- 學位論文
粒子族群演算法應用於冰水主機負載分配最佳化
Optimal Chiller Loading by Particle Swarm Algorithm for Reducing Energy Consumption
摘要
目前冰水主機負載分配法普遍都採用平均負載法，也有研究提出拉格蘭傑乘數法、基因演算法來做冰水主機的負載分配的最佳化，這些方法皆有其缺點，雖然基因演算法克服拉格蘭傑乘數法其在低負載下無法收斂之缺點，但其程式撰寫及求解過程相當複雜，且基因演算法卻無法找到最佳值。 本論文之研究是以粒子族群演算法(Particle Swarm Algorithm)，以其程式架構簡單，適用於求解最佳化問題的特性，粒子族群演算法以滿足空調系統負載需求為條件，在條件下隨機搜尋各機主運轉最佳點，使總耗電量達到最小。粒子族群演算法可改進拉格蘭傑乘數法其在低負載下無法收斂之缺點，也可排除基因演算法無法找到最佳值之缺點。 本論文使用兩個案例來做分析，以粒子族群演算法之結果與拉格蘭傑乘數法(Lagrangian Multiplier Method)及基因演算法(Genetic Algorithm)計算出結果比較，粒子族群演算法計算出結果與拉格蘭傑乘數法演算法幾乎相似，且可計算出低負載下之結果，計算結果比基因演算法更能找出最佳值。
並列摘要
The Optimal Chiller Loading (OCL) method includes Average Loading (AVL)method, Lagrangian Multiplier Method(LGM) method and Genetic Algorithm(GA) method at present. These methods all have several disadvantages. Although the Genetic Algorithm method can overcome disadvantages that Lagrangian Multiplier Method can not converge in the low-load, it is very complicated and difficult to make the coding of program. Genetic Algorithm method is unable to find the optimal solutions. The paper presents a method by using Particle Swarm algorithm to improve these defects. Particle Swarm Algorithm can not only improve disadvantages that Lagrangian Multiplier Method can not converge in the low-load, but also solve problems that Genetic Algorithm method is unable to find the optimal solutions. This study employs the Particle Swarm Algorithm to solve the optimal chiller loading problem. The energy consumption is considered as the objective function, and the partial loading ratio of each chiller is considered as the optimum parameter. From the result of two case studies, the Particle Swarm algorithm overcomes the flaw of Largrangian Multiplier Method which the system may not converge at low demands, and reduces the energy consumption more comparing with Genetic Algorithm.