Research on soft sensing model via FCM-based distributed ANFIS and its application

Originated from the idea of combining several models to improve prediction accuracy and robustness, a new method for nonlinear soft sensing modeling was proposed. Fuzzy c-means clustering (FCM) algorithm was adopted to separate a whole training data set into several subsets with different centers, each subset was trained by adaptive neural-fuzzy inference system (ANFIS). Subsets outputs were integrated by fuzzy cluster so as to obtain the final result. This model has been evaluated and applied to loose of jig bed. The simulation and practical application demonstrate that this model has good generalization result, good prediction accuracy and wide potential application online.