近年來隨著高科技的發展、電力設備及電力電子元件廣泛被使用在工業上，使得諧波問題日益嚴重，而導致元件或設備誤動作情況發生。改善諧波問題之方法，通常是加裝濾波器，濾波器大致可區分主動式與被動式兩類，由於主動式濾波器造價昂貴，構造及維修較為複雜，就經濟考量方面，較不符合成本考量，而被動式濾波器具有結構簡單、維修方便和造價便宜等優點，因而廣泛被採用。本文提出利用粒子群演算法來設計被動式濾波器，根據台灣電力公司系統諧波管制標準(參照IEEE-519諧波管制標準)、系統頻率誤差、電容器製造與溫度誤差及電感器製造誤差，作為抑制諧波限制條件，以防止串聯共振發生，而導致燒毀濾波器情況發生。另外成本目標函數亦考慮濾波器成本及使用年限等情況，以更能貼近實際之應用。最後，本文以一低壓配電系統為例，加裝5次、7次與11次濾波器，並對蟻拓和免疫演算法比較，以驗證本文所提方法之效能。 Owing to the development of high technology, power device and power electronic device are widely used in the industry, which deteriorates the problem of harmonic and results in wrong operation of equipments and devices. The most common way to reduce harmonic distortion is to install the harmonic filters. Filters are divided into both active and passive types. Active filter does not meet the requirement of low cost consideration due to the high cost and complex construction and maintenance. Passive filter, however, because of its simple construction, convenient maintenance and cheaper cost, is widely adopted. This thesis utilizes the PSO algorithm to design the passive filter for the harmonic problems. According to the harmonic standard of Taiwan Power Company, system variation, manufacturing tolerance and temperature tolerance of capacitor and manufacturing tolerance of inductance are used to curb harmonic limit condition and to prevent series resonance, which cause the burning down of filters. This context also considers the cost and lifetime of filters as the object function. This study takes a low voltage distribution system as an example, using Matlab/Simulink software simulation to mount 5th, 7th and 11th of filters and get the results. In addition, this thesis compares the ant algorithm and immune algorithm and use optimal techniques to make the results perfect and practical, which prove the effective functions come up with by this context.