A Multiple Neural Network Architecture Based on Fuzzy C-Means Clustering Algorithm
Jian Cheng, Yinan Guo, Jiansheng Qian · 2006
Inspired by the idea of integrating several models to improve prediction robustness and accuracy, a new approach of a multiple neural network (MNN) for nonlinear modeling is proposed. A whole training sample data set is separated into several clusters with different centers using fuzzy c-means clustering (FCM) algorithm, and each cluster is trained by adaptive neuro-fuzzy inference system (ANFIS) to constitute the sub-model respectively. The degrees of memberships are used for combining the outputs of subnets to obtain the final result, which are gained from the relationship of a new input sample data and clustering samples. The model has been evaluated and applied to estimate the status-of-loose of jig washer bed. The simulation and practical application demonstrate that the model has good generalization abilities, good prediction accuracy and wide potential application online