應用類神經網路於醫院空調短期電力預測

Short-term Electricity Forecasting of Air-conditioners of Hospital Using Artificial Neural Networks

施順鐘(Shun-Chung Shih)

指導教授 ： 陳昭榮

國立臺北科技大學/機電學院/電機工程系所/碩士(2005年)

摘要

本文利用具有優異預測能力的類神經網路來預測實際醫院空調一小時後電力耗能，並指出影響實際醫院空調電力耗能的可變因素包括溫度、相對溼度、前一小時空調電力耗能及一些不可控制的因素，如醫院的手術量、開設的門診診間量等，而固定的因素則為室內面積、病床數等。本研究為了實現對醫院的能源管理，以類神經網路建構預測實際醫院空調下一小時耗能的三種模式：每一天（全日）、工作日、工作時，並以非工作時及不考慮門診量等進行測試及驗證，結果發現以工作日模式的預測絕對平均誤差率4.99%為最小，且不考慮門診因素下的工作日模式比考慮的預測絕對平均誤差率為高，此結果反應出醫院門診病患數是決定空調負載的主要因素之一，且全日及工作日模式的預測模式是可用於實際醫院短期空調用電預測，未來並可進一步應用於協助醫院電力管理者進行能源管理。

關鍵字

類神經網路 ； 電力預測 ； 熱能通風空氣調節(HVAC) ； 醫院

並列摘要

This thesis is proposed the practical predictions of hospital air-conditioner next one hour electricity using the artificial neural network, owing to its excellent predict ability. The influence variables of hospital air-conditioner electricity are included temperature, relative humidity, the previous one hour electricity, the time in day, and some uncontrolled variables, e.g. the number of surgical operations, the number of persons; and some fix variables, e.g., the area of indoors, the number of sickbeds, etc. This paper can realize energy management to be important. There are three types of artificial neural network models to do the prediction of the next hour electricity load in the hospital, they are each day, weekday and work-time with outpatient considering. Test and prove the result with the work-time and weekday without outpatient considering for the three types. It was found that the workday prediction absolute average error ratio (AMER) 4.99% is lower than other type. The AMER was higher when did not consider outpatient under the weekday. The results show the numbers of outpatient is the main factor for the air conditioner load in hospital. The most important result is the each day and weekday type can be used in hospital for short-term electricity forecasting. And the results show that the predictions are excellent and able to assist the operators to do energy management in future.

並列關鍵字

Artificial Neural Network (ANN) ； Electricity Forecast ； Heating ； Ventilating and Air-Conditioning (HVAC) ； Hospital

參考文獻

[15] 李靖男，「冷凍空調系統之「奧運」標誌」，中國冷凍空調雜誌，35期，1997，第116-121頁。

[17] 趙宏耀、嚴志偉，「中央空調系統節能連環炮」，中國冷凍空調雜誌，40期，1998，第98-104頁。

[1] Y. Y. Hsu and C. C. Yang, "Design of artificial neural networks for short-term load forecasting. part I: self-organising feature maps for day type identification," IEE PROCEEDINGS-C, vol. 138, Sep. 1991, pp. 407-413.

[2] I. Drezga and S. Rahman, "Short-term load forecasting with local ANN predictors," IEEE Trans. Power Systems, vol. 14, Aug. 1999, pp. 844-850.

[3] S. H. Ling, Frank H. F. Leung, H. K. Lam, Y. S. Lee, and P. K. S. Tam, "A novel genetic-algorithm-based neural network for short-term load forecasting," IEEE Trans. Industrial Electronics, vol. 50, Aug. 2003, pp. 793-799.