Recognition Of Flatness For Cold Strip Mill Based On BP Neural Network

Flatless, as the major index of rolling strips product, has caused serious attention of the most rolling automation working personnel. However, the present flatness theory is still in the period of experiment and research, flatness model distinguish techniques are unmature. Most of steel rolling industry in China is still in the level of open-loop control and manual adjustment.Recently, the research of Artificial Neural Network comes into the field of control. The new word of Neurocontrol springs up quickly. Neural Network has strong ability of self-study and adaptation. The satisfied result can be obtained by using Neural Network to control. This paper based on deep researching of cold-rolling Automatic Flatness Control system, in view of the complexity and uncertainty of rolling process,applied Neural Network into the Automatic Flatness Model Distinguish System.The main points of the discourse are as following:1. The summarize introduced the current status and the development trend of flatness control. And it discuss about the problem in distinguish of flatness defect.2. Analysis the flatness defect model. The main research is the construction of the flatness model.3. It introduced the BP Artificial Neural Network, and training the Artificial Neural Network with improved algorithm.4. The working principle of flatness control system including the acquisition of measurement and the generation of actuator command as well.To distinguish the flatness model with BP Artificial Neural Network, which could match the requirement of the highest precision and controlling on real time. Since it's very agility to construct the standard flatness model, if possible , it can distinguish any complex model of flatness defects.