結合可行方向法與基因演算法於濾波器之最佳規劃

Applying the Combined Feasible Direction Method and a Genetic Algorithm to Optimal Planning of Harmonic Filters Considering Uncertainty Conditions

本文提出結合可行方向法與基因演算法，探討大容量被動式諧波濾波器之最佳規劃問題。在豐富諧波電流源情況下，所提方法可獲得濾波器之最佳方案，並可避免系統諧波放大問題。濾波器最佳規劃是以電壓總諧波失真與電流諧波總需量失真為最小，並考慮濾波器損失、虛功率補償與各次諧波限制條件。為避免濾波器引發並聯共振產生諧波放大現象，於調諧點以下之諧波電流限制條件為管制標準1/3。為驗證所提方法之有效性，實際應用在某一負載為18 脈波整流器化工廠，並比較三種諧波濾波器設計方案。最後，以目標函數期望值及標準偏差反映濾波器參數失調、負載不確定性與系統阻抗變動的效應。

This paper presents an approach combining the feasible directionmethod and a genetic algorithm (FDM+GA) to investigate the planning oflarge-scale passive harmonic filters. The optimal filter scheme can beobtained for a system with abundant harmonic current sources whereharmonic amplification problems should be avoided. The objective is to minimize the total demand distortion of harmonic currents and totalharmonic distortion of load bus voltages. Filters loss, reactive powercompensation, and constraints of individual harmonics can be considered inthe design procedures. The constraints of harmonics with orders lower thanthe filter tuned-points have been set stricter, by one third, to avoidamplifying non-characteristic harmonics. The searching for an optimalsolution has been applied to the harmonic problems in a chemical plant,where three 6-pulse rectifiers are used. Three design schemes are comparedto demonstrate the performance of the proposed method. Finally,expectations and standard deviations of objective functions are used topresent effects of filter parameter detuning, loading uncertainty, andchanging of system impedance.