Physics-Informed Neural Network for Multicomponent Material Energy Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional artificial neural network methods face challenges in accurately predicting the potential energy of multicomponent materials, especially for organic compounds, due to limited prediction accuracy, low calculation efficiency, and inability to extrapolate energy for unlearned structures, which hampers simulations such as electrolyte decomposition reactions.
Innovation Solution
A physics-informed artificial neural network (PINN) method is employed, utilizing atom-centered symmetry functions to encode structural information, calculating bond order potentials, and optimizing neural network parameters to construct a potential energy surface, enabling efficient prediction of potential energy and force for multicomponent materials, including organic materials, with less learning data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional artificial neural network methods are used to predict potential energy of multicomponent materials, then calculation speed is improved, but prediction accuracy deteriorates
Solution Approach 1:
The patent combines conventional neural network methods with physics-based principles (bond order potentials, electron density constraints) to create a hybrid model. This composite approach integrates the speed of neural networks with the accuracy of physics-based methods, resolving the contradiction between calculation speed and prediction accuracy for multicomponent materials
Solution Approach 2:
The patent introduces additional parameters and constraints based on physical principles (electron density, bond order potentials) into the neural network framework. By changing the parameter space to include physics-based constraints, the model achieves both fast calculation and accurate prediction of potential energy
2Loss of time
If conventional artificial neural network methods are used with limited learning data, then training time is reduced, but prediction accuracy for unlearned structures deteriorates
Solution Approach 1:
The patent incorporates physics-based feedback mechanisms (electron density constraints, bond order potentials) that guide the neural network predictions even when structures are outside the training data. This feedback ensures predictions remain physically consistent, improving reliability without requiring extensive training data
Solution Approach 2:
The patent pre-encodes physical knowledge and constraints into the neural network architecture before training. By embedding physics principles in advance, the model can make accurate predictions for unlearned structures without requiring大量 training examples, reducing training time while maintaining extrapolation capability
3Loss of energy
If conventional artificial neural network methods are used for multicomponent materials, then computational cost is reduced, but ability to handle complex interatomic forces deteriorates
Solution Approach 1:
The patent segments the complex potential energy calculation into multiple components: neural network predictions for base energy, bond order potentials for specific interactions, and electron density constraints for electronic effects. This segmentation allows efficient handling of different physical phenomena while maintaining overall accuracy for multicomponent materials
Data Source
AI summary
The present disclosure provides a method for predicting a potential energy of a multicomponent material. The method includes encoding a structural information of a multicomponent material in which eigenvector values for each element are calculated using an atom-centered symmetry function (ACSF) of a structural information of atoms constituting the multicomponent material. The method also includes calculating a bond order potential (BOP) parameter for each element using the calculated eigenvector value as an input value and using a different artificial neural network model for each type of atom to construct a physics-informed artificial neural network (PINN) model. The method also includes calculating a bond order potential value for each element to construct a potential energy surface (PES). The method also includes performing structural optimization or physical property prediction of the multicomponent material using a potential energy surface.