Impact of Fuzzy and Neural Network Techniques in Induction Motor Load Modeling and Voltage Stability Analysis

Abstract

Dynamic load models, as well as the phenomena of voltage instability, are of growing importance to the studies of power system dynamics. If the load representation is not of sufficient accuracy, the simulation results obtained using such models do not correspond to the actual response of the load. This affects the assessment of power system stability limit. This thesis deals with modeling and description of the dynamic load characteristics for long-term voltage stability studies. The main component of the dynamic loads constitutes the induction motor loads. The research comprised of developing a neurofuzzy induction motor load model which was then incorporated in a Newton-Raphson power flow algorithm. The modified load flow incorporates a neuro-fuzzy induction motor load model that estimates the scheduled active and reactive power adjustable during each iteration process. The impact of the aggregated induction motor on the convergence characteristics of load flow are investigated using the IEEE-30 bus standard test system. Based on the load flow results, counter propagative artificial neural networks were used to classify buses in order of weakness. In this thesis, the neuro-fuzzy model of an induction motor load was extended into the Continuation Load flow algorithm, where the impact of different induction motor horse power ratings on the voltage stability margin was investigated. The thesis finally investigates the impact of the Fuzzy Logic based Power System Stabilizer (FLPSS) in maintaining short term voltage stability in a system with Induction Motor Loads. xix