Research on AFT of Unstructured Surface Mesh Based on ANN

Compared with other algorithms, Advance font method can better approximate the geometric surface features of complex models and has better universality, but the algorithm itself contains a large number of intersection judgments, which directly affects the efficiency of the algorithm. This paper proposes to apply neural networks to the Advance font method, and use the neural network to automatically calculate new surface elements instead of search, so as to reduce the number of front intersection judgments in the algorithm process and improve the efficiency of the algorithm. Compared with the traditional Advance font method, the new algorithm can reduce the intersection judgment by 10%~20% without reducing the quality of the grid element, and the efficiency of the algorithm is also improved by 9%~18%.