Estimating Low Frequency Oscillation using Bacterial Swarm Algorithm with Local Probability Likelihood Approach

This paper proposes a Low Frequency Oscillation (LFO) parameters estimation scheme based on Bacterial Swarm Algorithm (BSA). LFO is caused by a wide variety of events, including system faults and load switching. Thus, it is an important task to accurately estimate the parameters of LFO, and further perform fault diagnosis. Although the techniques such as Prony’s method could reconstruct the form of the signal, the estimated parameters are not accurate enough. In order to improve the estimation accuracy, this research improves the regression objective function, which aims to minimize the probability likelihood between a segment of the signal filtered by a Mathematica Morphology (MM) filter and its regression. In the experimental studies, BSA is used to optimize the proposed objective function. Comprehensive comparisons are taken among the proposed method, other Evolutionary Algorithms (EAs), and conventional signal processing techniques, which show BSA with Local Probability likelihood (BSA-LP) has the best performance on the estimation.