Accelerating Finite Element Analysis Through Precise Matrix Preidiction Based on Deep Learning
Please login to view abstract download link
Calculating the inverse of a matrix is a complex problem, especially when engineering properties are involved. This study propose a precise AI prediction method that can maintain engineering properties almost perfectly during inverse matrix calculations. To address the accuracy issues faced by existing AI methods, we adopted an approach that divides the elements of the matrix into segments and predicts each part separately, as illustrated in Fig. 1(a). This approach overcomes the limitations of previous methods and achieves higher accuracy. The proposed method has been applied to real engineering problems, demonstrating not only its accuracy but also its time efficiency, as shown in Fig. 1(b).
