當電力系統遭受重大的事故或發電機跳脫時,會造成系統頻率下降,甚至系統崩潰。此時需要使用適當的卸載策略,卸除部分的負載,避免事故擴大及讓系統恢復平衡狀態。本論文藉由單機-無限匯流排之簡化系統來模擬系統低頻響應,利用基因法則找出最佳低頻電驛之低頻卸載段數、各段卸載頻率及各段卸載率。 本論文針對2007年台電系統做研究,以十二個月份分別加以規劃,探討各種不同跳機事故。利用本論文提出的方法與現有台電卸載策略做模擬比較,在各電驛卸載率可任意設定、每月份可修改低頻電驛設定以及限制條件較為寬鬆的假設下,由模擬結果可知,應用基因法則之方法可在較少的卸載量下,仍維持系統穩定運轉。
The system frequency will be declined rapidly and even system collapsed due to huge contingencies or generator trippings. The suitable shedding strategy is needed to use, therefore to avoid the fault areas expansion and recover the system by shedding the part of load. This thesis analysis the system underfrequency responses for simplified system of single machine infinite bus, and proposes genetic algorithm to do the setting of optimal underfrequency load shedding, each section of shedding frequency and shedding rate. This thesis aims to plan the setting of 12 months separately within various contingency on Taipower 2007’s system. The simulation results of comparison between the method proposed by the thesis and Taipower company. It assumed that each random relay shedding rate could be set, the underfrequency relay setting could be modified every month, and under a smaller constraints. From the simulation results, GA method could be kept system stable with the few load-shedding quantity.