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

Optimal Chiller Loading by Genetic Algorithm Considering Convex and Non-Convex Characteristic Curves

莊虔毅(Chien-Yi Chuang)

指導教授 ： 張永宗

國立臺北科技大學/機電學院/冷凍空調工程系所/碩士(2005年)

https://doi.org/10.6841/NTUT.2005.00041

摘要

冰水主機為空調系統上主要耗能設備，而傳統的冰水主機負載分配以平均負載法（ELD）作為運轉依據，其每部主機性能與COP係假設一致，皆運轉於相同條件之情況之下並不見得是最佳運轉策略，由於其他因素使得每部主機性能效率並不一致，因此平均負載率並非各冰水機組之最佳效率點，故須整合各機組最佳運轉點以提供空調系統負荷需求，方能獲得最佳負載分配及最小總消耗功率。
拉格蘭傑乘數法對於kW－PLR性能曲線為凸函數時，有最佳主機負載分配耗電，但當kW－PLR性能曲線為凹凸函數共同存在時，拉格蘭傑乘數法卻無法求出最佳解，故本文使用基因演算法做為研究方法，利用kW－PLR性能曲線為凸函數與拉格蘭傑乘數法做負載分配比較，針對基因演算法所求近似最佳解與拉格蘭傑乘數法所求最佳解之間精確度誤差多少，進而在將基因演算法與平均負載法做主機最佳化負載分配耗電量之比較。
本文之研究方法為基因演算法，其所求近似解具高精確之特性而且以隨機性搜尋各機組最佳運轉點而滿足空調系統負載需求，而冰水主機負載分配最佳化之目的在於滿足空調系統負載需求條件下，決定每台冰水主機運轉之最佳負載率，而使系統總耗電量達到最小。

關鍵字

最佳化負載分配 ； 平均負載法 ； 拉格蘭傑乘數法 ； 基因演算法

並列摘要

Chiller is the major power consumption equipment of the HVAC system. Conventional chiller loading distribution is based on the Equal Loading Distribution（ELD）method. When every chiller is designing value of the performance and COP identically and operating a circumstance of the same condition, it is not the optimal operating ploy. Every chiller’s performance efficiency is different due to the other. However the ELD method can not result the most efficient loading condition due to the chiller performance characteristics being not identical and the more their runtimes the more differences between their characteristics.
Lagrangian Multiplier Method（LGM） is evaluating the power consumption with the optimal chiller loading distribution towards the kW-PLR performance curves is a non-convex function. When the kW-PLR performance curves is existing a convex and non-convex function together, the LGM can not obtain the optimum solution. So this research method of the thesis is based on the Genetic Algorithm（GA）. In connection with the non-convex of the kW-PLR performance curves is compared between Genetic Algorithm（GA） and LGM. Comparing Genetic Algorithm（GA） and LGM to discuss the both accurate optimal solutions, and then using Genetic Algorithm（GA） and Equal Loading Distribution（ELD） to finish Optimal Chiller Load Distribution（OCLD）, power consumption comparison.
The research method of the thesis is based on the Genetic Algorithm（GA）. Its solution has high accurate characteristic and can find the near optimal operation and conform the HVAC system load. And then the purpose of the OCLD is to meet the HVAC system load and decides the chiller’s optimal part load ratio（PLR）to reduce the system power consumption.

並列關鍵字

Optimal Chiller Loading Distribution ； Equal Loading Distribution ； Lagrangian Multiplier Method ； Genetic Algorithm

參考文獻

[6] 黃博興，「空調主機節約能源方法」，中國冷凍空調雜誌，Jun，1992。

[2] J.E. Braun, S.A. klein, J.W. Mitchell and W.A. Beckman, “Applications of optimal control to chilled water systems without storage”, ASHRAE Transactions. Vol. 95. , pp 663-675, 1989.

[12] 陳森煌，「以現場實用測量結果作冰水主機部分負載性能分析」，台北科技大學碩士論文，2001。

[5] 李良梧，「漫談建築、空調系統與節約能源」，冷凍空調雜誌，頁101-104，Jun，1991。

[15] ASHRAE Handbook, ”Supervisory Control Strategies and Optimization”, CH,40, 1999。