結合粒子群演算法及大渦模擬進行橢圓柱排列隊形最佳化之研究

The Research on Optimizing Formation of Elliptic Cylinders by Combining Particle Swarm Method and Large Eddy Simulation

本研究主要目的主要是以單目標與多目標粒子群演算法為基礎，探討三維流場中多個橢圓柱隊形的阻力最佳化問題，運用C#為程式架構主體撰寫粒子群演算法，結合大尺度渦流模擬法及繪圖軟體進行二次開發，除此，根據使用者限制條件及適應函數的需求，透過CFD軟體FLUENT計算並回饋回粒子群演算法進行最佳化的搜索，比較在層流及紊流等不同流場狀況下分析雷諾數對橢圓柱之影響，爾後比較並驗證其結果，以增加數值模擬的可信度。引入粒子群演算法，以阻力之最佳化為主，並加入升力以及力矩進行第二目標函數的考量，最後，將其最佳之隊形排列輸出，修正原始隊形以彙整出多目標最佳化之柏拉圖解集，根據最佳隊形之優化結果，達到省時省能的預期目標。 This research aims to adopt the single objective particle swarm optimization (PSO) and multi-objective PSO (MOPSO) to optimize the formation of elliptic cylinders in the three-dimensional flow. In addition, Large Eddy Simulation (LES) model is included to calculate turbulence flow. The simulation of optimization is originally based on the two-dimensional flow, and then extended to three-dimensional flow. The difference of elliptic cylinders in the flows of high and low Reynolds numbers is also discussed. In order to search the feasible solution, PSO and MOPSO are suggested for adjustment and for reducing computing resources. The difference of results between the PSO and MOPSO would be discussed. The framework of program is written in Microsoft Visual Studio 2008 C#, including PSO algorithm, changing the positons of elliptic cylinders, importing the files to ANSYS FLUENT and returning the data of forces back to PSO for correcting the next calculation.