Virtual Design And Research Of Automobile Windshield Mold Surface

The traditional design method of automobile windshield mold is trial and error. This design method is featured by a long cycle and high cost and requires multiple corrections, which will influence the development cycle of the automobile. In this connection, this method has become increasingly incapable of meeting the fierce competitive environment of the automobile market and individual requirements. The design of automobile windshield mold has become an important factor affecting the development of the automobile industry at home and abroad. To solve this problem, this paper uses the virtual design method, namely the artificial intelligence method, to study the virtual design of automobile windshield mold.This paper has created the intelligent algorithm model of three layer BP neural network of automobile windshield mold using virtual design. Through neural network, the training to the input samples and the target output samples can be realized and the mapping relationships between input samples and the target output samples can also be obtained. The input samples are coordinates of the points on the surface of the automobile windshield mold. And the target output samples are coordinates of the points on the surface of the corresponding mold. The extractions of the training samples are realized by programming utilizing UG secondary development tool UG/Open GRIP. And then, UG/Open UIStyle, UG/Open API are utilized to write the interactive interface that can extract the training samples and generate the mold surface. In addition, on the basis of Levenberg-Marquardt neural network training algorithm, the paper adopts the momentum method and self-adaptive learning rate algorithm to optimize the weights and learning efficiency of neural network, which has improved the training reliability and shortened the learning time of neural network. Lastly, this paper has summarized the general steps of using artificial intelligence method to realize the virtual design of automobile windshield mold. It shows that this method is feasible through testing the training results of neural network, analyzing the simulation results and verifying the forming experiments of automobile windshield glass.This paper combines the artificial intelligence method with automobile windshield mold design for the first time, which can be a reference for the study of virtual design of automobile windshield mold in the future. It has significant theoretical value. At the same time, compared with the traditional method of trial and error, using artificial intelligence method to realize the virtual design of automobile windshield mold can shorten the design cycle, lower the cost and satisfy the accuracy requirements more easily, which has an important practical significance.