Summary: | 博士 === 國立臺灣科技大學 === 電機工程技術研究所 === 86 === The optimal dispatch of real/reactive power in a deregulated power system over a short period of time is investigated in this dissertation. The generation plants, transmission system and distribution systems will be separately owned under this condition. The consideration of profits of power producers, system- wide benefits and system security to determine the optimal dispatch pattern then becomes a significant task of the independent system operator (ISO). In this dissertation, the sensitivity factors are used to solve the optimal dispatch of real/reactive power over a short period of time. The traditional optimal power flow shows high accuracy, but it takes much computation time. Therefore, it cannot be applied to real-time operation, not to mention dynamically over the whole day scheduling period. In this dissertation, the artificial neural network is used for load forecasting and a two-step dynamic real power dispatch method considering inconforming load changes is developed to solve real power dispatch, and then, the sensitivity factors are used to solve reactive power dispatch with constraints. Finally, the dynamic programming technique is used to minimize control actions and to ensure bus voltagemagnitudes within the secure range over the scheduling time period. For wheeling, if there is any line flow beyond the secure range in the vested generation contracts, generation shifting factors are employed to move the excess line flow to other transmission lines to relieve the transmission congestion. Then, the reactive power/voltage sensitivity factors and dynamic programming technique are used to solve the reactive power dispatch with constraints. The proposed approach is applied to Taipower system. The results show that this methodology can minimize thecontrol actions in the whole day. It is simple and canenhance the operators'' planning greatly and can save a lot of switching cost.
|