當台電系統發生發電機跳脫或輸電線路故障,導致負載量大於發電量時,發電機為了增加出力會使得系統頻率急速下降。但每台發電機可容許頻率下降的程度有限,若頻率的即時恢復處理不當則會使發電機的使用壽命縮短,更嚴重還有可能導致整個系統解聯、崩潰。 為了避免上述情形發生,必須採用低頻電驛做為防護措施,依照低頻電驛的設定可藉由卸除負載以防止事故擴大、保全電力系統的穩定運作。但過少的卸載量會導致系統頻率回復速度過慢;過多不必要的卸載卻會增加用戶因停電而遭受的財產損失,因此,低頻卸載的規劃與執行就顯得相當重要。 本論文提出以免疫演算法應用於台電2007年尖峰與離峰系統下,發生嚴重發電機跳脫事故時之最佳化低頻負載卸載策略。免疫演算法以模仿生物基因演化的方式計算,並藉由記憶細胞保存每代演算結果之菁英解、抑制細胞篩選可行解的雜異度等功能,計算出適應值最佳化的最佳解,再利用每代計算出之結果配合電力系統模擬軟體PSS/E,模擬電力系統低頻響應,得到可將卸載的傷害程度減到最小的低頻卸載率。
When a power system experiences malfunctions, the power generator may trip and the transmission lines may break down, resulting in a load far exceeding the generated power. In such a situation, the generator’s attempt to increase its output can lead to a drastic drop in the system frequency. There is, however, a cap on the decline of frequency for each generator. Failure to facilitate instant recovery of frequency may reduce the generator’s life; the system may even collapse due to huge contingencies or generator tripping. Underfrequency relays are therefore used as protection mechanism against such accidents as they are capable of facilitating load shedding to contain accidents and keep the power system in stable operation. However, low shedding rates may reduce the speed of frequency recovery while excessive shedding can exacerbate the general public’s property loss during power outage. Optimal planning for and effective execution of underfrequency load shedding are thus an issue of crucial importance. The paper aims at using immune algorithm to develop optimal underfrequency load shedding strategy in the scenario of serious generator tripping for the 2007 Taipower systems. Imitating biological evolution, immune algorithm strives to reach the optimal solution by utilizing the function of memory cells to preserve the elite solutions of each generation and the function of suppressor cells to restrain the high affinity solutions that can increase diversity of solutions. Application of the immune algorithm is then coupled with the use of PSS/E power system simulation software to simulate the underfrequency responses of the power systems and to obtain the optimal underfrequency load shedding rate capable of minimizing the damages caused by load shedding.