- 作者: 楊百川; 楊宏澤; 黃慶連
- 作者服務機構: 成功大學電機工程學系
- 中文摘要: 本文主要在應用進化規劃法(Evolutionary Programming)求解火力機組調派的問題,在滿足負載需求與運轉限制條件下,規劃一天至一週內系統應上機的機組,以獲得最低之系統總發電成本。各機組上機或不上機的狀況事先編碼成0與1的組合,同時將最小上機停機時間的限制隱藏於所設計的二次碼內,解決了傳統上對該兩類限制式處理上的困擾。進化規劃法在尋優的過程中,係同時對一組解而非單一點進行搜尋,且直接利用其目標函數,因此無論發電成本函數是否連續可微,皆能適用;又其由某一解尋優至另一解採用隨機的方式,而非確定性之移動規則,因此基本上可找到最佳或接近最佳解。測試於台電38台火力機組,24小時規劃期的結果顯示,使用進化規劃法可得到接近最佳解且快速收斂。
- 英文摘要: This paper describes the application of the evolutionary programming (EP) algorithm to solve theoptimization problem of thermal unit commitment(UC)The EP algorithm tends to generate new creaturesof increasing fitness over time, and performs a global search over the solution space. The proposed constrainthandling techniques are integrated in the EP searching model to efficiently deal with diverse constraintsof the UC problem. Numerical results tested on the practical Taiwan Power(Taipower)system of 38thermal units over a 24-hour period show that the features of easy implementation, fast convergence, andnear-optimal solution of the UC problem can be achieved by EP. Results of comparison with the relatedstochastic search methods, simulated annealing and genetic algorithm methods, and the Lagrangianrelaxation method are also presented
- 中文關鍵字: unit commitment; mixed interger programming; evolutionary programming; stochastic search
- 英文關鍵字: --