Method for predicting thickness of chemical reaction area of composite explosive based on machine learning model

Abstract

The invention provides a method for predicting the thickness of a chemical reaction area of a composite explosive based on a machine learning model, and the method employs the machine learning model, and the input data of the machine learning model comprises the composition of a composite explosive formula, the charge density and the CJ detonation velocity; the output result of the machine learning model comprises the thickness of the chemical reaction area of the composite explosive. The machine learning model adopts a random forest model. The model parameters of the random forest model comprise the number of decision trees, the maximum feature number, the maximum depth of the decision trees and the minimum leaf node sample number. The machine learning model construction method is distinct in hierarchy and clear in process, can accurately predict the thicknesses of chemical reaction areas of various mixed explosives, and has wide applicability. The machine learning model involved in the invention is constructed based on a random forest model, shows good prediction performance and over-fitting resistance, has good noise immunity, and is insensitive to missing values of sample data.

Description

Technical Field

[0001] The present invention belongs to the technical field of mixed explosives, relates to simulation calculation of explosive formula performance, and particularly relates to a method for predicting the thickness of a chemical reaction zone of a mixed explosive based on a machine learning model. Background Art

[0002] Detonation characteristics are an essential and crucial aspect of fundamental research on the performance of mixed explosives. Characterizing and calculating the structural characteristics of the detonation chemical reaction zone further enhances this understanding. The detonation chemical reaction zone is the specific region of a mixed explosive where the detonation energy is concentrated. Understanding and predicting the thickness of this zone is crucial for understanding and mastering the detonation energy release process of mixed explosives. This knowledge provides essential technical support for the design of the detonation dynamics of mixed explosives, furthering fundamental scientific research on the detonation process of mixed explosives.

[0003] Currently, the prediction of the thickness of the detonation chemical reaction zone relies primarily on experimental techniques. Generally, due to the very limited thickness of the detonation chemical reaction zone of mixed explosives and the relatively short duration of the reaction, experimental testing is technically challenging. Furthermore, the detonation chemical reaction of mixed explosives introduces complex safety risks, requiring a high level of safety precautions and risk mitigation measures. These uncertainties can easily lead to test failures, significantly increasing personnel and financial costs, and making repeated technical research difficult.

[0004] Chinese invention patent application number CN 202410033604.X discloses a device for testing the shock initiation, detonation velocity, and detonation reaction zone parameters of solid propellants. The device includes a shock initiation test module, a detonation velocity test module, and a detonation reaction zone test module, organically combining the functions of shock reaction growth pressure history testing, detonation velocity testing, and detonation reaction zone parameter testing. However, this invention only involves an experimental device and does not propose a method for predicting the thickness of the detonation chemical reaction zone from a basic theoretical or practical perspective. Experimental methods are still required to obtain results. Summary of the Invention

[0005] In view of the shortcomings of the existing technology, the purpose of the present invention is to provide a method for predicting the thickness of the chemical reaction zone of mixed explosives based on a machine learning model, so as to solve the technical problem that the accuracy of the thickness prediction method in the existing technology needs to be further improved.

[0006] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0007] A method for predicting the thickness of a chemical reaction zone of a mixed explosive based on a machine learning model. The method adopts a machine learning model, wherein the input data of the machine learning model include the composition of the mixed explosive formula, charge density and CJ detonation velocity; and the output result of the machine learning model includes the thickness of the chemical reaction zone of the mixed explosive.

[0008] The machine learning model adopts a random forest model.

[0009] The model parameters of the random forest model include the number of decision trees, the maximum number of features, the maximum depth of the decision tree and the minimum number of leaf node samples.

[0010] The present invention also has the following technical features:

[0011] Specifically, the method includes the following steps:

[0012] Step 1: Determine the input data and output results.

[0013] Step 2: Get the sample data set:

[0014] Collect and pre-process sample data to form a sample data set used for model training, tuning, and model validation testing.

[0015] Step 3: Model structure design and training optimization:

[0016] Combined with the test set in the sample data set obtained in step 2, the structure of the random forest model is designed and the parameters of the random forest model are trained and tuned.

[0017] Step 4: Model validation, testing and saving:

[0018] Based on the validation set in the sample data set obtained in step 2, the prediction performance of the random forest model trained and tuned in step 3 is tested, the formed model structure and parameters are saved, and the final machine learning model for predicting the thickness of the chemical reaction zone of the mixed explosive is obtained.

[0019] Step 5: Prediction of the thickness of the chemical reaction zone of the mixed explosive:

[0020] Based on the machine learning model for predicting the thickness of the chemical reaction zone of the mixed explosive obtained in step 4, the input data of the mixed explosive to be tested is input, and the thickness of the chemical reaction zone of the mixed explosive to be tested is output.

[0021] Compared with the prior art, the present invention has the following technical effects:

[0022] (I) The machine learning model construction method constructed by the present invention has a clear hierarchy and process, can accurately predict the thickness of the chemical reaction zone of various mixed explosives, and has a wide range of applicability.

[0023] (II) The machine learning model involved in the present invention is constructed based on a random forest model, exhibits good prediction performance and resistance to overfitting, has good noise resistance, and is insensitive to missing values ​​in sample data.

[0024] The specific contents of the present invention are further explained in detail below with reference to the embodiments. DETAILED DESCRIPTION

[0025] It should be noted that, unless otherwise specified, all raw materials, models and methods in the present invention are those known in the prior art.

[0026] Research has shown that machine learning methods have shown promising technical promise in predicting the detonation performance of mixed explosives, with the ability to accurately predict detonation performance parameters. Therefore, the present invention provides a machine learning model and construction method for predicting the thickness of the chemical reaction zone of mixed explosives, offering a new technical solution for predicting the thickness of the detonation reaction zone of mixed explosives.

[0027] The overall inventive concept of the present invention is to utilize appropriate model input data based on a regression prediction model within a machine learning model to generate an output result of a predicted value for the thickness of the chemical reaction zone of the mixed explosive through prediction by the machine learning model. Currently, the present invention utilizes a random forest regression prediction model.

[0028] The present invention provides a method for predicting the thickness of the chemical reaction zone of a mixed explosive based on a machine learning model, the method comprising the following steps:

[0029] Step 1: Determine the input data and output results:

[0030] The input data of the machine learning model include the composition of the mixed explosive formula, charge density and CJ detonation velocity; the output results of the machine learning model include the thickness of the chemical reaction zone of the mixed explosive.

[0031] In step 1, the mixed explosive formula preferably comprises the chemical structural formula of each component and the corresponding mass percentage. The chemical elements included in the chemical structural formula of each component include C, H, N, O and Al.

[0032] In the present invention, the relative molecular mass of each component is calculated using the relative atomic mass of each chemical element as a constant. Assuming the total mass fraction of the mixed explosive is 100 parts, the mass fraction of each component is calculated based on the mass percentage of each component. The molar fraction of each component is then calculated based on the mass fraction and relative molecular mass of each component. The molar content of each chemical element in each component is calculated by combining the obtained molar fractions and the simplified chemical structure formula. The molar content of each chemical element in each component is then added together for each chemical element of the same type, thereby obtaining the simplified chemical structure formula of the mixed explosive formulation.

[0033] In the present invention, the composition of the mixed explosive formula is a dimensionless quantity, and the unit of charge density can be set to g / cm 3 The unit of CJ detonation velocity can be set to m / s. In addition, the thickness of the chemical reaction zone of the mixed explosive is a model output result, and its unit can be set to the length unit mm.

[0034] In the present invention, charge density refers to the actual formula density of the formed mixed explosive.

[0035] In the present invention, the CJ detonation velocity refers to the propagation velocity of the detonation wave at the CJ point when the mixed explosive is assumed to meet the ideal detonation Chapman-Jouguet theory conditions.

[0036] Step 2: Get the sample data set:

[0037] Collect and pre-process sample data to form a sample data set used for model training, tuning, and model validation testing.

[0038] Step 2 includes the following steps:

[0039] Step 201: Raw data collection:

[0040] Retrieve technical information on the thickness of the chemical reaction zone in the field of mixed explosives, obtain data information on the formula composition of mixed explosives, charge density, CJ detonation velocity and thickness of the chemical reaction zone, and collect the original data.

[0041] In the present invention, the sources of technical information include public academic papers, technical reports, books, monographs, patents, and relevant technical content involved in conferences.

[0042] Step 202: Data pre-processing:

[0043] Based on the original data collected in step 201, the original data is standardized to form a sample data set. The sample data set is divided into a test set and a validation set.

[0044] In the present invention, the test set is used for model training and tuning, and the validation set is used for model validation testing.

[0045] In the present invention, the ratio of the number of data entries contained in the test set to that contained in the validation set can be set in the range of 7:3 to 8:2.

[0046] Step 3: Model structure design and training optimization:

[0047] The machine learning model uses the random forest model. Combined with the test set in the sample data set obtained in step 2, the structure of the random forest model is designed and the parameters of the random forest model are trained and optimized.

[0048] In the present invention, the random forest model is a machine learning model based on ensemble learning. It improves the accuracy and stability of prediction results by combining multiple decision tree models and is suitable for regression-type machine learning tasks. The problem of predicting the thickness of the chemical reaction zone of mixed explosives is essentially a regression-type task that uses input data as multidimensional variables, predicts using a machine learning model, and obtains a one-dimensional output result. The random forest model can demonstrate the technical advantages of high accuracy and low overfitting in this type of regression-type task. By comprehensively averaging the prediction results of multiple decision trees to obtain the model output result, it can significantly reduce the computational variance of a single decision tree model and improve the generalization ability of the ensemble model. Bootstrap sampling is used when selecting the sample set to ensure sample randomness, ensuring that each tree model uses a different training set. At the same time, randomness is also maintained in feature prediction. When splitting a node, each tree model searches for the optimal split point from a randomly selected feature subset.

[0049] Step 3 includes the following steps:

[0050] Step 301: Model structure design:

[0051] First, based on the random forest model, sample data is randomly sampled from the test set; second, some features are randomly selected for each decision tree in the random forest model for optimal division; then, model parameters are set for each decision tree and the model is trained; finally, the output results obtained from each decision tree are averaged to obtain the model output result.

[0052] In step 301, the model parameters include the number of decision trees, the maximum number of features, the maximum depth of the decision tree, and the minimum number of leaf node samples.

[0053] Step 302: Model training and optimization:

[0054] The parameter optimization method is used to optimize and adjust the model parameters of each decision tree to improve the accuracy of the output results of the random forest model.

[0055] In step 302, the parameter optimization method includes: a grid search method, a random search method and a Bayesian parameter optimization method.

[0056] Step 4: Model validation, testing and saving:

[0057] Based on the validation set in the sample data set obtained in step 2, the prediction performance of the random forest model trained and tuned in step 3 is tested, the formed model structure and parameters are saved, and the final machine learning model for predicting the thickness of the chemical reaction zone of the mixed explosive is obtained.

[0058] Step 4 includes the following steps:

[0059] Step 401: Model verification test:

[0060] Randomly select verification cases from the verification set to test the prediction performance of the random forest model trained and tuned in step 3. Use statistical indicators to evaluate the performance of the random forest model output results to complete the verification test of the random forest model.

[0061] In step 401, the statistical indicators include: mean square error and coefficient of determination.

[0062] The expressions of mean square error and determination coefficient are based on the common expressions of mean square error MSE and determination coefficient R commonly used in the art. 2 Common expressions of .

[0063] Step 402: Save the model.

[0064] For the random forest model verified in step 401, all model parameters contained in the random forest model are saved in the local hardware in the form of a file to obtain a final machine learning model for predicting the thickness of the chemical reaction zone of the mixed explosive.

[0065] Step 5: Prediction of the thickness of the chemical reaction zone of the mixed explosive:

[0066] Based on the machine learning model for predicting the thickness of the chemical reaction zone of the mixed explosive obtained in step 4, the input data of the mixed explosive to be tested is input, and the thickness of the chemical reaction zone of the mixed explosive to be tested is output.

[0067] In the present invention, when the user adds new input data according to the format requirements, the saved machine learning model for predicting the thickness of the chemical reaction zone of the mixed explosive is called to perform prediction, obtain the model output result, and give the prediction result of the thickness of the chemical reaction zone of the mixed explosive to be tested.

[0068] Specific embodiments of the present invention are given below. It should be noted that the present invention is not limited to the following specific embodiments, and all equivalent changes made on the basis of the technical solution of this application fall within the protection scope of the present invention.

[0069] Example 1:

[0070] This embodiment provides a method for predicting the thickness of the chemical reaction zone of a mixed explosive based on a machine learning model according to the above-mentioned solution of the present invention. The method comprises the following steps:

[0071] Step 1: Determine the input data and output results:

[0072] The input data of the machine learning model include the composition of the mixed explosive formula, charge density and CJ detonation velocity; the output results of the machine learning model include the thickness of the chemical reaction zone of the mixed explosive.

[0073] In this embodiment, the mixed explosive is code-named Case 1, and the formula mass composition is RDX (RDX) / binder = 95 / 5, and the converted chemical structure formula is C 1.6396 H 3.2792 N 2.5663 O 2.5663 , the charge density is 1.66g / cm 3 , CJ explosion speed is about 7800m / s.

[0074] In this embodiment, the adhesive is a commonly used adhesive known in the art, such as paraffin adhesive.

[0075] Step 2: Get the sample data set:

[0076] Collect and pre-process sample data to form a sample data set used for model training, tuning, and model validation testing.

[0077] In this embodiment, the number of data entries contained in the sample data set is about 30, which is divided into a test set and a validation set, and the ratio of the number of data entries contained in the test set to the validation set is set to 7:3.

[0078] Step 3: Model structure design and training optimization:

[0079] The machine learning model uses the random forest model. Combined with the test set in the sample data set obtained in step 2, the structure of the random forest model is designed and the parameters of the random forest model are trained and optimized.

[0080] In this embodiment, the number of decision trees is 150, the maximum number of features is 0.8, the maximum depth of the decision tree is 6, and the minimum number of leaf node samples is 7.

[0081] In this embodiment, the parameter optimization method uses a grid search method.

[0082] Step 4: Model validation, testing and saving:

[0083] Based on the validation set in the sample data set obtained in step 2, the prediction performance of the random forest model trained and tuned in step 3 is tested, the formed model structure and parameters are saved, and the final machine learning model for predicting the thickness of the chemical reaction zone of the mixed explosive is obtained.

[0084] Step 5: Prediction of the thickness of the chemical reaction zone of the mixed explosive:

[0085] Based on the machine learning model for predicting the thickness of the chemical reaction zone of the mixed explosive obtained in step 4, the input data of the mixed explosive to be tested is input, and the thickness of the chemical reaction zone of the mixed explosive to be tested is output.

[0086] In this embodiment, the predicted thickness of the chemical reaction zone of the mixed explosive is approximately 0.140902 mm.

[0087] Verification of prediction results: For the mixed explosive of this embodiment, the thickness of the chemical reaction zone of the mixed explosive obtained by testing using the traditional method (specifically, see the passivated RDX explosive disclosed in the paper "Guo Wei, Cao Wei, Tan Kaiyuan, et al. Detonation Reaction Zone Parameter Measurement of RDX-based Metallized Explosives [J]. Energetic Materials, 2021, 29(5): 389-393") is 0.15 mm. This shows that the machine learning model for predicting the thickness of the chemical reaction zone of the mixed explosive in this embodiment is accurate and shows good applicability.

[0088] Example 2:

[0089] This embodiment provides a method for predicting the thickness of the chemical reaction zone of a mixed explosive based on a machine learning model according to the above-mentioned solution of the present invention. The method comprises the following steps:

[0090] Step 1: Determine the input data and output results:

[0091] The input data of the machine learning model include the composition of the mixed explosive formula, charge density and CJ detonation velocity; the output results of the machine learning model include the thickness of the chemical reaction zone of the mixed explosive.

[0092] In this embodiment, the mixed explosive is code-named Case 2, and the formula mass composition is DNAN (2,4-dinitroanisole) / HMX (octogen) / Al / binder = 31 / 40 / 26 / 3, and the converted chemical structure formula is C 1.8494 H 2.4471 N 1.393 5O 1.8628 Al 0.9636 , the charge density is 1.86g / cm 3, CJ explosion speed is about 7120m / s.

[0093] In this embodiment, the adhesive is a commonly used adhesive known in the art, such as paraffin adhesive.

[0094] Step 2: Get the sample data set:

[0095] Collect and pre-process sample data to form a sample data set used for model training, tuning, and model validation testing.

[0096] In this embodiment, the number of data entries contained in the sample data set is about 40, which is divided into a test set and a validation set, and the ratio of the number of data entries contained in the test set to the validation set is set to 8:2.

[0097] Step 3: Model structure design and training optimization:

[0098] The machine learning model uses the random forest model. Combined with the test set in the sample data set obtained in step 2, the structure of the random forest model is designed and the parameters of the random forest model are trained and optimized.

[0099] In this embodiment, the number of decision trees is 200, the maximum feature number is 0.7, the maximum depth of the decision tree is 11, and the minimum number of leaf node samples is 9.

[0100] In this embodiment, the parameter optimization method uses a random search method.

[0101] Step 4: Model validation, testing and saving:

[0102] Based on the validation set in the sample data set obtained in step 2, the prediction performance of the random forest model trained and tuned in step 3 is tested, the formed model structure and parameters are saved, and the final machine learning model for predicting the thickness of the chemical reaction zone of the mixed explosive is obtained.

[0103] Step 5: Prediction of the thickness of the chemical reaction zone of the mixed explosive:

[0104] Based on the machine learning model for predicting the thickness of the chemical reaction zone of the mixed explosive obtained in step 4, the input data of the mixed explosive to be tested is input, and the thickness of the chemical reaction zone of the mixed explosive to be tested is output.

[0105] In this embodiment, the predicted thickness of the chemical reaction zone of the mixed explosive is approximately 0.936686 mm.

[0106] Verification of prediction results: For the mixed explosive of this embodiment, the thickness of the chemical reaction zone of the mixed explosive was 1.073 mm using the traditional method (see the RBOL-2 explosive disclosed in the paper "Yang Yang, Duan Zhuoping, Zhang Liansheng, et al. Detonation Performance of Two DNAN-Based Aluminum-Containing Explosives [J]. Energetic Materials, 2019, 27(8): 679-684."). This shows that the machine learning model for predicting the thickness of the chemical reaction zone of the mixed explosive in this embodiment is accurate and shows good applicability.

[0107] Example 3:

[0108] This embodiment provides a method for predicting the thickness of the chemical reaction zone of a mixed explosive based on a machine learning model according to the above-mentioned solution of the present invention. The method comprises the following steps:

[0109] Step 1: Determine the input data and output results:

[0110] The input data of the machine learning model include the composition of the mixed explosive formula, charge density and CJ detonation velocity; the output results of the machine learning model include the thickness of the chemical reaction zone of the mixed explosive.

[0111] In this embodiment, the mixed explosive is code-named Case 3, and the formula mass composition is TATB (1,3,5-triamino-2,4,6-trinitrobenzene) / binder = 95 / 5. The converted chemical structure formula is C 2.5645 H 2.9210 N 2.2081 O 2.2081 , the charge density is 1.895g / cm 3 , CJ explosion speed is about 7660m / s.

[0112] In this embodiment, the adhesive is a commonly used adhesive known in the art, such as paraffin adhesive.

[0113] Step 2: Get the sample data set:

[0114] Collect and pre-process sample data to form a sample data set used for model training, tuning, and model validation testing.

[0115] In this embodiment, the number of data entries contained in the sample data set is about 35, which is divided into a test set and a validation set, and the ratio of the number of data entries contained in the test set to the validation set is set to 8:2.

[0116] Step 3: Model structure design and training optimization:

[0117] The machine learning model uses the random forest model. Combined with the test set in the sample data set obtained in step 2, the structure of the random forest model is designed and the parameters of the random forest model are trained and optimized.

[0118] In this embodiment, the number of decision trees is 180, the maximum number of features is 0.6, the maximum depth of the decision tree is 15, and the minimum number of leaf node samples is 5.

[0119] In this embodiment, the parameter optimization method uses the Bayesian parameter optimization method.

[0120] Step 4: Model validation, testing and saving:

[0121] Based on the validation set in the sample data set obtained in step 2, the prediction performance of the random forest model trained and tuned in step 3 is tested, the formed model structure and parameters are saved, and the final machine learning model for predicting the thickness of the chemical reaction zone of the mixed explosive is obtained.

[0122] Step 5: Prediction of the thickness of the chemical reaction zone of the mixed explosive:

[0123] Based on the machine learning model for predicting the thickness of the chemical reaction zone of the mixed explosive obtained in step 4, the input data of the mixed explosive to be tested is input, and the thickness of the chemical reaction zone of the mixed explosive to be tested is output.

[0124] In this embodiment, the predicted thickness of the chemical reaction zone of the mixed explosive is approximately 1.593801 mm.

[0125] Verification of prediction results: For the mixed explosive of this embodiment, the thickness of the chemical reaction zone of the mixed explosive was 1.5±0.2 mm using the traditional method (see the JB-9014 explosive disclosed in the paper "Pei Hongbo, Huang Wenbin, Qin Jincheng, et al. Research on the Reaction Zone Structure of JB-9014 Explosive Based on Doppler Velocimetry Technology [J]. Explosion and Shock Waves, 2018, 38(3): 485-490."). This shows that the machine learning model for predicting the thickness of the chemical reaction zone of the mixed explosive in this embodiment is accurate and shows good applicability.

Claims

1. A method for predicting the thickness of the chemical reaction zone of a mixed explosive based on a machine learning model, characterized in that: The method adopts a machine learning model, the input data of which include the composition of the mixed explosive formula, charge density and CJ detonation velocity; the output result of the machine learning model includes the thickness of the chemical reaction zone of the mixed explosive.

2. The method for predicting the thickness of the chemical reaction zone of a mixed explosive based on a machine learning model as claimed in claim 1, wherein: The machine learning model adopts a random forest model.

3. The method for predicting the thickness of the chemical reaction zone of a mixed explosive based on a machine learning model as claimed in claim 2, wherein: The model parameters of the random forest model include the number of decision trees, the maximum number of features, the maximum depth of the decision tree and the minimum number of leaf node samples.

4. The method for predicting the thickness of the chemical reaction zone of a mixed explosive based on a machine learning model as claimed in claim 3, wherein: The method comprises the following steps: Step 1: Determine input data and output results; Step 2: Get the sample data set: Collect and pre-process sample data to form a sample data set used for model training, tuning, and model validation testing; Step 3: Model structure design and training optimization: Combined with the test set in the sample data set obtained in step 2, the structure of the random forest model is designed and the parameters of the random forest model are trained and optimized; Step 4: Model validation, testing and saving: Based on the validation set in the sample data set obtained in step 2, the prediction performance of the random forest model trained and tuned in step 3 is tested, the formed model structure and parameters are saved, and the final machine learning model for predicting the thickness of the chemical reaction zone of the mixed explosive is obtained; Step 5: Prediction of the thickness of the chemical reaction zone of the mixed explosive: Based on the machine learning model for predicting the thickness of the chemical reaction zone of the mixed explosive obtained in step 4, the input data of the mixed explosive to be tested is input, and the thickness of the chemical reaction zone of the mixed explosive to be tested is output.

5. The method for predicting the thickness of the chemical reaction zone of a mixed explosive based on a machine learning model as claimed in claim 4, wherein: In step 1, the mixed explosive formula comprises the chemical structural formula of each component and the corresponding mass percentage.

6. The method for predicting the thickness of the chemical reaction zone of a mixed explosive based on a machine learning model as claimed in claim 5, wherein: In step 1, the chemical elements included in the simplified chemical structure formula of each component include C, H, N, O and Al.

7. The method for predicting the thickness of the chemical reaction zone of a mixed explosive based on a machine learning model as claimed in claim 4, wherein: Step 2 includes the following steps: Step 201: Raw data collection: Obtain data information about the composition of mixed explosives, charge density, CJ detonation velocity and thickness of the chemical reaction zone, and collect them to form raw data; Step 202: Data pre-processing: Based on the original data collected in step 201, the original data is standardized to form a sample data set; the sample data set is segmented to form a test set and a validation set.

8. The method for predicting the thickness of the chemical reaction zone of a mixed explosive based on a machine learning model as claimed in claim 4, wherein: Step 3 includes the following steps: Step 301: Model structure design: First, based on the random forest model, sample data is randomly sampled from the test set. Second, some features are randomly selected for each decision tree in the random forest model to perform optimal partitioning. Then, model parameters are set for each decision tree and the model is trained. Finally, the output results of each decision tree are averaged to obtain the model output result. Step 302: Model training and optimization: The parameter optimization method is used to optimize and adjust the model parameters of each decision tree.

9. The method for predicting the thickness of the chemical reaction zone of a mixed explosive based on a machine learning model as claimed in claim 8, wherein: In step 302, the parameter optimization method includes: a grid search method, a random search method and a Bayesian parameter optimization method.

10. The method for predicting the thickness of the chemical reaction zone of a mixed explosive based on a machine learning model as claimed in claim 4, wherein: Step 4 includes the following steps: Step 401: Model verification test: Randomly select validation cases from the validation set to test the prediction performance of the random forest model trained and tuned in step 3. Use statistical indicators to evaluate the performance of the random forest model output results to complete the validation test of the random forest model. In step 401, the statistical indicators include: mean square error and coefficient of determination; Step 402: Save the model. For the random forest model verified in step 401, all model parameters contained in the random forest model are saved in the local hardware in the form of a file to obtain a final machine learning model for predicting the thickness of the chemical reaction zone of the mixed explosive.