Finite Element Analysis And Research On Bending Based On Artificial Neural Network

Bending is carried out by testing samples traditionally. There are several malpractice including high cost and long testing cycle because of so many times of tests. And they are limits to develop new products. Conceive of controlling the parameters by Artificial Neural Networks (ANN) is put forward for realizing flexible producing and intelligentizing the control of parameters.The Finite Element study of bending based on ANN is the typical multi-subjects field. It involves bending plane, building model of Finite Element Analysis, ANN and so on. The method for controlling the parameters makes the producing process simple greatly and saves the cost of producing. It also shorts the period of developing new products. It has great economic effects and good foreground. And in the same time the new field of applying new technology on the traditional producing is pioneered.Since the computer science is broadly used and the functions of the software become stronger, the technique of computer simulation and Computer-Aided Manufacturing are successfully applied in the control of parameters. And the simulations of Finite Element and ANN are studied. The function of the intelligent control of the parameters is realized.The basic theory of bending is introduced, especially the bending to make shapes at one time. In the analysis of metal bending, the Finite Element Method (FEM) is used. And the simulating software of ANSYS is used too. The stress field , the strain field (including the elastic and plastic stress) of the panel and the force and the cause of invalidation are particularly simulated in 3-D. The Finite Element model of bending panel is built. Furthermore , the model is proved to be effective and accurate.Secondly, the testing process is replaced by the Finite Element model to acquire the sample data. The Orthogonal Experimental Design is applied to get enough data information and simplify the process to handle the data.Thirdly, the elementary study on the ANN is carried on. BP neural network is focused on. The ANN model between the bending angle and technological parameters is built by using the technology and sample data which has been acquired. The processing result can be achieved quickly by inputting the processing parameters into the model. Then the forecasted result and the tested result are compared in order to offer quick direction to adjust the processing parameters.Finally, the application of the model is explained briefly. And the purpose and meaning are showed.The manpower and material resource waste and the high cost because of testing the samples to acquire the processing parameters several times is avoided by using the model. The efficiency of production is increased and the cost of production is saved. It makes great economic benefits and social benefits.