Prediction And Research Of Sheet Metal’s Bending Springback Based On Neural Network

Springback problems have always existed in bending forming process. This phenomenon is obviously affect precision of parts and manufacture efficiency. So it is necessary to do further research to control the springback. In order to multiple processing and repeated tests to reduce or eliminate the influence of springback, experiences usually rely on the operator’s experience during the production. But since the beginning of 20th, springback problem has daily risen to an important research project.Leading artificial intelligence technology and method into bending springback prediction is a study hotspot at present. Artificial neural network(ANN) is one kind of methods which stood up by imitating human brain nerve delivering information. It is one kind of distributed parallel processing system, the acquired results save in the matrix with weight values distributed. By the network, an optional nonlinear input-output mapping relationship can be realized. Due to the strong self-adaptability and self-learning-ability as well as excellent and ribosomes and tolerance ability, it can not only replace many traditional algorithm which is very complicated and time consumption. Because of these advantages, in this paper, the author using artificial intelligence techniques and methods predict and research the springback. The author combined BP neural network and Genetic Algorithms (GA), Particle Swarm Optimization (PSO) to predict the springback and then compared and analyzed the predicted results. Finally, the author do experiment to demonstrate the accuracy and reliability of the model established in this paper.