調和搜尋演算法(Harmony Search; HS)是將樂團演奏表現調整至最協調且美妙的現象引用到最佳化演算系統當中，而發展出一套全新的啟發式演算法。然而在最佳化問題的求解過程中，如何有效率的搜尋到目標函數的全域最佳解並不是一件容易的事，而調和搜尋演算法有不錯的搜尋精確度，但調和搜尋演算法由於微調方向依賴隨機化的方式求得，使得搜尋速度較慢，求解時間上消耗多，這對於決策者而言，是求解過程中的最大問題。本研究提出直交調和搜尋演算法(Orthogonal Harmony Search; OHS)，主要透過結構上重新修正並應用直交表實驗設計(Orthogonal Experimental Design; OED)之技術以改良調和搜尋演算法。直交調和搜尋演算法主要以三項機制運作搜尋：(1)以直交交配來做全域搜尋(exploration)，(2)以直交微調來做區域搜尋(exploitation)，(3)以隨機方式尋找其他可行解。因此，直交調和搜尋演算法承襲調和搜尋演算法的搜尋精確度，結合直交表實驗設計具有優良經驗的推理能力與主效果分析，促使演算法在廣大的搜尋空間中，能夠為子代判斷正確的搜尋區域與方向，更快速向最佳解逼近以達到收斂並強化搜尋最佳解的能力。本研究於直交表改良演算法的測試函數中，選出八個函數做實驗，實驗函數包含單一區域解與多重區域解問題。實驗結果除了與調和搜尋演算法比較之外，並與直交基因演算法(Orthogonal Genetic Algorithm; OGA)、直交模擬退火演算法(Orthogonal Simulated Annealing Algorithm; OSA)比較，整體結果直交調和搜尋演算法優於其他演算法，並證實本研究所提出之演算法在各種測試函數中能迅速且穩定向最佳解逼近。 Harmony search is a new heuristic algorithm, and it is conceptualized using the musical process of searching for a perfect state of harmony. How to search the global optimization solution efficiently in object function is difficult in the process of solving optimization problems. Harmony search have ability to find solutions closer to the optima but it consumes a lot of time. And it’s a big problem of HS for decision makers.This paper brings up an Orthogonal Harmony Search (OHS). The main focal point of OHS is to revise harmony search and apply method of orthogonal experimental design to enhance it. OHS has three operators: (1) orthogonal clossover for global search, (2) orthogonal pitch adjustment for local search, (3) random search for feasible space. OHS use the systematic reasoning methods, ODE and factor analysis, to find the right direction to approach the optimal solution and speed the search ability.We execute the proposed algorithm to solve eight test functions include of unimodal and multi-modal. Comparing with HS, OGA and OHS, we can find that OHS can quicker slove problems than them and more stable find optimal or close-to-optimal solutions.
