中央空調直接負載控制績效分類與評估系統

Performance Classification and Evaluation Systems for Direct Load Control of Central Air Conditioners

陳世杰(Shih-Cheih Chen)

指導教授 ： 楊宏澤

中原大學/工學院/電機工程研究所/博士(2005年)

https://doi.org/10.6840/cycu200500627

摘要

中文摘要
電能需求為現代人生活基本必要條件，近年來由於生活水準提高用電量隨之大幅提高，尤其以夏季時期空調負載用電更為明顯。尖峰負載持續攀升現象常造成電力公司供電不足問題，為解決此一問題，電力公司乃實施各種負載管理措施，以期達到抑低尖峰負載與拉高離峰負載的目的。
本文開發一負載監測系統，可藉由萬用序列匯流排（universal serial bus, USB）介面為基礎之虛擬儀表監控和收集參與空調直接負載控制用戶用電資料。虛擬負載監控系統所收集用電資料提供自動分類負載曲線系統作實施分類，鑑別符合與不符合負載控制策略之用戶。對於符合負載控制策略之用戶，本文擬定績效評估模式，以估算其抑低負載量與轉移用電量；對於不符合負載控制策略之用戶則予以剔除。
為有效達成負載曲線之分類目標，本文分別發展兩種自動化負載曲線分類系統，以供正確剔除不符合負載控制策略之用戶。第一種分類系統是以非線性主要成份分析及週期性分析作為負載曲線特徵之擷取，再透過自我組織映射網路(self-organization map, SOM)分類工具，配合分類適確性Davies-Bouldin（DB）指標指出該資料最佳分類群聚，以建構自我組織映射網路分類系統。另一套分類系統則是以應用基因演算法於k-means群聚分析法則作為負載曲線分類方法，其特徵擷取方法是採用自我迴歸移動平均（auto-regressive moving average, ARMA）建立負載曲線模型，並以Akaike資訊準則（Akaike information criterion, AIC）指標找出最佳模型維度，再取模型係數作為特徵向量。
為驗證所提直接負載控制曲線分類系統與績效評估之可行性，本論文所提方法實際測試於台電夏季實施中央空調負載管理措施資料，其中包含中央空調週期性暫停用電、中央空調遙控降載週期性暫停用電與儲冷式空調系統等負載管理措施。同時本文亦使用原始資料直接與不同分類演算法進行分類結果測試，以驗證特徵擷取方法以及分類演算法的有效性。分類辨識率結果可以明顯看出本文所提出之特徵擷取方法以及分類演算法之效能。
透過上述分類系統篩選出符合負載控制策略之用戶，本文建立參與空調直接負載控制措施之績效評估模式，並利用所建構之分類系統，評估各參與用戶之負載抑低及轉移效果，期改善各項空調直接負載控制等負載管理策略之實施成效。

關鍵字

自我迴歸移動平均模式 ； 中央空調 ； 自我組織映射網路 ； k-means ； Akaike資訊準則 ； 直接負載控制

並列摘要

Abstract
Nowadays, electrical energy has been an essential living demand. In recent years, growth of living standards makes electrical energy increase rapidly. Especially, load demand of the air-conditioning in has increased apparently. Dramatic growth of the peak load leads to deficient power supply of utilities. To solve the problem, various load demand-side management (DSM) programs have been adopted. Expected to be achieved are reducing the peak load, whereas increasing the off-peak load.
The dissertation develops a load monitoring system, where a universal serial bus (USB) interface based virtual instrument is used to monitor and collect the data for the power consumption of the customers involved in the programs of demand management. The data collected for the direct load control (DLC) curves are then classified by an automatic classification system. Through the system, identified is either the DLC curve that complies with the control model of a specific DLC program or the other one that does not. For the one complying with the DLC model, the performance evaluation models are developed to evaluate the amounts of load reduction or load shifting of the customers that commit the DLC programs. The load curve that does not comply with the DLC model would be excluded from the database and not taken into account in the performance evaluation.
To achieve the above purpose of effective DCL curve classification, two classification systems are developed for excluding the one that does not comply with the DLC model. The first classification system uses the methods of nonlinear principal component analysis (NLPCA) and periodic analysis for feature extraction. Then, the features of DLC curves are classified through self-organizing maps (SOM) networks of the classification system. In the system, the Davies-Bouldin (DB) index is employed to determine the best number of the clusters to achieve better classification. The other classification system established is the genetic k-means algorithm (GKA) based classification system. In the system, auto-regression moving averaging (ARMA) modeling techniques are employed to extract the features of the DLC curves. To ensure the adequacy of the ARMA models used to represent DLC curves, Akaike information criterion (AIC) is relied on in the dissertation.
To verify the feasibility of the above proposed approaches, the approaches have been tested on the practical DLC database for the summer loads of central air-conditioning customers of Taiwan Power Company. The database includes the data for central air conditioning duty cycling control, paging system in central air conditioner duty cycling control, and ice storage central air conditioning systems. To demonstrate the effectiveness of the proposed methods the classification algorithms, also given are the results of tests using raw load-curve data and various classification algorithms. The results reveal the effectiveness of the proposed feature extraction and the classification algorithms.
For the load curves complying with the DLC programs through the classification system, the performance evaluation models of various DLC programs have been built up. Performance of load reduction and effectiveness of load shifting can thus be assessed. Accordingly, efficiency of the DLC programs can be improved expectedly.

並列關鍵字

central air conditioning ； ARMA model ； k-means ； AIC ； direct load control ； SOM network

參考文獻

[1] F. Kreith, “Integrated Resource Planning,” Journal of Energy Resources Technology, Vol. 115, pp 80-85, June 1993.

[2] S. Rehman and K. Rinaldy, “An Efficient Load Model for Analyzing Demand Side Management Impact,” IEEE Trans. on Power Systems, Vol.8, No. 3, pp.1219-1227, August 1993.

[6] S. Majumdar, D. Chattopadhyay, and I. Parikh, “Interruptible Load Management Using Optimal Power Flow Analysis,” IEEE Trans. on Power Systems, Vol.11, No. 2, pp.715-720, May 1996.

[7] M. W. Gusftafson, J. S. Baylor, and G. Epstein, “Estimating Air Conditioning Load Control Effectiveness Using An Engineering Model,” IEEE Trans. on Power Systems, Vol.8, No. 3, pp.972-978, August 1993.

[8] A. T. Almeida and E. L. Vine, “Advanced Monitoring Technologies for The Evaluation of Demand-side Management Programs,” IEEE Trans. on Power Systems, Vol.9, No. 3, pp.1691-1697, August 1994.