Fuzzy Models for the Study of Hydro Power Plant Dynamics

Abstract

In this paper, the hydro power plant dynamics is identified using fuzzy models. The plant data is generated from Pade and H-infinity approximated first, second, third and fourth-order rational transfer function models. The models are simulated as (i) gate-servo motor position and turbine speed with random load disturbance and (ii) gate position and developed turbine power. Takagi-Sugeno fuzzy model structures are identified with smooth stepped wave signal input and the identified model is generalized on its validation data set and with random stepped wave signal as input. The fuzzy rules are extracted from data by means of Gustafson-Kessel clustering with antecedents determined using product-space and point-wise projection techniques