差分演化演算法應用於冰水主機負載分配最佳化

Optimal Chiller Loading by differential evolution algorithm for Reducing Energy Consumption

陳怡婷(Yi-Ting Chen)

指導教授 ： 李文興

國立臺北科技大學/機電學院/機電科技研究所/博士(2011年)

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、基因演算法和粒子族群演算法來做冰水主機負載分配的最佳化，除了粒子族群演算法能找出最佳值之外，平均負載法、拉格蘭傑乘數法和基因演算法皆有其缺點。
因此，本研究利用差分演化演算法(differential evolution algorithm；DE)進行冰水主機運轉時的負載分配以節約能源使用。在滿足空調系統負載需求條件下利用「突變、交配、天擇」等運算，使得整個群體漸漸地向最佳解收斂。
本論文使用兩個案例來做分析，以差分演化演算法之結果分別與平均負載法、拉格蘭傑乘數法、基因演算法和粒子族群演算法做比較，結果顯示，差分演化演算法在冰水主機負載最佳分配的計算上與拉格蘭傑乘數法及粒子群演算法相同都能找出最佳的解。因此，相較於平均負載法於不同負載下其最佳值之性能改善率平均分別為7.81%和9.4%；相較於基因演算法之性能改善率平均分別為7.25%和5.8%。然而在求解的平均值上，差分演化演算法相較於粒子族群演算法其計算的平均值之性能改善率平均為1.43%和1.02%。另外，利用拉格蘭傑乘數法在低負載下，因主機停機會造成主機性能成為不連續函數，無法直接應用求得最佳解。因此，本研究於低負載時以拉格蘭傑乘數法結合停機策略進行求解，此方法雖可求得與差分演化演算法相同之最小耗電率，但運算過程繁瑣。綜上所述，本研究結果發現應用差分演算法於冰水主機負載分配最佳化可改善目前所使用演算法的缺點，具有良好的計算性能。

關鍵字

差分演化演算法 ； 冰水主機負載分配 ； 最佳化

並列摘要

The Optimal Chiller Loading (OCL) method includes Average Loading Method (AVL), Lagrangian Method (LGM), Genetic Algorithm(GA) and Particle Swarm Algorithm (PSO) at present. Although the Genetic Algorithm method can overcome disadvantages that LGM can not obtain the analytic solution in the low-load, it is very complicated and difficult to make the coding of program. GA method is unable to find the optimal solutions.
This study employs differential evolution algorithm to solve the optimal chiller loading problem for reducing energy consumption. To testify the performance of the proposed method, the paper adopts two case studies to compare the results of the developed optimal model with those of the Lagrangian method, genetic algorithm and particle swarm algorithm. The result shows that the proposed differential evolution algorithm can find the optimal solution as the particle swarm algorithm can, but obtain better average solutions. Moreover, it not only outperforms the genetic algorithm in finding optimal solution, but also overcomes the non-solutions of analytic solution, which caused by the Lagrangian method occurring at low demands. Therefore, this study uses the Lagrangian method with the on-off strategy to obtain the analytic solution in the low-load. However, the complicated computation process will not be suitable for the multiple-chiller system.

並列關鍵字

differential evolution algorithm ； chiller loading ； engineering optimization

參考文獻

[9] 張永宗，冰水主機最佳排序，冷凍與空調雜誌，2002，pp.87-92.。

[16] 莊虔毅，以基因演算法求解含凹凸特性曲線之冰水主機最佳負載分配，台北科技大學碩士論文，台北，2004。

[11] Y. C. Chang, “A novel energy conservation method – optimal chiller loading,” Electric Power Systems Research, Vol. 69. 2004, pp.221–226.

[14] Y. C. Chang, J. K. Lin, M. H. Chuang, “Optimal chiller loading by genetic algorithm for reducing energy consumption.” Energy and Buildings, Vol. 37, 2005, pp.147–155.

[17] Y. C. Chang, W. H. Chen, C. Y. Lee, C. N. Huang, “Simulated annealing based optimal chiller loading for saving energy,” Energy Conversion and Management, Vol. 47, 2006, pp. 2044–2058.