- 作者: 劉承宗; 葉增雄
- 作者服務機構: 國立中山大學電機工程研究所
- 中文摘要: 為獲得詳盡之電力系統動態特性分析結果,提高系統元件之數學描述模型的複雜度將是必需的,然而‧模擬求解之計算負荷將因此變得極為沈重。為加速求解時間並保持足夠之精確度,應用多處理器平行計算技巧是一頗受矚目的趨勢。本文探討兩種平行預測-修正法,名為多步距預測-修正演算法(Parallel Multistep Predictor-Corrector Algorithm,MPC)及時間區塊預測-修正演算法(Parallel Block Predictor-Corrector Algorithm, BPC),在分別含二及四個處理器之兩部計算機上實現電力系統動態模擬之平行處理。IEEE Second Benchmark Model System #2及一含汽電共生機組之工業電力系統被選為數值模擬例證,包含串列及平行演算法之平行計算效能及其間之交叉比較與討論均於文中列舉。經由此一例證,將可提供多處理器平行處理於電力系統分析上的一些極有價值之參考導引。
- 英文摘要: To provide detailed power system dynamic analysis, more rigorous system representations aregenerally required. Consequently, the computational burden becomes heavy. To speedup the solutiontime while maintaining the required accuracy, an attractive method is to implement parallel computationtechniques on the multiprocessor architecture. This paper investigates the application of two parallelpredictor-corrector algorithms, namely the multistep predictor-corrector algorithm (MPC) and the blockpredictor-corrector algorithm (BPC), on a generalized power system dynamic simulation program runningon both the 2-CPU Alliant FX/80 and the 4-CPU Convex C3840 systems. The IEEE Second BenchmarkModel System #2 and an industrial cogeneration system are selected for detailed numerical verification,and cross comparisons and discussions also provided. From these illustrations, valuable guidance isobtained for the associated multiprocessor parallel processing schemes applied to power system transientcharacteristic analyses.
- 中文關鍵字: parallel processing; power system dynamics; parallel predictor-corrector algorithm
- 英文關鍵字: --