Method and device for predicting adsorption coefficient of nuclide on granite and storage medium
Through machine learning technology, extreme gradient enhancement models are trained to predict the adsorption coefficient of nuclides on granite, solving the problems of low prediction accuracy and high cost in the existing technology, and achieving efficient and accurate prediction of adsorption coefficients.
Patent Information
- Application Number
- CN202411777783.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-05
- Publication Date
- 2025-03-14
AI Technical Summary
The prior art has problems such as time-consuming and laborious experiments, low prediction accuracy of mechanism model, and poor generalization ability of multiple regression methods when predicting the adsorption coefficient of nuclides on granite.
Using machine learning technology, the specific steps include obtaining experimental data of historical nuclide adsorption on granite, extracting characteristic variables and target variables, training extreme gradient enhancement models, and using the trained model to predict the adsorption coefficient.
Through machine learning model prediction, the experimental cost is significantly reduced, the prediction efficiency and accuracy are improved, and it is suitable for different types of nuclides and has wide applicability.
Smart Images
Figure CN119626378A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of machine learning, and in particular relates to a method and device for predicting the adsorption coefficient of nuclides on granite, and a storage medium. Background Art
[0002] Deep geological disposal is a long-term disposal method for high-level radioactive waste recognized by nuclear countries in the world. This method effectively isolates radionuclides from the biosphere by placing them in rock formations deep underground. Granite has the characteristics of low permeability, high mechanical strength and long-term stability and is selected as the main surrounding rock type by many countries. The migration of nuclides in granite media involves mechanisms such as convection, diffusion, adsorption, and matrix diffusion. Among them, adsorption is the most important mechanism to block the migration of radionuclides in granite. The adsorption coefficient is a parameter used to quantify the adsorption performance of nuclides on granite, and its value will show significant differences under different environmental conditions. Therefore, accurately predicting the adsorption coefficient of nuclides on granite under different environmental conditions is of great significance for simulating the migration behavior of nuclides in granite and evaluating the safety of geological disposal repositories for high-level radioactive waste. The main methods for obtaining the adsorption coefficient of nuclides on granite include experimental method, mechanism model method, multiple regression method, etc. These traditional methods usually have certain limitations in obtaining the adsorption coefficient, such as: the experimental method is time-consuming and labor-intensive, the prediction accuracy of the mechanism model method is low, and the generalization ability of the multiple regression method is poor. Summary of the invention
[0003] The technical problem to be solved by the present invention is to provide a method and device for predicting the adsorption coefficient of nuclides on granite, and a storage medium, which can predict the adsorption coefficients of various types of adsorptive solutes on granite, and is particularly suitable for nuclides involved in the geological disposal of high-level radioactive waste.
[0004] To achieve the above object, the present invention adopts the following technical solution:
[0005] A method for predicting the adsorption coefficient of nuclides on granite, comprising:
[0006] Step S1, obtaining experimental data of historical nuclides adsorbed on granite;
[0007] Step S2, extracting characteristic variables and target variables based on the experimental data of historical nuclides adsorbed on granite;
[0008] Step S3, training an extreme gradient boosting model according to the characteristic variables and target variables of historical nuclides adsorbed on granite;
[0009] Step S4: input the characteristic variables of the target nuclide adsorption on the granite into the trained extreme gradient boosting model to predict the adsorption coefficient.
[0010] Preferably, the experimental data of historical nuclides adsorbed on granite include: granite particle size, solid-liquid ratio, ionic strength, pH, cesium concentration, contact time, temperature, and adsorption coefficient.
[0011] Preferably, in step S2, according to the experimental data of historical radionuclides adsorption on granite, the correlation coefficients between all different variables are calculated based on Pearson correlation analysis, the adsorption coefficient is used as the target variable, the variables with too low correlation with the target variable are eliminated, and the granite particle size, solid-liquid ratio, ionic strength, pH, radionuclide concentration, and contact time are determined as characteristic variables.
[0012] The present invention also provides a device for predicting the adsorption coefficient of nuclides on granite, comprising:
[0013] The acquisition module is used to obtain the experimental data of historical nuclides adsorbed on granite;
[0014] An extraction module is used to extract characteristic variables and target variables based on the experimental data of historical nuclide adsorption on granite;
[0015] A training module, used to train the extreme gradient boosting model based on the characteristic variables and target variables of historical nuclides adsorbed on granite;
[0016] The prediction module is used to input the characteristic variables of the target nuclide adsorption on granite into the trained extreme gradient boosting model to predict the adsorption coefficient.
[0017] Preferably, the experimental data of historical nuclides adsorbed on granite include: granite particle size, solid-liquid ratio, ionic strength, pH, cesium concentration, contact time, temperature, and adsorption coefficient.
[0018] Preferably, the extraction module calculates the correlation coefficients between all different variables based on the Pearson correlation analysis according to the experimental data of historical radionuclides adsorption on granite, takes the adsorption coefficient as the target variable, eliminates the variables with too low correlation with the target variable, and determines the granite particle size, solid-liquid ratio, ionic strength, pH, radionuclide concentration, and contact time as characteristic variables.
[0019] An embodiment of the present invention further provides a storage medium, on which a computer program is stored, and when the computer program is run, a method for predicting the adsorption coefficient of nuclides on granite is executed.
[0020] Compared with the prior art, the present invention has the following technical effects:
[0021] 1. Reduce costs: Predictions through machine learning models avoid a large amount of experimental materials and manpower investment, significantly reducing experimental costs.
[0022] 2. Improve efficiency: The model has a fast prediction speed and can complete the prediction of a large amount of data in a short time, which improves the evaluation efficiency.
[0023] 3. Improved accuracy: Machine learning algorithms can capture nonlinear relationships between variables, improving the accuracy and reliability of predictions.
[0024] 4. Strong universality: The present invention is not only applicable to a specific nuclide, but can be extended to the prediction of adsorption coefficients of different types of nuclides on granite, and has wide applicability. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.
[0026] Figure 1 The figure is a flow chart of a method for predicting the adsorption coefficient of nuclides on granite according to an embodiment of the present invention. DETAILED DESCRIPTION
[0027] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0028] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.
[0029] Embodiment 1:
[0030] like Figure 1 As shown, an embodiment of the present invention provides a method for predicting the adsorption coefficient of nuclides on granite, comprising:
[0031] Step S1, obtaining experimental data of historical nuclides adsorbed on granite;
[0032] Step S2, extracting characteristic variables and target variables based on the experimental data of historical nuclides adsorbed on granite;
[0033] Step S3, training an extreme gradient boosting model (XGBoost model) according to the characteristic variables and target variables of historical nuclides adsorbed on granite;
[0034] Step S4: input the characteristic variables of the target nuclide adsorption on the granite into the trained extreme gradient boosting model to predict the adsorption coefficient.
[0035] As an implementation of an embodiment of the present invention, in step S1, the experimental data of historical nuclides adsorbed on granite include: granite particle size, solid-liquid ratio, ionic strength, pH, nuclide concentration, contact time, temperature, and adsorption coefficient.
[0036] Furthermore, the experimental data of historical nuclides adsorbed on granite were cleaned and normalized; data cleaning was used to remove outliers and missing values, and normalization was used to improve the accuracy and stability of the prediction.
[0037] As an implementation method of an embodiment of the present invention, in step S2, based on the experimental data of historical radionuclides adsorbed on granite, the correlation coefficient between all potential characteristic variables and the target variable is calculated based on Pearson correlation analysis, the adsorption coefficient is used as the target variable, and the variables with too low correlation with the target variable are eliminated, and the granite particle size, solid-liquid ratio, ionic strength, pH, radionuclide concentration, and contact time are determined as characteristic variables.
[0038] The calculation equation of Pearson's correlation coefficient is:
[0039]
[0040] Among them, ρ is the correlation coefficient; n represents the number of all potential feature variables; x and y represent the potential feature variables and the target variable respectively, x i and i represents the true value of the latent feature variable and the target variable, and is the mean of the latent feature variable and the target variable.
[0041] As an implementation method of an embodiment of the present invention, in step S3, according to the characteristic variables and target variables of the historical nuclides adsorbed on granite, the value range of the initial hyperparameter is set, and the learning rate is set to [0.01, 0.1, 0.2]. A smaller learning rate can ensure stable model optimization. The number of trees is set to [100, 200, 300], and the maximum depth is set to [3, 5, 7], so as to control the complexity of the tree and prevent the model from overfitting. Construct an objective function, use the extreme gradient boosting algorithm to iteratively optimize the objective function, gradually improve the model prediction accuracy, and make the objective function reach the minimum value. The expression of the objective function is:
[0042]
[0043] Where L(F) is the objective function, λ is the regularization coefficient, ∥F∥2 is a regularization term for model complexity. In each iteration, the residuals of the current model are calculated, a new tree is fitted to predict these residuals, and the model is updated. This process is repeated until the number of trees reaches a predetermined value or the error convergence condition is met.
[0044] The hyperparameter combination is optimized through the grid search method, and a grid of hyperparameter combinations with different learning rates, numbers, and maximum depths is constructed. Each combination is trained one by one, and the model performance under different combinations is evaluated based on model performance evaluation indicators such as the certainty coefficient. The calculation equation for the certainty coefficient is:
[0045]
[0046] Among them, R 2 is the coefficient of certainty, n is the number of experimental samples, K d,i is the measured value of the adsorption coefficient of the ith sample, is the predicted value of the adsorption coefficient of the ith sample, is the mean of the measured values of the adsorption coefficient.
[0047] Finally, the hyperparameter combination with the highest coefficient of certainty was obtained, and the extreme gradient boosting model trained under this parameter combination was used as the optimal model to quickly and accurately predict the adsorption behavior of nuclides on granite under different conditions.
[0048] Based on experience, a reasonable prediction scenario is set, such as: granite particle size is 0.1mm, ionic strength is 0.18mol / L, solid-liquid ratio is 100g / L, pH is 7.26, cesium concentration is 2.00e-5mol / L, and contact time is 1h. The constructed XGBoost model is used to predict the adsorption coefficient of cesium on granite under the above conditions, and the predicted result is 9.26e-4m 3 / kg. The accuracy of the prediction was tested by conducting static adsorption experiments, and the adsorption coefficient obtained was 9.37e-4m 3 / kg, the relative error of the prediction by the machine learning algorithm was within 2%.
[0049] Embodiment 2:
[0050] The embodiment of the present invention further provides a device for predicting the adsorption coefficient of nuclides on granite, comprising:
[0051] The acquisition module is used to obtain the experimental data of historical nuclides adsorbed on granite;
[0052] An extraction module is used to extract characteristic variables and target variables based on the experimental data of historical nuclide adsorption on granite;
[0053] A training module, used to train the extreme gradient boosting model based on the characteristic variables and target variables of historical nuclides adsorbed on granite;
[0054] The prediction module is used to input the characteristic variables of the target nuclide adsorption on granite into the trained extreme gradient boosting model to predict the adsorption coefficient.
[0055] As an implementation method of the embodiment of the present invention, the experimental data of the adsorption of historical nuclides on granite include: granite particle size, solid-liquid ratio, ionic strength, pH, cesium concentration, contact time, temperature, and adsorption coefficient.
[0056] As an implementation method of an embodiment of the present invention, the extraction module calculates the correlation coefficient between all different variables based on the Pearson correlation analysis according to the experimental data of historical radionuclides adsorbed on granite, takes the adsorption coefficient as the target variable, eliminates the variables with too low correlation with the target variable, and determines the granite particle size, solid-liquid ratio, ionic strength, pH, radionuclide concentration, and contact time as characteristic variables.
[0057] Embodiment 3:
[0058] An embodiment of the present invention further provides a storage medium, on which a computer program is stored, and when the computer program is run, a method for predicting the adsorption coefficient of nuclides on granite is executed.
[0059] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.
Claims
1. A method for predicting the adsorption coefficient of nuclides on granite, characterized in that: include: Step S1, obtaining experimental data of historical nuclides adsorbed on granite; Step S2, extracting characteristic variables and target variables based on the experimental data of historical nuclides adsorbed on granite; Step S3, training an extreme gradient boosting model according to the characteristic variables and target variables of historical nuclides adsorbed on granite; Step S4: input the characteristic variables of the target nuclide adsorption on the granite into the trained extreme gradient boosting model to predict the adsorption coefficient.
2. The method for predicting the adsorption coefficient of nuclides on granite according to claim 1, characterized in that: The experimental data of historical nuclides adsorbed on granite include: granite particle size, solid-liquid ratio, ionic strength, pH, cesium concentration, contact time, temperature, and adsorption coefficient.
3. The method for predicting the adsorption coefficient of nuclides on granite according to claim 2, characterized in that: In step S2, according to the experimental data of historical radionuclides adsorption on granite, the correlation coefficients between all different variables are calculated based on Pearson correlation analysis, the adsorption coefficient is used as the target variable, the variables with too low correlation with the target variable are eliminated, and the granite particle size, solid-liquid ratio, ionic strength, pH, radionuclide concentration, and contact time are determined as characteristic variables.
4. A device for predicting the adsorption coefficient of nuclides on granite, characterized in that: include: The acquisition module is used to obtain the experimental data of historical nuclides adsorbed on granite; An extraction module is used to extract characteristic variables and target variables based on the experimental data of historical nuclide adsorption on granite; A training module, used to train the extreme gradient boosting model based on the characteristic variables and target variables of historical nuclides adsorbed on granite; The prediction module is used to input the characteristic variables of the target nuclide adsorption on granite into the trained extreme gradient boosting model to predict the adsorption coefficient.
5. The device for predicting the adsorption coefficient of nuclides on granite according to claim 4, characterized in that: The experimental data of historical nuclides adsorbed on granite include: granite particle size, solid-liquid ratio, ionic strength, pH, cesium concentration, contact time, temperature, and adsorption coefficient.
6. The device for predicting the adsorption coefficient of nuclides on granite according to claim 5, characterized in that: The extraction module calculates the correlation coefficients between all different variables based on the Pearson correlation analysis according to the experimental data of historical radionuclides adsorption on granite, takes the adsorption coefficient as the target variable, eliminates the variables with too low correlation with the target variable, and determines the granite particle size, solid-liquid ratio, ionic strength, pH, radionuclide concentration, and contact time as characteristic variables.
7. A storage medium, characterized in that: The storage medium stores a computer program, which executes the method for predicting the adsorption coefficient of nuclides on granite as claimed in any one of claims 1 to 3 when running.