Indirect Adaptive Neurofuzzy Estimation of Nonlinear Time Series
Zhijie Wu, Chris J. Harris · ePrints Soton (University of Southampton) · 1996
Some classes of nonlinear systems or time series can be represented by an operating point dependent ARMA model. In this paper a neurofuzzy network structure is configured to identify such a model and the neural network is trained by the normalized back-propagation algorithm. The identified model is then converted to its equivalent state-space representation. Using this state-space form, a Kalman filter can be applied to estimate the state indirectly. A simulated example is given.