Neural Network Analysis Of The Mechanical Behavior Of Plate Rolling Mill

According to the facts of medium plate mill at Wu-Yang, a mathematical model of rolling process is founded using the technology of neural network. Some developments are made in the modeling:(1) Presenting a new mathematical model of medium plate rolling process according to neural networks, which was successfully used in sheet mill.(2) Developing the application field of neural networks, from rolling force to temperature, width and metal flow stress.(3) The improvement is made of adopting LM(levenberg-marquardt) algorithm, which accelerates the studying process of the model. So, it is fit for real-time control.(4) This article uses Elman neural network and GRNN neural network to build model of medium plate rolling process, proves the feasibility.After the studying and testing of this system using data on the spot of medium plate mill, the fact is that the maximum error is below 10 percent .That is to say the model is proper to characterize .The process of medium plate rolling, and the accuracy is fit for engineering demands.