Independent component analysis for harmonic source estimation from piecewise constant parameter mixed measurements

The presence of harmonic currents and voltages in power electronic equipment is responsible for major power quality problems. Recently, the Independent Component Analysis (ICA) has been used for harmonic source estimation. It estimates the harmonic sources in a power system assuming that the harmonic sources are statistically independent, non-Gaussian and that the measurements are linear combinations of the sources with constant weights. In a practical system, the mixing parameters are dependent on the elements and topology of the power grid and when the topology and element values change due to, for example, switching operations, the mixing parameters also change. In this paper, a modified algorithm using independent component analysis is developed for the estimation of harmonic sources with piecewise constant mixing models. The JADE algorithm is used to estimate harmonic sources in piecewise manner. Overlap segment method is used to order and scale the estimated mixing matrices. The voltage measurements are obtained at three buses in the standard IEEE 14-bus system. The change in network topology is identified. The effect of segment length on estimation algorithm is determined. Absolute square error (ASE) is used to measure the performance of the estimation algorithm. The simulation results for various segment lengths and network changes are presented.