A DEEP LEARNING-BASED ALGORITHM FOR HYBRID APPROACH TO SOLUTION OF MULTIPHYSICS PROBLEMS

The solution of Multiphysics problems, which involve the simultaneous modelling of multiple physical phenomena, is a challenging task due to the complexity and nonlinearity of the underlying equations. In recent years, deep learning techniques have shown great potential in various scientific and engineering applications. This paper proposes a deep learning-based algorithm for a hybrid approach to the solution of Multiphysics problems. The hybrid approach combines the strengths of traditional numerical methods and deep learning algorithms to achieve accurate and efficient solutions. The algorithm leverages the ability of deep learning models to learn complex patterns and relationships from data, while also utilizing the well-established numerical techniques for handling specific physics equations. The proposed algorithm consists of two main steps. In the first step, a deep learning model is trained using available data from the Multiphysics problem. This model is capable of capturing the intricate relationships between different physical quantities and generating approximations of the solution. In the second step, the deep learning model is integrated with traditional numerical methods to refine and improve the solution. The effectiveness of the proposed algorithm is demonstrated through several case studies involving different Multiphysics problems, including fluid-structure interaction, heat transfer, and electromagnetics. The results obtained from the hybrid approach show significant improvements in accuracy and computational efficiency compared to traditional numerical methods alone. The deep learning-based algorithm presented in this paper offers a promising solution to the challenges associated with solving Multiphysics problems. By combining the strengths of deep learning and traditional numerical methods, it provides a powerful framework for accurate and efficient simulations of complex physical systems. This approach has the potential to advance research and development in various fields, including aerospace, automotive, and renewable energy, where Multiphysics problems are prevalent.