Method and device for predicting stability of porous material
By constructing a porous material stability prediction model using deep learning neural networks and utilizing SMILES feature data, the problems of low accuracy, high cost, and long time consumption in porous material stability prediction are solved, achieving fast and accurate stability prediction and improving the efficiency of porous material development.
Patent Information
- Application Number
- CN202511035897.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-25
- Publication Date
- 2025-11-11
AI Technical Summary
Existing technologies lack rapid and accurate methods for predicting the stability of porous materials, resulting in high costs and long processing times, which hinders the research and development and application of porous materials.
By reading, recognizing, and training the Molecular Linear Input Specification (SMILES), a stability prediction model is constructed using a deep learning neural network. Feature data of atoms, groups, branched groups, and rings are extracted to achieve rapid and accurate prediction of the stability of porous materials.
This enables rapid and accurate prediction of the stability of porous materials, reducing experimental costs and time, and improving the efficiency of porous material development.
Smart Images
Figure CN120932785A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of artificial intelligence technology, and specifically to a method and apparatus for predicting the stability of porous materials. Background Technology
[0002] Porous materials are a class of materials containing abundant micro- and nano-sized pores. Due to their abundant porosity, large specific surface area, and tunable structure, they possess excellent mass transfer properties and have important applications in separation, catalysis, and energy. However, the stability of porous materials during use has become a bottleneck restricting their research and application.
[0003] Therefore, to better study and develop porous materials, it is necessary to quickly and accurately predict the stability of porous materials during the adsorption process. However, existing technologies lack methods for accurately predicting stability using artificial intelligence, relying solely on characterization analysis after experimental synthesis, which is costly and time-consuming. Therefore, this invention proposes a method for predicting the stability of porous materials. This method achieves accurate and rapid stability prediction solely through the reading, recognition, and training of strings from the Simplified Molecular Input Line Entry System (SMILES), thus solving the technical problems of low accuracy, high cost, and long time consumption in traditional porous material stability prediction methods. Summary of the Invention
[0004] The purpose of this invention is to provide a method and apparatus for predicting the stability of porous materials, so as to at least partially solve the technical problems of low accuracy, high cost and long time consumption in the method for predicting the stability of porous materials.
[0005] To achieve the above objectives, the present invention provides the following technical solution:
[0006] This invention provides a method for predicting the stability of porous materials, the method comprising:
[0007] Obtain the molecular linear input specifications (SMILES) of repeating structural units of a porous material to be predicted, wherein the porous material to be predicted has a five-membered ring or a six-membered ring molecular structure;
[0008] Extract feature data of atoms, groups, branched groups, and rings from the molecular linear input specification (SMILES);
[0009] The extracted feature data is input into a pre-trained stability prediction model to obtain the stability prediction result of the porous material to be predicted.
[0010] The stability prediction model is obtained by training a pre-built deep learning neural network based on the molecular linear input specification (SMILES) of the repeated result unit samples of porous materials and the corresponding stability prediction results of porous materials.
[0011] In some embodiments, the stability prediction model is obtained by training a pre-built deep learning neural network, specifically including:
[0012] A massive amount of molecular linear input specifications (SMILES) and corresponding porous material stability prediction results were obtained from a database of repeating structural units of porous materials.
[0013] The massive molecular linear input specifications (SMILES) and the corresponding porous material stability prediction results were used to construct a sample set, which was then divided into a training set and a test set.
[0014] Extract the feature data of atoms, groups, branched groups, and rings from all molecular linear input specifications (SMILES) in the training set;
[0015] The extracted feature data of atoms, groups, branched groups and rings are input into a deep learning neural network. Multivariate nonlinear regression modeling is performed on the training set to obtain a multivariate nonlinear regression model. The multivariate nonlinear regression model is then trained to obtain a stability prediction model.
[0016] The stability prediction model is evaluated using the test set based on the evaluation index of the multivariate nonlinear regression model.
[0017] In some embodiments, feature data of atoms, groups, branched groups, and rings of all molecular linear input specifications (SMILES) in the training set are extracted, specifically including:
[0018] Extract the percentages of carbon, nitrogen, oxygen, hydrogen, chlorine, and fluorine atoms, the percentages of methyl, amino, hydroxyl, and carboxyl atoms, the percentage of branched groups, and the number of rings from the molecular linear input specification (SMILES).
[0019] In some embodiments, the evaluation metrics include mean squared error (MSE), mean absolute error (MAE), and adjusted coefficient of determination (Adjusted R²). 2 ).
[0020] In some embodiments, the stability prediction result is the difference in pore size of the porous material before and after adsorption, in angstroms.
[0021] The present invention also provides a device for predicting the stability of porous materials, the device comprising:
[0022] A molecular linear input specification reading unit is used to read the molecular linear input specifications (SMILES) of repeating structural units of a porous material to be predicted, wherein the porous material to be predicted has a five-membered ring or a six-membered ring molecular structure.
[0023] The feature data extraction unit is used to extract feature data of atoms, groups, branched groups and rings from the read molecular linear input specifications (SMILES);
[0024] The stability prediction result generation unit is used to input the extracted feature data into a pre-trained stability prediction model to obtain the stability prediction result of the porous material to be predicted.
[0025] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method described above.
[0026] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described above.
[0027] The advantages of the porous material stability prediction method and apparatus provided by this invention are that it can achieve rapid and accurate prediction of porous material stability solely through the reading and recognition of molecular linear input specification (SMILES) strings and training on feature data. When developing new porous materials, the stability of repeating structural units can be predicted first using the method and apparatus of this invention, thereby reducing the trial-and-error costs caused by experiments. This solves the technical problems of low accuracy, high cost, and long time consumption in porous material stability prediction methods. Attached Figure Description
[0028] To clearly illustrate the embodiments of the present invention, a brief description of the embodiments will be provided below. Obviously, the accompanying drawings described below are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without any creative effort.
[0029] The structures, proportions, sizes, etc. illustrated in this specification are only for the purpose of assisting those skilled in the art in understanding and reading the content disclosed herein, and are not intended to limit the conditions under which the present invention can be implemented. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in the proportions, or adjustments to the size, without affecting the effects and objectives that the present invention can produce, should still fall within the scope of the technical content disclosed in the present invention.
[0030] Figure 1One of the flowcharts for a method for predicting the stability of porous materials provided by the present invention;
[0031] Figure 2 A second flowchart of a method for predicting the stability of porous materials provided by the present invention;
[0032] Figure 3 One of the model training curves for the training and test sets;
[0033] Figure 4 The second training curve for the model on the training and test sets;
[0034] Figure 5 The third training curve for the model on the training and test sets;
[0035] Figure 6 A structural block diagram of a porous material stability prediction device provided by the present invention;
[0036] Figure 7 A structural block diagram of a computer device provided by the present invention. Detailed Implementation
[0037] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings and specific embodiments. Obviously, the embodiments described below are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0038] Please see Figure 1 , Figure 1 This is one of the flowcharts for a method to predict the stability of porous materials provided by the present invention.
[0039] In one specific embodiment, the method for predicting the stability of porous materials provided by the present invention includes the following steps:
[0040] S110: Obtain the molecular linear input specifications (SMILES) of repeating structural units of the porous material to be predicted, wherein the porous material to be predicted has a five-membered ring or a six-membered ring molecular structure.
[0041] S120: Extracts feature data of atoms, groups, branched groups, and rings from the molecular linear input specification (SMILES).
[0042] S130: Input the extracted feature data into the pre-trained stability prediction model to obtain the stability prediction result of the porous material to be predicted.
[0043] The stability prediction model is obtained by training a pre-built deep learning neural network based on the molecular linear input specification (SMILES) of the repeated result unit samples of porous materials and the corresponding stability prediction results of porous materials.
[0044] In some embodiments, the deep learning neural network algorithm used in the stability prediction model is a back propagation neural network, which is suitable for learning and solving complex nonlinear regression problems. The model only requires the input of molecular linear input specifications (SMILES) to predict stability results. The model is trained on feature data extracted from the read molecular linear input specifications (SMILES), and the model's prediction results show good consistency across various molecular linear input specifications (SMILES) characters, meaning the error between the model's prediction results and the actual results is very small.
[0045] In some embodiments, such as Figure 2 As shown, the stability prediction model is obtained by training a pre-constructed deep learning neural network based on the extracted feature data, specifically including:
[0046] S210: Obtain a massive amount of molecular linear input specifications (SMILES) and corresponding porous material stability prediction results from the porous material repeating structural unit database.
[0047] S220: The massive molecular linear input specifications (SMILES) and the corresponding porous material stability prediction results are used to construct a sample set, which is then divided into a training set and a test set.
[0048] S230: Extract feature data of atoms, groups, branched groups, and rings from all molecular linear input specifications (SMILES) in the training set.
[0049] S240: Input the extracted feature data of atoms, groups, branched groups and rings into the deep learning neural network, perform multivariate nonlinear regression modeling on the training set to obtain a multivariate nonlinear regression model, and train the multivariate nonlinear regression model to obtain a stability prediction model.
[0050] S250: The stability prediction model is evaluated using the test set based on the evaluation index of the multivariate nonlinear regression model.
[0051] In some embodiments, feature data of atoms, groups, branched groups, and rings of all molecular linear input specifications (SMILES) in the training set are extracted, specifically including:
[0052] Extract the percentages of carbon, nitrogen, oxygen, hydrogen, chlorine, and fluorine atoms, the percentages of methyl, amino, hydroxyl, and carboxyl atoms, the percentage of branched groups, and the number of rings from the molecular linear input specification (SMILES).
[0053] In some embodiments, the evaluation metrics include: mean squared error (MSE), mean absolute error (MAE), and adjusted coefficient of determination (Adjusted R²). 2 ).
[0054] In some embodiments, the stability prediction result can be the difference in pore size of the porous material before and after adsorption, in angstroms.
[0055] In a specific use case, a stability prediction model is first constructed using a back propagation neural network. The training samples contain 12 feature data read from all molecular linear input specifications (SMILES) of the training set, including: carbon atom fraction, nitrogen atom fraction, oxygen atom fraction, hydrogen atom fraction, chlorine atom fraction, fluorine atom fraction, methyl atom fraction, amino atom fraction, hydroxyl atom fraction, carboxyl atom fraction, branched group atom fraction, number of polycyclic benzene rings or aliphatic rings, and stability result data of molecular structures corresponding to all molecular linear input specifications (SMILES): pore size variation difference.
[0056] During model training, the sample set is randomly divided into a training set and a test set, with the training set accounting for 90% and the test set accounting for 10%. The 12 feature data corresponding to each molecule's linear input specifications (SMILES) and the pore size variation difference are normalized and then input into the stability prediction model. The model is trained using a deep learning algorithm. The test set is then input into the trained stability prediction model to evaluate it. Some of the stability prediction values obtained from the model evaluation are shown in Table 1. The stability prediction model is then encapsulated and loaded into the porous material stability prediction device provided by this invention.
[0057] The deep learning algorithm is used to train the model, specifically by inputting the training set into the model and performing training and evaluation using 10-fold cross-validation. This is done in 10 iterations, with each iteration dividing the training set into 10 equal parts: one part for validation and the remaining nine parts for training. The model training result curves for the training and validation data are shown below. Figure 3 , Figure 4 , Figure 5 As shown, where: Figure 3The curve showing the relationship between mean squared error (MSE) and the number of training iterations (Epoch) indicates that the model fitting error is reduced and there is no overfitting. Figure 4 The curve showing the relationship between mean absolute error (MAE) and the number of training iterations (Epoch) illustrates that the absolute error of the model prediction gradually decreases and tends to stabilize. Figure 5 To adjust the coefficient of determination (Adjusted R) 2 The curve showing the relationship between the model and the number of training iterations (Epochs) indicates that the model has a strong ability to explain the variation in the true values, the model's fitting effect on the training data continues to improve, and the model's generalization ability gradually increases.
[0058] The process of using the stability prediction model includes: encapsulating the trained model and evaluating it on the test set into a module, inputting the molecular linear input specification (SMILES) of the repeating structural unit of the porous material to be predicted, running the model, and outputting the predicted difference in pore size, which is the stability prediction result of the porous material to be predicted.
[0059] The stability assessment index is the difference in pore size, and its unit is _____. The reason for the difference in pore size (angiometer) is that after the adsorbate molecules enter the pores of the porous material, the material undergoes structural changes during the adsorption process due to swelling or contraction, resulting in unequal average pore sizes before and after adsorption.
[0060] Table 1. Predicted results of partial data on the stability assessment of porous materials using SMILES.
[0061]
[0062] The relative error refers to the percentage obtained by multiplying the absolute error caused by the aperture variation difference prediction to the true value of the aperture variation difference by 100%.
[0063] In addition to the methods described above, the present invention also provides a device for predicting the stability of porous materials, such as... Figure 6 As shown, the device includes:
[0064] The molecular linear input specification reading unit 101 is used to read the molecular linear input specifications (SMILES) of the repeating structural units of the porous material to be predicted, wherein the porous material to be predicted has a five-membered ring or a six-membered ring molecular structure.
[0065] Feature data extraction unit 102 is used to extract feature data of atoms, groups, branched groups and rings from the read molecular linear input specifications (SMILES);
[0066] The stability prediction result generation unit 103 is used to input the extracted feature data into a pre-trained stability prediction model to obtain the stability prediction result of the porous material to be predicted.
[0067] In one embodiment, the present invention also provides a computer device, which may be a server, and its internal structure diagram may be as follows: Figure 7 As shown, the computer device includes a processor, memory, and network interface connected via a system bus. The processor provides computing and control capabilities; the memory includes a storage medium and internal memory; the storage medium includes programs, an operating system, and a database. The storage medium can be memory, such as volatile or non-volatile memory, or both. The internal memory provides an environment for the operating system and programs running in the storage medium. The processor of the computer device can be an integrated circuit chip, a general-purpose processor, a digital signal processor, an application-specific integrated circuit (ASIC), or other logic device. The processor reads information from the storage medium and, in conjunction with its hardware, completes the steps of the above method embodiments. The database of the computer device stores static and dynamic information data. The network interface of the computer device is used for communication with external terminals via a network connection. When the program of the computer device is executed by the processor, it can implement the steps of the above method embodiments.
[0068] Those skilled in the art will understand that Figure 7 The structure shown is merely a partial structural block diagram related to the present invention and does not constitute a limitation on the computer device to which the present invention is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.
[0069] Corresponding to the above embodiments, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the method described above.
[0070] For those skilled in the art, based on the ideas of the embodiments of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.
Claims
1. A method for predicting the stability of porous materials, characterized in that, The method includes: Obtain the molecular linear input specifications (SMILES) of repeating structural units of a porous material to be predicted, wherein the porous material to be predicted has a five-membered ring or a six-membered ring molecular structure; Extract feature data of atoms, groups, branched groups, and rings from the molecular linear input specification (SMILES); The extracted feature data is input into a pre-trained stability prediction model to obtain the stability prediction result of the porous material to be predicted. The stability prediction model is obtained by training a pre-built deep learning neural network based on the molecular linear input specification (SMILES) of the repeated result unit samples of porous materials and the corresponding stability prediction results of porous materials. The stability prediction model is obtained by training a pre-constructed deep learning neural network, specifically including: A massive amount of molecular linear input specifications (SMILES) and corresponding porous material stability prediction results were obtained from a database of repeating structural units of porous materials. The massive molecular linear input specifications (SMILES) and the corresponding porous material stability prediction results were used to construct a sample set, which was then divided into a training set and a test set. Extract the feature data of atoms, groups, branched groups, and rings from all molecular linear input specifications (SMILES) in the training set; The extracted feature data of atoms, groups, branched groups and rings are input into a deep learning neural network. Multivariate nonlinear regression modeling is performed on the training set to obtain a multivariate nonlinear regression model. The multivariate nonlinear regression model is then trained to obtain a stability prediction model. The stability prediction model is evaluated using the test set based on the evaluation index of the multivariate nonlinear regression model. The evaluation metrics include mean squared error (MSE), mean absolute error (MAE), and adjusted coefficient of determination (R²). 2 ).
2. The method for predicting the stability of porous materials according to claim 1, characterized in that, Extract the feature data of atoms, groups, branched groups, and rings from all molecular linear input canonicals (SMILES) in the training set, specifically including: Extract the percentages of carbon, nitrogen, oxygen, hydrogen, chlorine, and fluorine atoms, the percentages of methyl, amino, hydroxyl, and carboxyl atoms, the percentage of branched groups, and the number of rings from the molecular linear input specification (SMILES).
3. The method for predicting the stability of porous materials according to claim 1, characterized in that, The stability prediction result is the difference in pore size of the porous material before and after adsorption, expressed in angstroms.
4. A device for predicting the stability of porous materials, characterized in that, The device includes: A molecular linear input specification reading unit is used to read the molecular linear input specifications (SMILES) of repeating structural units of a porous material to be predicted, wherein the porous material to be predicted has a five-membered ring or a six-membered ring molecular structure. The feature data extraction unit is used to extract feature data of atoms, groups, branched groups and rings from the read molecular linear input specifications (SMILES); The stability prediction result generation unit is used to input the extracted feature data into a pre-trained stability prediction model to obtain the stability prediction result of the porous material to be predicted.
5. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method as described in any one of claims 1 to 3.
6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 3.