Ceramic core material physical property prediction and material optimization method based on neural network

Abstract

The invention provides a ceramic core material physical property prediction and material optimization method based on a neural network. The method comprises the following steps: constructing a neural network optimization prediction model, and training the neural network optimization prediction model until the training reaches the standard; the raw material components of the ceramic core are determined, the raw material components comprise the adding amount of a mineralizer, the raw material components are input into the trained neural network optimization prediction model, and a plurality of corresponding performance parameters are obtained; inputting the performance parameters into a comprehensive performance evaluation model to obtain comprehensive scores of the components of the raw materials; and comparing the comprehensive scores of the components of the multiple groups of raw materials, and screening the component of the raw material with the highest comprehensive score as the optimal material ratio, thereby determining the addition amount of the mineralizer with the optimal comprehensive performance. According to the method, the physical performance of the ceramic core material is predicted by using the neural network, and the optimal addition amount of the mineralizer can be determined in an auxiliary manner.

Description

Technical Field

[0001] This application relates to the field of materials technology. Specifically, it relates to a method for predicting the physical properties of a silicon-based ceramic core material based on a neural network and optimizing the material. Background Art

[0002] Ceramic cores play a crucial role in many industries such as aerospace and automotive, especially when manufacturing complex and delicate structures. Depending on the matrix material, aluminum-based, silicon-based, and magnesium-based ceramic cores are widely used. Aluminum-based ceramic cores maintain structural stability during sintering and use and have good high-temperature resistance, but it is difficult to remove the core during the manufacture of single-crystal blades; the manufacturing process of magnesium-based ceramic cores is simple and the chemical structure is stable, but its magnesium oxide impurities affect the hydration performance and core removal performance; while silicon-based ceramic cores have the advantages of stable high-temperature physical and chemical properties, low thermal expansion coefficient, and high purity, overcoming the disadvantages of the other two materials, so they are widely used.

[0003] The advantages of silicon-based ceramic cores are also reflected in their better mechanical strength and corrosion resistance, enabling them to maintain extremely high stability in complex high-temperature environments. Its low thermal expansion coefficient can not only effectively reduce the deformation caused by temperature changes during the manufacturing process, but also reduce the influence of thermal stress on components under more severe high-temperature conditions, thereby improving the accuracy and reliability of the final product. At the same time, the high purity of the silicon-based ceramic core material makes it have better chemical stability at high temperatures, able to resist the erosion of chemical reactions such as oxidation and sulfidation, thus extending the service life of the core.

[0004] With the continuous development of manufacturing processes, the application fields of silicon-based ceramic core materials are also constantly expanding. Especially in the manufacture of advanced turbine blades, silicon-based ceramic cores can support higher casting precision and more complex geometric shape requirements, providing guarantee for manufacturing more efficient and high-performance engine components.

[0005] The superior performance of silicon-based ceramic cores not only makes them perform excellently in the manufacture of turbine blades, but also shows great potential in other high-temperature applications. Especially in high-temperature and high-pressure environments, silicon-based ceramic cores can withstand extreme working conditions and ensure the accuracy of the casting process. With the increasing requirements of turbine engines and high-performance combustion equipment, the stability, strength, and heat resistance of silicon-based ceramic cores make them an ideal material for key components in aeroengines, gas turbines, and the energy field.

[0006] In addition, compared with traditional materials, the production process of silicon-based ceramic cores is more environmentally friendly. Their high purity and excellent chemical stability result in less generation of harmful gases and substances at high temperatures, meeting the high standards of environmental protection in modern industry. It can also reduce waste and energy consumption, having unique advantages in green manufacturing and sustainable development.

[0007] The traditional experimental method can only find the optimal solution among the few experimental variables given, and the result depends to a large extent on the level values selected artificially. Therefore, this method has great limitations and inaccuracies. At the same time, the traditional method of making artificial judgments based on performance parameters lacks certain judgment criteria, and its accuracy and efficiency are both low. Summary of the Invention

[0008] To solve the technical problems existing in the above background art, the present application provides a method for predicting the physical properties of a ceramic core material based on a neural network and a material optimization method, an electronic device, a computer storage medium, and a computer program product, so as to predict the composition of the optimal performance under all parameters with only a few experimental parameters.

[0009] The present application provides a method for predicting the physical properties of a ceramic core material based on a neural network and a material optimization method, the method comprising:

[0010] Construct a neural network optimization prediction model and train the neural network optimization prediction model until the training reaches the standard;

[0011] Determine the raw material composition of the ceramic core, the raw material composition including the addition amount of a mineralizer, and input the raw material composition into the trained neural network optimization prediction model to obtain corresponding performance parameters;

[0012] Input the performance parameters into a comprehensive performance evaluation model to obtain the comprehensive score of the raw material composition;

[0013] Compare the comprehensive scores of multiple groups of the raw material compositions, and select the raw material composition with the highest comprehensive score as the optimal material ratio, that is, determine the addition amount of the mineralizer with the best comprehensive performance.

[0014] Optionally, the performance parameters include shrinkage rate, porosity, room temperature bending strength, and high temperature bending strength.

[0015] Optionally, the comprehensive performance evaluation model calculates the scores of each performance parameter using a constraint equation to obtain the comprehensive score of the raw material composition, and the constraint equation is:

[0016] Score=W1*f A (x1)+W2*f B (x2)+W3*f C (x3)+W4*f D (x4)

[0017] In the formula, Score is the comprehensive score, f A (x1), fB (x2), f C (x3), f D (x4) are the scores of each of the said performance parameters, and W1, W2, W3, and W4 are the weight coefficients respectively.

[0018] Optionally, the calculation method of the scores of each of the said performance parameters is specifically as follows:

[0019] Calculation method of the score of shrinkage rate:

[0020]

[0021] Calculation method of the score of porosity:

[0022]

[0023] The calculation method of the score of room temperature bending strength is:

[0024]

[0025] Calculation method of the score of high temperature bending strength:

[0026]

[0027] In the formula, the k value is the change trend factor for adjusting the decision function, and x1, x2, x3, and x4 are the shrinkage rate, porosity, room temperature bending strength, and high temperature bending strength predicted and output by the said neural network optimization prediction model respectively.

[0028] Optionally, the weight coefficients of each of the said performance parameters are specifically as follows:

[0029] The weight coefficient W1 of the shrinkage rate is 0.30, the weight coefficient of the porosity W2 is 0.15, the weight coefficient W3 of the room temperature bending strength is 0.30, and the weight coefficient W4 of the high temperature bending strength is 0.25.

[0030] Optionally, in the raw material composition of the ceramic core, high-purity quartz glass powder is used as the matrix material, zirconium silicate is used as the additive, and the mineralizer is fused alumina.

[0031] Optionally, training the said neural network optimization prediction model until the training reaches the standard includes:

[0032] Prepare multiple groups of ceramic core materials, with different addition amounts of the mineralizer in each group of ceramic core materials, detect each of the said performance parameters of each group of ceramic core materials, and construct the addition amount of the mineralizer and each of the said performance parameters into a piece of training data;

[0033] Divide all the training data into a training set and a validation set, use the training set to train the neural network optimization prediction model, and use the validation set to conduct a validation test on the neural network optimization prediction model that has been trained to a certain extent;

[0034] If the result of the validation test meets the standard, it is determined that the training is qualified and the training ends; otherwise, continue to use the remaining training data in the training set to continue the training.

[0035] This application also provides an electronic device, which includes: a memory storing executable program code; a processor coupled to the memory; the processor calls the executable program code stored in the memory and executes the method as described in any one of the previous items.

[0036] This application also provides a computer storage medium, which stores a computer program, and the computer program is executed by a processor to implement the method as described in any one of the previous items.

[0037] This application also provides a computer program product, including a computer program stored on a non-transitory computer-readable medium, and the computer program implements the method as described in any one of the previous items when executed by a processor.

[0038] The beneficial effects of the present invention are as follows:

[0039] The present invention uses a neural network to realize the prediction of the physical properties of ceramic core materials and can assist in determining the optimal addition amount of mineralizer. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings required in the embodiments. It should be understood that the following drawings only show some embodiments of the present application and should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0041] Figure 1 It is a schematic flowchart of a method for predicting the physical properties of ceramic core materials and material optimization based on a neural network disclosed in an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0042] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. Usually, the components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.

[0043] Accordingly, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the claimed present application, but merely represents selected embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts shall fall within the scope of protection of the present application.

[0044] It should be noted that like reference numerals and letters denote like items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0045] In the description of the present application, it should be noted that if terms such as "upper", "lower", "inner", "outer", etc. are used to indicate the orientation or positional relationship, it is based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship in which the inventive product is habitually placed during use. It is only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation to the present application.

[0046] In addition, terms such as "first", "second", etc. are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance. It should be noted that the features in the embodiments of the present application can be combined with each other without conflict.

[0047] Please refer to Figure 1 , the embodiments of the present application disclose a method for predicting the physical properties and optimizing the material of a neural network-based ceramic core material, and the method includes:

[0048] Construct a neural network optimization prediction model and train the neural network optimization prediction model until the training reaches the standard;

[0049] Determine the raw material composition of the ceramic core, where the raw material composition includes the addition amount of a mineralizer, and input the raw material composition into the trained neural network optimization prediction model to obtain corresponding performance parameters;

[0050] Input the performance parameters into a comprehensive performance evaluation model to obtain the comprehensive score of the raw material composition;

[0051] Compare the comprehensive scores of multiple groups of the raw material compositions, and select the raw material composition with the highest comprehensive score as the optimal material ratio, that is, determine the addition amount of the mineralizer with the best comprehensive performance.

[0052] Optionally, the performance parameters include shrinkage rate, porosity, room temperature bending strength, and high temperature bending strength.

[0053] In the embodiments of the present application, 1) the shrinkage rate is an important indicator to measure the volume change of the ceramic core during sintering, which directly affects the matching degree between the core and the casting. Since the ceramic core will experience different degrees of volume shrinkage during sintering, if the shrinkage rate is too high, it will cause a mismatch between the core and the casting, forming a gap and thus affecting the accuracy of the core and the casting during the casting process.

[0054] 2) The porosity directly relates to the mechanical properties of the core. If the porosity is too high, the internal pore structure of the core is uneven, resulting in a decline in the mechanical properties of the ceramic core. A higher porosity may cause the core to crack or deform easily under high-temperature or high-pressure environments, affecting the quality of the casting.

[0055] 3) The room-temperature flexural strength refers to the ability of the ceramic core to resist bending stress under room-temperature conditions and is one of the key indicators for evaluating the mechanical properties of the ceramic core material, which affects the reliability of the core during the casting process. The higher the room-temperature flexural strength, the less likely the ceramic core is to break during processing and the greater the mechanical stress it can withstand without breaking.

[0056] 4) The high-temperature flexural strength refers to the bending strength of the ceramic core in a high-temperature environment, which determines whether the ceramic core can maintain structural stability and mechanical strength during high-temperature casting. As the temperature rises, the mechanical properties of the ceramic core material usually decline. The excellent performance of the high-temperature flexural strength means that the ceramic core can still maintain high strength and stability under extreme working conditions. For high-performance ceramic cores in industries such as aerospace and automotive, especially in high-temperature casting, it is necessary to ensure that they have sufficient high-temperature flexural strength to guarantee the reliability of the core during high-temperature operations.

[0057] Optionally, the comprehensive performance evaluation model calculates the scores of each performance parameter using a constraint equation to obtain the comprehensive score of the raw material composition. The constraint equation is:

[0058] Score = W1*f A (x1) + W2*f B (x2) + W3*f C (x3) + W4*f D (x4)

[0059] In the formula, Score is the comprehensive score, and f A (x1), f B (x2), f C (x3), f D (x4) are the scores of each performance parameter respectively, and W1, W2, W3, and W4 are the weight coefficients respectively.

[0060] Optionally, the calculation method of the score of each of the performance parameters is specifically as follows:

[0061] Calculation method of the score of shrinkage rate:

[0062]

[0063] Calculation method of the score of porosity:

[0064]

[0065] The calculation method of the score of room temperature bending strength is:

[0066]

[0067] Calculation method of the score of high temperature bending strength:

[0068]

[0069] In the formula, the k value is a change trend factor for adjusting the decision function, and x1, x2, x3, and x4 are the shrinkage rate, porosity, room temperature bending strength, and high temperature bending strength predicted and output by the neural network optimization prediction model respectively.

[0070] Optionally, the weight coefficients of each of the performance parameters are specifically as follows:

[0071] The weight coefficient W1 of shrinkage rate is 0.30, the weight coefficient of porosity W2 is 0.15, the weight coefficient W3 of room temperature bending strength is 0.30, and the weight coefficient W4 of high temperature bending strength is 0.25.

[0072] Optionally, in the raw material composition of the ceramic core, high-purity quartz glass powder is used as the matrix material, zirconium silicate is used as the additive, and the mineralizer is fused alumina.

[0073] Quartz glass powder has extremely high purity, which can ensure the stable performance of the material. In order to optimize the characteristics of the ceramic core, zirconium silicate is added as an additive to enhance specific properties of the material. Fused alumina is selected as the mineralizer and is divided into multiple groups according to different experimental requirements. In the experiment, the influence of the addition amount of fused alumina on the material performance is explored by adjusting the addition amount of fused alumina.

[0074] Optionally, training the neural network optimization prediction model until the training reaches the standard includes:

[0075] Prepare multiple groups of ceramic core materials with different addition amounts of the mineralizer in each group, detect each of the performance parameters of each group of ceramic core materials, and construct the addition amount of the mineralizer and each of the performance parameters into a piece of training data;

[0076] Divide all the training data into a training set and a validation set, use the training set to train the neural network optimization prediction model, and use the validation set to conduct a validation test on the neural network optimization prediction model that has been trained to a certain extent;

[0077] If the validation test result meets the standard, it is determined that the training is qualified and the training ends; otherwise, continue to use the remaining training data in the training set to continue training.

[0078] In this embodiment, multiple groups of ceramic core materials added with different amounts of mineralizer are prepared in advance and fired to obtain the ceramic core materials. Each performance parameter of the ceramic core materials with different fused corundum addition amounts is detected, including shrinkage rate, porosity, room temperature bending strength, and high temperature bending strength. Each group of ceramic core materials is fired multiple times, and the corresponding performance parameter detections are carried out multiple times. These measured data constitute the training data, and the training data is divided into a training set and a validation set according to 80% and 20% respectively.

[0079] Then, use the training set to train the neural network optimization prediction model. When training reaches a certain extent, for example, when 50% of the training data in the training set has been input, use the validation set to conduct a performance test on the neural network optimization prediction model. If the test meets the standard, stop training; otherwise, continue training and use the validation set for testing again.

[0080] It should be noted that the input layer parameters of the neural network optimization prediction model are the composition components of the ceramic core raw materials added with mineralizer. The number of input layer neurons is 3. The output layer parameters are the key performance parameters of the ceramic core made under the same process conditions, which are shrinkage rate, porosity, room temperature strength, and high temperature strength respectively. The number of neurons in the output layer is 4.

[0081] In the structure of the neural network, each layer contains one or more neurons, and each neuron has an activation function. Information is first received in the input layer, then passes through the hidden layer, and finally is transmitted to the output layer. During this process, the connection weights between layers are adjusted and trained according to the input data. After the forward propagation is completed, the network executes the backpropagation algorithm based on the error between the actual output and the expected output. The Sigmoid function is used as the activation function, and the mean square error is used as the loss function during the optimization process of the network.

[0082] The above process can be implemented based on the Python language and the TensorFlow component. The following is the key code:

[0083] w1 = tf.Variable(tf.random.normal([d, q], stddev = 1, seed = 1))

[0084] b1 = tf.Variable(tf.constant(0.0, shape=[q]))

[0085] w2 = tf.Variable(tf.random.normal([q, l], stddev=1, seed=1))

[0086] b2 = tf.Variable(tf.constant(0.0, shape=[l]))

[0087] The algorithm starts from the output layer and constructs a loss function to calculate the reverse layer-by-layer adjustment of weights and biases to minimize the prediction error. It is implemented based on the Python language and TensorFlow components. The following is the key code:

[0088] loss = tf.reduce_mean(tf.square(y_ - y))

[0089] train_step = tf.compat.v1.train.AdamOptimizer(eta).minimize(loss)

[0090] The Sigmoid function is selected as the activation function, and the mean squared error is used as the loss function during the optimization process of the network. It is implemented based on the Python language and TensorFlow components. The following is the key code:

[0091] a = tf.nn.sigmoid(tf.matmul(x, w1) + b1)

[0092] y = tf.nn.sigmoid(tf.matmul(a, w2) + b2)

[0093] A training efficiency threshold is set during the training process to avoid ineffective training and overfitting.

[0094] After the training is completed, an independent test set is used to verify the model to evaluate its generalization ability and accuracy. If the verification result of the test set shows that the model can accurately predict the unseen data and the model has not overfitted, it is considered that the prediction result of the neural network model is reliable and can predict the relationship between the raw material ratio of the ceramic core and its performance indicators.

[0095] Optionally, validating and testing the optimized prediction model of the neural network that has been trained to a certain extent using the validation set includes:

[0096] Determine the number of detections of the same performance parameter corresponding to each group of ceramic core materials, and determine the usage percentage value of the training data set according to the number of detections; wherein, the usage percentage value is negatively correlated with the number of detections.

[0097] When the percentage of the training data input for training in all the training data in the training set reaches the usage percentage value, use the validation set to perform a validation test on the neural network optimization prediction model that has been trained to a certain extent.

[0098] In this embodiment, "a certain extent" is further quantified. Specifically, for each group of ceramic core materials determined above, divide them into multiple portions and fire them separately, and then correspondingly detect multiple sets of performance parameter sets. Each set of performance parameter sets contains a set of detection values of shrinkage rate, porosity, room temperature bending strength, and high temperature bending strength.

[0099] When the number of portions into which each group of ceramic core materials is divided is larger, the number of sets of performance parameter sets obtained correspondingly is larger. In this way, the diversity of the training data corresponding to the same group of ceramic core materials constructed is higher, and generally the training effect on the neural network optimization prediction model is better. Correspondingly, the usage percentage value is set smaller. That is, after using a smaller proportion of the training data in the training set, the validation set is started to be used for validation testing.

[0100] On the contrary, when the number of portions into which each group of ceramic core materials is divided is smaller, the number of sets of performance parameter sets obtained correspondingly is smaller. In this way, the diversity of the training data corresponding to the same group of ceramic core materials constructed is lower, and generally the training effect on the neural network optimization prediction model is worse. The usage percentage value is set larger. That is, after using a larger proportion of the training data in the training set (i.e., after more sufficient training), the validation set is started to be used for validation testing.

[0101] By setting like this, it is possible to dynamically determine the usage percentage value corresponding to "a certain extent" based on the high or low diversity of the training data, so as to achieve a balance between the training effect and the training efficiency.

[0102] In addition, if it is shown during one validation that the training still does not meet the standard, the usage percentage values adopted in subsequent secondary training, tertiary training, etc. can be fixed values. For example, when the cumulative usage proportion of the training data increases to 10% again, the validation test is started.

[0103] This application also provides an electronic device, which includes: a memory storing executable program code; a processor coupled to the memory; the processor calls the executable program code stored in the memory and executes the method as described in any of the previous items.

[0104] The present application also provides a computer storage medium storing a computer program, and the computer program is executed by a processor to implement the method described in any of the preceding items.

[0105] The present application also provides a computer program product, including a computer program stored on a non-transitory computer-readable medium, and the computer program implements the method described in any of the preceding items when executed by a processor.

[0106] The present application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the specified functions in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0107] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device, and the instruction device implements the specified functions in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0108] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the specified functions in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0109] As described above, the above are only specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed in the present application should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for predicting physical properties of ceramic core materials and optimizing materials based on neural networks, characterized in that: The method comprises: Constructing a neural network optimization prediction model, and training the neural network optimization prediction model until the training meets the standards; Determine the raw material composition of the ceramic core, wherein the raw material composition includes the addition amount of the mineralizer, and input the raw material composition into the trained neural network optimization prediction model to obtain corresponding performance parameters; Inputting the performance parameters into a comprehensive performance evaluation model to obtain a comprehensive score of the raw material components; The comprehensive scores of the multiple groups of raw material components are compared, and the raw material component with the highest comprehensive score is selected as the optimal material ratio, that is, the addition amount of the mineralizer with the best comprehensive performance is determined.

2. The method for predicting physical properties of ceramic core materials and optimizing materials based on neural network according to claim 1, characterized in that: The performance parameters include shrinkage rate, porosity, room temperature flexural strength, and high temperature flexural strength.

3. The method for predicting physical properties of ceramic core materials and optimizing materials based on neural network according to claim 1, characterized in that: The comprehensive performance evaluation model uses a constraint equation to calculate the score of each performance parameter to obtain a comprehensive score of the raw material component. The constraint equation is: Score=W1*f A (x1)+W2*f B (x2)+W3*f C (x3)+W4*f D (x4) In the formula, Score is the comprehensive score, f A (x1),f B (x2), f C (x3),f D (x4) are the scores of each performance parameter, and W1, W2, W3, and W4 are weight coefficients.

4. The method for predicting physical properties of ceramic core materials and optimizing materials based on neural network according to claim 1, characterized in that: The calculation method of the score of each performance parameter is as follows: The shrinkage score is calculated as follows: The porosity score is calculated as follows: The room temperature flexural strength score is calculated as: The calculation method of high temperature flexural strength score is: Wherein, the k value is the change trend factor of the adjustment judgment function, and x1, x2, x3, and x4 are respectively the shrinkage rate, porosity, room temperature flexural strength, and high temperature flexural strength predicted and output by the neural network optimization prediction model.

5. The method for predicting physical properties of ceramic core materials and optimizing materials based on neural network according to claim 1, characterized in that: The weight coefficients of the performance parameters are specifically: The weight coefficient W1 of the shrinkage rate is 0.30, the weight coefficient W2 of the porosity is 0.15, the weight coefficient W3 of the room temperature flexural strength is 0.30, and the weight coefficient W4 of the high temperature flexural strength is 0.

25.

6. The method for predicting physical properties of ceramic core materials and optimizing materials based on neural network according to claim 1, characterized in that: The raw material components of the ceramic core include high-purity quartz glass powder as a matrix material, zirconium silicate as an additive, and the mineralizer is fused corundum.

7. The method for predicting physical properties of ceramic core materials and optimizing materials based on neural network according to claim 1, characterized in that: The step of training the neural network optimization prediction model until the training reaches the target includes: Preparing multiple groups of ceramic core materials, each group of ceramic core materials has a different amount of mineralizer added, detecting each performance parameter of each group of ceramic core materials, and constructing the amount of mineralizer added and each performance parameter into a training data; Dividing all the training data into a training set and a validation set, using the training set to train the neural network optimization prediction model, and using the validation set to perform a validation test on the neural network optimization prediction model that has been trained to a certain extent; If the verification test result meets the standard, the training is determined to have met the standard and the training is terminated; otherwise, the training is continued using the remaining training data in the training set.

8. An electronic device, characterized in that: The electronic device comprises: a memory storing executable program code; a processor coupled to the memory; the processor calls the executable program code stored in the memory to execute the method according to any one of claims 1-7.

9. A computer storage medium storing a computer program, characterized in that: The computer program is executed by a processor to implement the method according to any one of claims 1 to 7.

10. A computer program product comprising a computer program stored on a non-transitory computer readable medium, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images