Fuzzy modeling and pattern recognition for dynamic processes and its application for an activated sludge process

Understanding process dynamic characteristics is very important to realize system dynamic control. The modeling method is one of the effective ways to characterize and understand a controlled process. However, dynamic characteristics for some industrial processes are too complex to be characterized by the traditional mathematical modeling method. In order to handle on-line recognition problems of some complex industrial process, a new modeling method with a self-learning function is proposed by using the theories of fuzzy sets and pattern recognition. Although subtle changes in processes are not the goal to be recognized by this method, general changing trends in processes will be understood using the proposed fuzzy recognition model, which is good to realize suboptimal process control. Modeling and simulation results for an activated sludge process are introduced.