應用類神經網路於高齡者住宅負載估測

人口老化是世界共同的趨勢，民國100年台灣高齡化比率已達到10.9%，高齡人口總數高達253萬；且經內政部預計在民國125年，高齡人口比例將達21.6%，屆時全國高齡人口將會達到520萬人，高齡人口成長相當快速。由於家庭結構及社經環境的變遷影響下，對於高齡者住宅的需求日益增大，有良善經營管理的高齡者住宅是影響選擇入住意願的重要因素。營運成本對於高齡者住宅的經營管理是相當重要的環節，而電能消耗在營運成本中所佔的比例是最多的部分。因此，本論文以電力監控系統歷史紀錄之用電量、中央氣象局之溼度、溫度與住戶數等資料做為輸入層來訓練類神經網路，輸出層為估測日之用電量，藉以推估往後之負載用電量。由估測結果顯示，應用倒傳遞類神經網路訓練模組所估測之負載用電量，平均誤差最低可收斂至0.1%以內，準確度達可接受的程度。估測之負載用電量資訊，可提供高齡者住宅經營部門作為電能成本之參考與營運管理之應變，以達到永續經營的目標。

關鍵字

人口老化；高齡者住宅；用電量；類神經網路

並列摘要

Population aging is a common trend in the world, in 2011 the aging ratio has reached 10.9%, the total number of the elderly population up to 2.53 million; and is expected by the Ministry of the Interior in 2036, the proportion of elderly population will reach 21.6%, when the national elderly population will reach 5.2 million, the elderly population is growing very fast. Elderly housing due to the impact of family structure and socio-economic changes in the environment, the demand for elderly housing is increasing, the good management is an important factor to influence the choice of admission to the wishes.Therefore, this thesis, the power consumption of the power monitoring system history, the Central Weather Bureau of humidity, temperature and number of households and other information as the input layer to train the neural network, the output layer to estimate electricity consumption in order to estimate the future electricity consumption of the load.The estimation results show that the application back-propagation neural network training module estimate of the load power consumption, the lowest average error to converge to less than 0.1% accuracy up to an acceptable level. Estimate of the load power consumption information to provide elderly housing operations as a reference for energy costs and operational management of the strain in order to achieve the objectives of sustainable management.

並列關鍵字

Population Aging ； Elderly Housing ； Electricity Consumption ； Artificial Neural Network

參考文獻

[13] 簡君芳，從經營管理觀點檢討高齡者住宅政策與相關法令之研究，碩士論文，國立臺北科技大學建築與都市設計研究所，台北，2008。

[19] 吳惠萍，高齡安養照護選擇因素之研究與分析，碩士論文，國立臺北科技大學工業工程與管理研究所，台北，2007。

[29] 柯致平，結合類神經網路及灰關聯分析法估測鉛酸蓄電池放電時間，碩士論文，國立台北科技大學電機工程系碩士班，2010。

[30] 劉乃銘，以類神經網路應用於分碼多工系統近似位元錯誤率之研究，碩士論文，國立台北科技大學電機工程系碩士班，2006。

[1] H. S. Hippert and C. E. Pedreira, "Estimation temperature profiles for short-term load forecasting: neural networks compared to linear models," IEE Proceedings Generation, Transmission & Distribution, vol. 151, no. 4, 2004, pp. 543-547.

國際替代計量

應用類神經網路於高齡者住宅負載估測

未授權

主題瀏覽