Method and system for predicting performance of two-electron oxygen reduction carbon-based electrocatalyst based on machine learning
Through machine learning-based methods, the preparation conditions of two electron oxygen reduction carbon-based electrocatalysts are optimized, and the problems of preparation and optimization in the prior art are solved, and the rapid improvement of catalyst performance and cost reduction are achieved.
Patent Information
- Application Number
- CN202510140278.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-08
- Publication Date
- 2025-05-27
AI Technical Summary
The prior art has problems such as time-consuming and labor-intensive and difficult to optimize the two-electron oxygen reduction carbon-based electrocatalysts, resulting in low efficiency and high cost in improving the performance of the catalyst.
Through machine learning-based methods, the properties parameters, preparation conditions and performance data of two electrons oxygen reduction of carbon-based electrocatalysts are collected, and data preprocessing and machine learning model training are carried out to obtain the best carbon-based electrocatalysts and preparation conditions.
It achieves rapid and accurate improvement of the performance of the two electron oxygen reduction reaction, reduces manual and experimental costs, and replaces the traditional trial and error experiment methods.
Smart Images
Figure CN120048377A_ABST
Abstract
Description
Technical Field:
[0001] The present invention relates to the technical fields of machine learning and big data applications, and particularly to a method and system for predicting the performance of a two-electron oxygen reduction carbon-based electrocatalyst based on machine learning. Background Art:
[0002] Carbon-based catalysts have great potential and development prospects in the efficient production of hydrogen peroxide in the 2e - ORR. With the global emphasis on green chemistry and sustainable development, the electrochemical method for preparing hydrogen peroxide has received extensive attention due to its environmental friendliness, high safety, and strong operability. Carbon materials themselves have good electrical conductivity, a large specific surface area, and excellent chemical stability, making them ideal catalytic carriers and support materials. However, the preparation of catalysts is often time-consuming and laborious, and there are various parameters in each stage of catalyst design that need to be further optimized, which increases the workload.
[0003] Therefore, there is an urgent need to study a method and system for predicting the performance of a two-electron oxygen reduction carbon-based electrocatalyst based on machine learning. Summary of the Invention:
[0004] The purpose of the invention is to overcome the defects of the above-mentioned prior art and provide a method and system for predicting the performance of a two-electron oxygen reduction carbon-based electrocatalyst based on machine learning. By optimizing each machine learning model, the best carbon-based electrocatalyst and corresponding preparation conditions can be obtained, which can quickly and accurately improve the performance of the two-electron oxygen reduction reaction and reduce labor and experimental costs.
[0005] The purpose of the present invention can be achieved by the following technical solutions:
[0006] The present invention provides a method for predicting the performance of a two-electron oxygen reduction carbon-based electrocatalyst based on machine learning, including the following steps:
[0007] S1: Collect the property parameters, preparation conditions, and corresponding performance data of the two-electron oxygen reduction of the carbon-based electrocatalyst;
[0008] S2: Convert the data collected in S1 into numerical data and perform normalization processing;
[0009] S3: Import the data processed in S2 into a machine learning model, adjust the hyperparameters of the machine learning model, and then perform training;
[0010] S4: According to the accuracy of the machine learning model in S3, use algorithm evaluation indicators to determine the optimal machine learning model in each data application scenario, and optimize under the optimal machine learning model to find the best carbon-based electrocatalyst and preparation conditions.
[0011] Further, the data collected in S1 includes classification data, and the classification data includes whether the catalyst is pure carbon and the source of heteroelements.
[0012] Further, the classification data is converted into numerical data using boolean values.
[0013] In S2, the normalization process is to normalize all numerical data to a similar order of magnitude by taking the logarithm.
[0014] Further, in S3, the machine learning models include Extreme Gradient Boosting (XGBoost), Random Forest, Gradient Boosting Tree, and Support Vector Regression.
[0015] Further, in S3, the hyperparameters of the machine learning model are adjusted using the grid search method.
[0016] Further, in S4, the algorithm evaluation metrics include R 2 , Root Mean Square Error (RMSE), and Mean Absolute Error (MAE).
[0017] The present invention provides a system for predicting the performance of two-electron oxygen reduction carbon-based electrocatalysts based on machine learning, including a data acquisition module, a data preprocessing module, a machine learning module, and an optimization feedback module;
[0018] The data acquisition module is used to collect the property parameters, preparation conditions of the carbon-based electrocatalyst, and the performance data of two-electron oxygen reduction, and the collected data is transmitted to the data preprocessing module;
[0019] The data preprocessing module is used to convert the classification data into numerical data, normalize the numerical data, and then transmit it to the machine learning module;
[0020] The machine learning module and the optimization feedback module are used to determine the optimal machine learning model under each data application scenario according to the accuracy of the machine learning model, and optimize under the optimal machine learning model to find the best carbon-based electrocatalyst and preparation conditions.
[0021] Further, the data acquisition module includes a carbon-based electrocatalyst parameter acquisition module, a preparation condition information collection module, and a two-electron oxygen reduction performance analysis module;
[0022] The carbon-based electrocatalyst parameter acquisition module is used to collect the source of the precursor of the carbon-based electrocatalyst and the carbon material structure;
[0023] The preparation condition information collection module is used to collect the temperature and time for preparing the carbon-based catalyst;
[0024] The two-electron oxygen reduction performance collection module is used to collect the H of the two-electron oxygen reduction reaction 2O 2 Selectivity and current density.
[0025] Beneficial effects:
[0026] Through the optimization process of each machine learning model, the optimal machine learning model is obtained, and under the optimal machine learning model, optimization is carried out to obtain the best carbon-based electrocatalyst and the corresponding preparation conditions, so as to achieve the purpose of quickly and accurately improving the performance of the two-electron oxygen reduction reaction. At the same time, the labor and experimental costs are reduced, replacing the traditional trial-and-error experiments. Description of the drawings:
[0027] Figure 1 It is a Pearson correlation coefficient analysis chart.
[0028] Figure 2 For H 2 O 2 Performance evaluation indicators of four machine learning models for selectivity.
[0029] Figure 3 Performance evaluation indicators of four machine learning models for current density.
[0030] Figure 4 Verification curves of 5 catalysts. Specific implementation manners:
[0031] The following further elaborates on the specific implementation manners of the present invention through examples. These examples are implemented on the premise of the solution described in the present invention, and the detailed implementation manners and specific operation processes are given. However, the protection scope of the present invention is not limited to the following examples.
[0032] The present invention will be further described below in conjunction with the drawings and specific examples. In the technical solution, features such as component models, material names, structure / module names, control modes, algorithms, process procedures, or composition ratios that are not clearly stated are regarded as common technical features disclosed in the prior art.
[0033] The present invention provides a method for predicting the performance of a two-electron oxygen reduction carbon-based electrocatalyst based on machine learning, including the following steps:
[0034] S1: Collect the parameters, preparation conditions, and corresponding performance data of the two-electron oxygen reduction of the carbon-based electrocatalyst;
[0035] S2: Convert the data collected in S1 into numerical data and perform normalization processing;
[0036] S3: Import the data processed in S2 into the machine learning model, adjust the hyperparameters of the machine learning model, and then perform training;
[0037] S4: According to the accuracy of the machine learning model in S3, use algorithm evaluation metrics to determine the optimal machine learning model for each data application scenario, and optimize under the optimal machine learning model to find the best carbon-based electrocatalyst and preparation conditions.
[0038] Further, in S2, the collected data includes categorical data and numerical data, and the categorical data includes whether the catalyst is pure carbon and the source of heteroelements.
[0039] Further, convert the categorical data into numerical data using boolean values.
[0040] Further, in S3, the machine learning models include Extreme Gradient Boosting, Random Forest, Gradient Boosting Tree, and Support Vector Regression.
[0041] Further, in S3, use the grid search method to adjust the hyperparameters of the machine learning model.
[0042] Further, in S4, the algorithm evaluation metrics include R 2 , RMSE, MAE.
[0043] The present invention provides a system for predicting the performance of two-electron oxygen reduction carbon-based electrocatalysts based on machine learning, including a data acquisition module, a data preprocessing module, a machine learning module, and an optimization feedback module;
[0044] The data acquisition module is used to collect the property parameters, preparation conditions of the carbon-based electrocatalyst, and the performance of two-electron oxygen reduction;
[0045] The data preprocessing module is used to convert categorical data into numerical data and normalize the numerical data;
[0046] The machine learning module and the optimization feedback module are used to determine the optimal machine learning model for each data application scenario according to the accuracy of the machine learning model, and optimize under the optimal machine learning model to find the best carbon-based electrocatalyst and preparation conditions.
[0047] Further, the data acquisition module includes a carbon-based electrocatalyst parameter acquisition module, a preparation condition information collection module, and a two-electron oxygen reduction performance analysis module;
[0048] The carbon-based electrocatalyst parameter acquisition module is used to collect the source of the precursor of the carbon-based electrocatalyst and the carbon material structure;
[0049] The preparation condition information collection module is used to collect the temperature and time for preparing the carbon-based catalyst;
[0050] The two-electron oxygen reduction performance collection module is used to collect H 2 O2 Selectivity and current density.
[0051] Example 1:
[0052] This example provides a method for predicting the performance of a two-electron oxygen reduction carbon-based electrocatalyst based on machine learning, including the following steps:
[0053] S1: Collect the parameters, preparation conditions, and corresponding two-electron oxygen reduction performance of the carbon-based electrocatalyst. The parameters of the carbon-based electrocatalyst include whether it is pure carbon, the source of heteroelements, specific surface area, carbon element content, oxygen element content, nitrogen element content, I D / I G 、d002 layer angle; the preparation conditions include temperature and time; the two-electron oxygen reduction performance collection module is used to collect the H 2 O 2 selectivity and current density of the two-electron oxygen reduction reaction. A total of 12 features were collected in this example.
[0054] S2: Use the data preprocessing module to convert the classification data collected in S1 into numerical data and perform normalization processing. Specifically: convert the classification data into numerical data using boolean values.
[0055] S3: Import the data preprocessed in S2 into four machine learning models commonly used to handle regression problems, namely Extreme Gradient Boosting (XGBoost), Random Forest (RF), Gradient Boosting Decision Tree (GBDT), and Support Vector Regression (SVR), and use the grid search method to adjust the hyperparameters of the above models and train the above machine learning models. The accuracies of the four models are as Figure 2 and Figure 3 shown.
[0056] From Figure 2 and Figure 3 it can be seen that the machine learning model with the best prediction performance for H2O2 selectivity and current density is the XGBoost model. Therefore, the two trained XGBoost models are combined to recommend the best carbon-based electrocatalyst and preparation conditions.
[0057] This example uses BCN-1100, CNB-ZIL 8, N,O-CNS0.2, o-CQD-4, and meso-PC as examples. The verification results are as Figure 4 shown. BCN-1100 is pure carbon, the source of heteroelements is endogenous, the specific surface area is 860.8m 2 / g, the carbon element content is 78.91%, the oxygen element content is 1.93%, the nitrogen element content is 8.38%, ID / IG is 1.03, and the d002 layer angle is 27°; the corresponding preparation conditions are 1100°C and 2 hours; the corresponding H 2O 2 The selectivity is 82.3%, and the current density is -2.1 mA / cm -2 . CNB-ZIL 8 is not pure carbon, the source of heteroelements is endogenous, the specific surface area is 525 m 2 / g, the carbon element content is 27%, the oxygen element content is 27.2%, the nitrogen element content is 24.6%, ID / IG is 1.35, and the d002 layer angle is 26.6°; the corresponding preparation conditions are 800 °C and 2 hours; the corresponding H 2 O 2 The selectivity is 81.6%, and the current density is -1.8 mA / cm -2 . N,O-CNS0.2 is pure carbon, the source of heteroelements is endogenous, the specific surface area is 583 m 2 / g, the carbon element content is 94.51%, the oxygen element content is 9.58%, the nitrogen element content is 5.91%, ID / IG is 2.33, and the d002 layer angle is 22°; the corresponding preparation conditions are 800 °C and 5 hours; the corresponding H 2 O 2 The selectivity is 87%, and the current density is -1.9 mA / cm -2 . o-CQD-4 is not pure carbon, the source of heteroelements is endogenous, the specific surface area is 570 m 2 / g, the carbon element content is 73.56%, the oxygen element content is 26.44%, the nitrogen element content is 0%, ID / IG is 0.6, and the d002 layer angle is 25°; the corresponding preparation conditions are 180 °C and 12 hours; the corresponding H 2 O 2 The selectivity is 76%, and the current density is -0.6 mA / cm -2 . meso-PC is not pure carbon, the source of heteroelements is endogenous, the specific surface area is 594 m 2 / g, the carbon element content is 73.17%, the oxygen element content is 24.6%, the nitrogen element content is 1.92%, ID / IG is 0.76, and the d002 layer angle is 24°; the corresponding preparation conditions are 900 °C and 2 hours; the corresponding H 2 O 2 The selectivity is 87%, and the current density is -1.6 mA / cm -2 . As Figure 4 shown, the gap between the results predicted by machine learning and the actual results is small, showing good performance.
[0058] This embodiment also provides a system for predicting the performance of a two-electron oxygen reduction carbon-based electrocatalyst based on machine learning, including a data acquisition module, a data preprocessing module, a machine learning module, and an optimization feedback module. The data acquisition module, the data preprocessing module, the machine learning module, and the optimization feedback module all adopt mainstream microprocessors in the prior art. The data acquisition module is used to collect the parameters of the carbon-based electrocatalyst, the preparation conditions, and the performance of two-electron oxygen reduction; the data preprocessing module is used to convert categorical data into numerical data and normalize the numerical data.
[0059] The machine learning module and the optimization feedback module are used to determine the optimal machine learning model for each data application scenario according to the evaluation index of the machine learning model, and optimize under the optimal machine learning model to find the best carbon-based electrocatalyst and preparation conditions.
[0060] The data acquisition module includes collecting the parameters of the carbon-based electrocatalyst, the preparation conditions, and the corresponding performance of two-electron oxygen reduction. The parameters of the carbon-based electrocatalyst include whether it is pure carbon, the source of heteroelements, specific surface area, carbon element content, oxygen element content, nitrogen element content, ID / IG, d002 layer angle; the preparation conditions include temperature and time; the two-electron oxygen reduction performance collection module is used to collect the H 2 O 2 selectivity and current density of the two-electron oxygen reduction reaction.
[0061] The above description of the embodiments is to enable those of ordinary skill in the art to understand and use the invention. It is obvious that those skilled in the art can easily make various modifications to these embodiments and apply the general principles described herein to other embodiments without creative labor. Therefore, the present invention is not limited to the above embodiments, and the improvements and modifications made by those skilled in the art without departing from the scope of the present invention should be within the protection scope of the present invention.
Claims
1. A method for predicting the performance of a two-electron oxygen reduction carbon-based electrocatalyst based on machine learning, characterized in that: The following steps are involved: S1: Collect the property parameters, preparation conditions and corresponding performance data of two-electron oxygen reduction of carbon-based electrocatalysts; S2: convert the data collected in S1 into numerical data and perform normalization; S3: Import the processed data in S2 into the machine learning model, adjust the hyperparameters of the machine learning model and then perform training; S4: According to the accuracy of the machine learning model in S3, the algorithm evaluation index is used to determine the optimal machine learning model for each data application scenario, and then optimize under the optimal machine learning model to find the best carbon-based electrocatalyst and preparation conditions.
2. A method for predicting the performance of a two-electron oxygen reduction carbon-based electrocatalyst based on machine learning according to claim 1, characterized in that: The data collected in S1 includes classification data, and the classification data includes whether the catalyst is pure carbon and the source of impurities.
3. The method for predicting the performance of a two-electron oxygen reduction carbon-based electrocatalyst based on machine learning according to claim 2, characterized in that: Convert categorical data to numerical data using Boolean values.
4. The method for predicting the performance of a two-electron oxygen reduction carbon-based electrocatalyst based on machine learning according to claim 1, characterized in that: In S2, the normalization processing method is to normalize all numerical data to a similar order of magnitude by taking a logarithm.
5. The method of predicting the performance of a two-electron oxygen reduction carbon-based electrocatalyst based on machine learning according to claim 1, characterized in that: In S3, the machine learning models include XGBoost, random forest, gradient boosted tree, and support vector regression.
6. The method of predicting the performance of a two-electron oxygen reduction carbon-based electrocatalyst based on machine learning according to claim 1, characterized in that: In S3, the grid search method is used to adjust the hyperparameters of the machine learning model.
7. The method of predicting the performance of a two-electron oxygen reduction carbon-based electrocatalyst based on machine learning according to claim 1, characterized in that: In S4, the algorithm evaluation index includes R 2 , RMSE, MAE.
8. A system based on the method according to any one of claims 1 to 7, characterized in that: It includes a data acquisition module, a data preprocessing module, a machine learning module and an optimization feedback module; the data acquisition module is used to collect the property parameters, preparation conditions and performance data of two-electron oxygen reduction of the carbon-based electrocatalyst, and the collected data is transmitted to the data preprocessing module; The data preprocessing module is used to convert the classified data into numerical data, normalize the numerical data, and then transmit it to the machine learning module; the machine learning module and the optimization feedback module are used to determine the optimal machine learning model for each data application scenario based on the accuracy of the machine learning model and use algorithm evaluation indicators, and optimize under the optimal machine learning model to find the best carbon-based electrocatalyst and preparation conditions.
9. The system according to claim 8, characterized in that The data acquisition module includes a carbon-based electrocatalyst parameter acquisition module, a preparation condition information collection module and a two-electron oxygen reduction performance analysis module; the carbon-based electrocatalyst parameter acquisition module is used to collect the source of the carbon-based electrocatalyst precursor and the carbon material structure; the preparation condition information collection module is used to collect the temperature and time for preparing the carbon-based catalyst; the two-electron oxygen reduction performance collection module is used to collect the H2O2 selectivity and current density of the two-electron oxygen reduction reaction.