Harmonic filter design using soft computing methods with design centering

Harmonic filter design has always been problematic due to inconsistent design parameters, power system uncertainty and differences in design techniques and application. Filter elements, especially the capacitor, are easily stressed and therefore fail. Load characteristics are constantly changing. Power electronic apparatus tends to inject varying amounts of harmonic current into the power system. A number of computer aided design and engineering methods have been employed to design passive element harmonic filters. However not all methods offer a viable solution and no method is ideal.; This dissertation explores fuzzy modeling in power system design specifically directed toward harmonic filter implementation. Fuzzy system inference is used in conjunction with design centering methodologies to develop a new technique which considerably refines the modeling of uncertainties and imprecision inherent in power system information. The fuzzy centered design tool solution adequately handles an important subset of inconsistencies and uncertainties in power systems. The technique designs a solution that will accept the four criteria used for comparison and still maintain performance corresponding to IEEE standards. We demonstrate that a design centered solution can be achieved through soft computing methods such as fuzzy inference systems. Moreover, the design method presented here increases the robustness of the filter design and out performs current design methods.