An inferential estimation strategy of Texaco gasification furnace temperature based on primary component analysis and new radial basis function networks is proposed. Firstly

the process prior knowledge and primary component analysis are used to simplify the networks input vector and to choose the secondary variable. Then a new recursive hybrid algorithm of radial basis function networks is developed. The algorithm includes the simplified rival penalized competitive learning method to make an adaptive clustering of networks input pattern and the recursive least squares method with forgetting factor to update the networks weights. The results show that this inferential estimator possesses features of fast tracking speed and high estimation accuracy

with performance much better than that of inferential estimators based on conventional radial basis function networks or ordinary backpropagation networks.