A neural-network-based fuzzy system (NNFS) is proposed in thispaper. It is a self-organizing neura-network which can partition theinput spaces in a flexible way, based on the distribution of thetraining data in order to reduce the number of rules without any lossof modeling accuracy. Associated with the NNFS is a two-phase hybridlearning algorithm which utilizes a nearest-neighborhood clusteringscheme for both structure learning and initial parameters setting anda gradient descent method for fine tuning the parameters of the NNFS.By combining the above two methods, the learning speed converges muchfaster than the original back-propagation algorithm. Simulationresults suggest that the NNFS has merits of simple structure, fastlearning speed, few fuzzy logic rules and relatively high modelingaccuracy. Finally, the NNFS is applied to the construction of thesoft sensor for a distillation column.
展开▼