The Study Of Modeling Method For Products Quality Fuzzy Model In Rolled Steel Production Proess

Rolling industrial production process is the process by a number of subsystems consisting of complex systems, with multi-parameter, nonlinear and uncertainties and other characteristics. To control the quality of the rolling industrial production process, it is necessary to establish quality model for its production process. As the rolling industrial production process have the complexity, it is very difficult to establish an appropriate quality model. In recent years, many domestic and foreign scholars has been carried out the study of qualitative modeling by using different methods and have achieved some results.But there is a certain gap between these quality model and actual production practice. In this paper, we researched steel rolling process fuzzy quality modeling method and obtained some meaningful results, the main work as follows:(1)Based on the analysis of rolling production process and its product quality modeling,we researched and analysised the factors of affecting product quality,and determined the main impact and key indicators of affecting product quality,finally pretreatment the rolling production process data.(2) Based on the analysis of the fuzzy neural network and its modeling method,a modeling method based on real constant value consequent fuzzy model was given. We improved its former parts and pieces of the structure, parameter identification method etc,and enhanced the efficiency of learning and training model.The simulation shows the validity and reliability of modeling method.(3) For the problem of "fuzzy rule explosion",an effective way is proposed to establish fuzzy quality model for rolling produce process based on input layer of rolling over the production process quality model for hierarchical fuzzy neural network modeling.And a hierarchical structure of fuzzy neural network model was given.Finally we established a rolling complex model of product quality by combining this hierarchical model and the real constant value consequent fuzzy model.The simulation with real rolling data shows the validity and reliability of modeling method.