An adaptive fuzzy model based process state identification for prediction and control

In this paper at first an integrated intelligent model for process state identification and behavior prediction for complex processes is introduced based on the results in (Tang and Koch, 2004). In the model, fuzzy neural networks (FNNs) are applied as process state classifiers for process state (fault) detection. Various neural networks (NNs) are used for system identification of process characteristics in different process states. The model detects process states and predicts process output according to process input variables and historical output. The whole model is constructed based on fuzzy TS dynamic nonlinear autoregressive with exogenous input (NARX) models. Secondly, two different model optimization schemes are investigated for model adaptability to cover time depending process changes. Thirdly, a specific state space equation of a discrete time varying system is being derived from the adaptive fuzzy model, based on this state space equation, corresponding process control methods can be used. Finally, an application case has been studied for products supply forecasting with this model. It indicated that the model has good performance and that it can be applied for process state (fault) detection, prediction and predictive control