similarity determination system, used to execute a casting solidification process similarity determination method, the system includes: a casting parameter acquisition module, a Niyama criterion value acquisition module, and a solidification similarity determination module;
[0084] The casting parameter acquisition module is used to obtain the modulus, pouring temperature, sand mold temperature and heat transfer coefficient of the two castings to be evaluated for solidification similarity, and send the modulus, pouring temperature, sand mold temperature and heat transfer coefficient of the two castings to be evaluated for solidification similarity to the niyama criterion value acquisition module.
[0085] The Niyama criterion value acquisition module is used to input the modulus, pouring temperature, sand mold temperature and heat transfer coefficient of each casting whose solidification similarity is to be evaluated into the trained intelligent casting analysis model to obtain the Niyama criterion value of each casting whose solidification similarity is to be evaluated, and input the Niyama criterion value of each casting whose solidification similarity is to be evaluated into the solidification similarity determination module.
[0086] The solidification similarity determination module is used to obtain the solidification similarity conclusion between two castings to be evaluated by utilizing the Niyama score of each casting to be evaluated for solidification similarity:
[0087] If the Niyama criteria for points P1, P2, and P3 on two castings to be evaluated for solidification similarity are all equal, it indicates that the solidification processes of the two castings to be evaluated for solidification similarity are similar. If the Niyama criteria for any point are not equal, it indicates that the solidification processes of the two castings to be evaluated for solidification similarity are not similar.
[0088] The standard equality is defined as follows: the error between the two Niyama criteria is within the preset error range.
[0089] Specific Implementation Method Three: The trained intelligent casting analysis model is obtained through the following method:
[0090] Step 1: Establish a solidification model of the casting and simulate the solidification process of the casting to obtain the simulation results of the solidification process of the casting.
[0091] Step 2: In the simulation results of the casting solidification process, obtain the casting modulus, pouring temperature, sand mold temperature and heat transfer coefficient, and obtain the temperature gradient G and cooling rate L of points P1, P2 and P3 on the casting solidification model under the current casting modulus, pouring temperature, sand mold temperature and heat transfer coefficient, so as to obtain the Niyama criterion value of points P1, P2 and P3.
[0092] Points P1, P2, and P3 in the solidification model of the casting are as follows:
[0093] Get a vertex P1 of the solidified casting model, and set the current vertex P1 as the origin of the spatial coordinate system, and set P2(0,L / 2,L / 2) and P3(L / 2,L / 2,L / 2);
[0094] The Niyama criterion value is obtained using the following formula:
[0095]
[0096] Step 3: Combine the casting modulus, pouring temperature, sand mold temperature, heat transfer coefficient and Niyama criteria for points P1, P2 and P3 obtained in Step 2 into a training set. Use the training set to train the intelligent casting analysis model and obtain the trained intelligent casting analysis model.
[0097] The intelligent casting analysis model is a three-layer artificial neural network, including: an input layer, a hidden layer, and an output layer;
[0098] The hidden layer consists of two layers;
[0099] The loss function of the intelligent casting analysis model is as follows:
[0100]
[0101] Where O is the target output value, a is the model prediction value, λ, θ i This is the regularization parameter.
[0102] Example:
[0103] Step 1: Create an original model of the actual casting, taking a cube as an example, such as... Figure 1 As shown.
[0104] Step 2: Simulate the solidification process using ProCAST software. The material is ZL205A. By changing four factors, namely the casting modulus M, pouring temperature, sand mold temperature, and heat transfer coefficient, the simulation results of the solidification process are obtained. The specific scheme is shown in Table 1.
[0105] Table 1
[0106] Casting module M Pouring temperature Sand mold temperature heat transfer coefficient 1.67 690 25 300 5 700 50 400 8.33 710 100 500 11.67 720 200 600 15 730 300 700
[0107] Step 3, with Figure 1 Taking P1 as the origin, we select three points: P1(0,0,0), P2(0,L / 2,L / 2), and P3(L / 2,L / 2,L / 2). We obtain the effects of four factors on the temperature gradient G, cooling rate L, and [other factors] at these three points. The influence of values, the research results are as follows Figures 2(a)-5(c) As shown, it can be seen that, except for the heat transfer coefficient which is positively correlated with the temperature gradient G and the cooling rate L, the other three are negatively correlated with the temperature gradient G and the cooling rate L. Figures 2(a)-5(c) The vertical axis corresponding to the curve formed by solid circles is the second vertical axis from left to right, the vertical axis corresponding to the curve formed by solid squares is the first vertical axis from left to right, and the vertical axis corresponding to the curve formed by solid triangles is the third vertical axis from left to right.
[0108] Step 4: Use data-driven intelligent casting analysis software to learn from the simulation data. The input factors are the modulus, pouring temperature, sand mold temperature, and heat transfer coefficient. The Niyama criteria for P1, P2, and P3 are used. The value is output. Figures 6(a)-6(c) The chart comparing the actual values with the predicted values obtained by the analysis software shows that the predictions made by the analysis software are very accurate.
[0109] Step 5: Obtain relevant data for the actual casting. Taking a cube as an example, a casting with a module of 6.67mm (size 40mm) needs to be approximated to an experimental model with a module of 5mm (size 30mm). The known pouring temperature for a 40mm casting is 710℃, the sand mold temperature is 25℃, and the heat transfer coefficient is 300W / (m²). 2 ·K), Table 2 was obtained by using analysis software for prediction and ProCAST software for simulation.
[0110] Table 2
[0111] Modulus Pouring temperature Sand mold temperature heat transfer coefficient <![CDATA[P1]]> <![CDATA[P2]]> <![CDATA[P3]]> Predicted value 6.67 710 25 300 1.355 2.0139 7.772 True value 6.67 710 25 300 1.36 2.108 8.04
[0112] This invention compares the Niyama criterion values of corresponding points on two castings. To determine whether the solidification processes of two castings are similar, if the criteria for all corresponding points are equal, it indicates that the solidification processes of the two castings are similar. Now, for an experimental model with a modulus of 5, it is necessary to adjust the pouring temperature, sand mold temperature, and heat transfer coefficient to make the two solidification processes similar. Table 3 shows the results obtained through this adjustment.
[0113] Table 3 shows castings P1, P2, and P3 with a module of 5mm. Predicted values and simulated true values
[0114]
[0115] For the entire dataset, the average percentage error of the Niyama criterion values of the trained model is approximately 2.90%. These results indicate that the predicted and trained values are consistent with the parameter variations during the casting solidification process. The effectiveness of the model was verified through casting experiments. Figure 7 The cross-sectional views of the actual castings under different moduli show that there are a large number of shrinkage cavities and porosity inside each casting, and their distribution areas are the same.
Claims
1. A method for determining the similarity of the solidification process of castings, characterized in that... The specific process of the method is as follows: S1. Obtain the modulus, pouring temperature, sand mold temperature and heat transfer coefficient of the two castings to be evaluated for solidification similarity. Input the modulus, pouring temperature, sand mold temperature and heat transfer coefficient of the two castings to be evaluated for solidification similarity into the trained intelligent casting analysis model to obtain the Niyama criterion value of the two castings to be evaluated for solidification similarity. S2. Using the niyama criterion value obtained in S1, obtain the solidification similarity conclusion of the two castings to be evaluated. The trained intelligent casting analysis model is obtained through the following method: Step 1: Establish a solidification model of the casting and simulate the solidification process of the casting to obtain the simulation results of the solidification process of the casting. Step 2: Obtain the casting modulus, pouring temperature, sand mold temperature, and heat transfer coefficient from the simulation results of the casting solidification process. Also, obtain the casting solidification model under the current casting modulus, pouring temperature, sand mold temperature, and heat transfer coefficient. point, point, The temperature gradient G and cooling rate L at the point are used to obtain... point, point, Niyama criterion for points; In the solidification model of casting point, point, The following are the details: Obtain a vertex of the solidification model of the casting. and the current vertex As the origin of the spatial rectangular coordinate system, we set P2(0,L / 2,L / 2) and P3(L / 2,L / 2,L / 2); Step 3: Combine the casting modulus, pouring temperature, sand mold temperature, heat transfer coefficient, and other parameters obtained in Step 2. point, point, The Niyama criteria of points form a training set, which is used to train an intelligent casting analysis model. The trained intelligent casting analysis model is then obtained.
2. The method for determining the similarity of the solidification process of castings according to claim 1, characterized in that: The point, point, The Niyama criterion for a point is obtained through the following formula: Where G is the temperature gradient and L is the cooling rate.
3. The method for determining the similarity of the solidification process of castings according to claim 2, characterized in that: The intelligent casting analysis model is a three-layer artificial neural network, including: an input layer, a hidden layer, and an output layer; The hidden layer consists of two layers.
4. The method for determining the similarity of the solidification process of castings according to claim 3, characterized in that: The loss function of the intelligent casting analysis model is as follows: in, It is the target output value. These are model predictions. , i This is the regularization parameter.
5. The method for determining the similarity of the solidification process of a casting according to claim 4, characterized in that: The solidification similarity conclusion of the two castings to be evaluated is obtained by using the Niyama criterion value obtained in S1 in S2, specifically as follows: If two castings to be evaluated for solidification similarity have point, point, If the Niyama criteria at any point are equal, it indicates that the solidification processes of the two castings to be evaluated for solidification similarity are similar. If the Niyama criteria at any point are not equal, it indicates that the solidification processes of the two castings to be evaluated for solidification similarity are not similar. The standard equality is defined as follows: the error between the two Niyama criteria is within the preset error range.
6. A system for determining the similarity of a casting solidification process, used to execute a method for determining the similarity of a casting solidification process as described in any one of claims 1-5, characterized in that: The system includes: a casting parameter acquisition module, a Niyama criterion value acquisition module, and a solidification similarity determination module; The casting parameter acquisition module is used to obtain the modulus, pouring temperature, sand mold temperature and heat transfer coefficient of the two castings to be evaluated for solidification similarity, and send the modulus, pouring temperature, sand mold temperature and heat transfer coefficient of the two castings to be evaluated for solidification similarity to the niyama criterion value acquisition module. The Niyama criterion value acquisition module is used to input the modulus, pouring temperature, sand mold temperature and heat transfer coefficient of each casting whose solidification similarity is to be evaluated into the trained intelligent casting analysis model to obtain the Niyama criterion value of each casting whose solidification similarity is to be evaluated, and input the Niyama criterion value of each casting whose solidification similarity is to be evaluated into the solidification similarity determination module. The solidification similarity determination module is used to obtain the solidification similarity conclusion of two castings to be evaluated by using the niyama score of each casting to be evaluated for solidification similarity.
7. The casting solidification process similarity determination system according to claim 6, characterized in that: The solidification similarity analysis software was obtained through the following methods: Step 1: Establish a solidification model of the casting and simulate the solidification process of the casting to obtain the simulation results of the solidification process of the casting. Step 2: Calculate the casting modulus, pouring temperature, sand mold temperature, and heat transfer coefficient from the simulation results of the casting solidification process. Then, obtain the casting solidification model under the current casting modulus, pouring temperature, sand mold temperature, and heat transfer coefficient. point, point, The temperature gradient G and cooling rate L at the point are used to obtain... point, point, The Niyama criterion value of a point; In the solidification model of casting point, point, The following are the details: Obtain a vertex of the solidification model of the casting. and the current vertex As the origin, construct the x-axis and y-axis along the side length of the casting; set P2(0,L / 2,L / 2) and P3(L / 2,L / 2,L / 2); The Niyama criterion value is obtained using the following formula: Step 3, combine the casting modulus, pouring temperature, sand mold temperature, heat transfer coefficient and... obtained in Step 2 point, point, The Niyama criteria of points form a training set, which is then used to train the intelligent casting analysis model, resulting in a well-trained intelligent casting analysis model.
8. The casting solidification process similarity determination system according to claim 7, characterized in that: The intelligent casting analysis model is a three-layer artificial neural network, including: an input layer, a hidden layer, and an output layer; The hidden layer consists of two layers.
9. A system for determining the similarity of casting solidification processes according to claim 8, characterized in that: The loss function of the intelligent casting analysis model is as follows: in, It is the target output value. These are model predictions. , This is the regularization parameter.
10. A system for determining the similarity of casting solidification processes according to claim 9, characterized in that: The method of using the Niyama score of each casting to be evaluated for solidification similarity to obtain the solidification similarity conclusion between the two castings to be evaluated is as follows: If two castings to be evaluated for solidification similarity have point, point, If the Niyama criteria at any point are equal, it indicates that the solidification processes of the two castings to be evaluated for solidification similarity are similar. If the Niyama criteria at any point are not equal, it indicates that the solidification processes of the two castings to be evaluated for solidification similarity are not similar. The standard equality is defined as follows: the error between the two Niyama criteria is within the preset error range.