Research On Intelligent Approach Of Flatness Pattern Recognition And Control

In order to meet the high demands of the cold rolling system, improving the flatness quality increasingly has become one of the most important questions that are urgent to be solved. Therefore, flatness control has become the key to the steel system. Flatness pattern recognition is the precondition and the important part of the flatness control system as well as the hot topic. Recently the artificial intelligence theory which has a good function on model setting, optimization, controlling was widely used in the field of flatness pattern recognition and flatness control system. An analysis has been made on the current situation both in home and abroad, the problems of the previous artificial intelligence methods have been shown in this paper and a study has been made on approaches to flatness pattern recognition and control respectively which were proposed in the paper.Above all, problems of the traditional model known as neural network with slow convergence and local minimum etc have been analyzed. The radial basis function (RBF) network based on support vector machines (SVM) model has been applied in the flatness pattern recognition system so that an effective model to flatness pattern recognition has been established.Secondly, the problem of input redundancy has been found. The fuzzy distances to every two typical patterns were calculated respectively. The deduction of the fuzzy distances was regarded as the input of RBF network, which has got the numbers of inputs declined by a half and made the structure of RBF network simpler. The approach has been proved with high precision and speed. It could also be put into other fields of pattern recognition in which reciprocal polynomials for every two typical patterns are existed.Moreover, the weakness of previous flatness control method has beenanalyzed and the dynamic matrix method of flatness control was presented. The real-time feature has been considered and the predicting model of flatness control was established to adjust the effective matrix and provide the flatness control with accurate information.Finally, the improved RBF network flatness pattern recognition based on SVM method and the intelligent control based on dynamic control matrix has been proved with a good correctness and feasibilities through the experimental results.