倒傳遞類神經網路於電量需求預測之應用

The Application of Back-propagation Neural Network on the Forecasts of Electricity Demand

黃豐裕(Feng-Yu Huang)

指導教授 ： 黃怡詔

國立屏東科技大學/管理學院/工業管理系所/碩士(2012年)

https://doi.org/10.6346/NPUST.2012.00102

摘要

由於產業生產時段差異與民眾生活步調不同，用電時段不一致造成用電量離尖峰的現象。為滿足用戶電力需求，台電必須預測電力需求並提前完成電力調度工作，以避免造成電力用戶缺電的困擾。為瞭解台灣地區用電量的影響因子，過去文獻中有學者以迴歸方法分析出氣象的變化和總用電量有一定的關聯性，因此運用電力公司用電量的資料，搭配中央氣象局氣象站測得的數據，以倒傳遞類神經網路求出溫度、濕度、風速、氣壓、雨量與用電量的關聯性。在求解過程中，以不同學習速率與慣性因子求解出最佳的權重值，電力公司根據此權重值建構出區域電量預測系統。當獲得氣象局的預報資料後，將預報資料轉檔輸入預測系統，藉此推算出區域的電量需求，讓電力調度單位提前做好電力調度的工作，以提高供電品質與用戶滿意度。

關鍵字

倒傳遞類神經網路 ； 電力調度系統 ； 電量預測

並列摘要

Due to electric power consumption alternates between domestic and industrial usage, there’d be peak and off-peak hours for electricity supply. To satisfy the user’s electricity demands, Taipower (Taiwan Power Company) must forecast the demands for electricity in advance in order to avoid causing power shortage problems for all its clients. In order to find out the impact factors of the regional electricity consumption in Taiwan, some studies show that the combination between the weather’s changes and the total electricity consumption could be analyzed by regression analysis. According to the electricity consumption data from Taiwan Power Company and the measurement reports from Central Weather Bureau, using Back-propagation Neural Network to evaluate the relations among the power consumption and the temperature, humidity, wind speed, atmospheric pressure and rainfall. In the whole procedure, taking different learning rate and momentum factors to evaluate the best weight for the Power Company to found a regional power consumption prediction system. With Central Weather Bureau’s forecast data, converse all the data to input the prediction system to estimate the demands of regional power consumption. Therefore the usage of power dispatch units could schedule and optimize the power availability to user’s demands beforehand.

並列關鍵字

Back-propagation Neural Network ； Power Dispatch System ； Power Prediction

參考文獻

5. Abu-Et-Magd M.A., R.D. Findlay, “A New Approach Using Artifical Neural Network and Time Series Models for Short Term Load Forecasting,” IEEE Electrical and Computer Engineering, 1723-1726(2003).

6. Affijulla S., C.Sushil, “A New Intelligence Solution for Power SystemEconomic Load Dispatch,” IEEE Environment and Electrical Engineering, 1-5(2011).

7. Amjady N., F. Keynia, “Short-term Load Forecasting of Power Systems by Combination of Wavelet Transform and Neuro-evolutionary Algorithm,” Engrgy 34, 46-57(2009).

8. Asbury C.E., “Weather Load Model Electric Demand and Energy Forecasting ,” IEEE Transactions on Power Apparatus and Systems, 94, no. 4, 1111-1116(1975).

9. Azdeh A., S.F. Ghaderi, S. Tarverdian, M. Saberi, “Integration of Artificial Neural Networks and Genetic Algorithm to Predict Electrical Energy Consumption,” Applied Mathematics and Computation, 186, 1731-1741(2007).