A Comparison of Hourly Wattage Prediction using Multiple Regression and Artificial Neural Network and ARIMA

• Main Authors: Jun-Yi Wu, 吳俊億
• Other Authors: Chung-Chian Hsu
• Format: Others
• Language: en_US
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/70958872756743094870

碩士 === 國立雲林科技大學 === 資訊管理系碩士班 === 100 === In recent years, demand for substitutable energy is increasing. For this reason, people begin to find the best substitutable energy. Among the substitutable energies, solar energy is a typical of its kind. Therefore, research issues relevant to solar energy were actively investigated recently. Predicting output of solar energy is the most widely discussed topic. Therefore, in this study, we attempt to use three techniques to predict output wattages. These models are applied in two experiments based on a collection of data from 09:00 to 15:00 hours. This work compares the performance on predicting wattage values. Experimental results show unsteady changes easily affect the prediction of one-step-ahead forecasting. Moreover, the prediction curves of MLR and BPNN model are intended to depend on the previous day’s values. In addition, the results also indicate that the radiation variable is an important index in the forecasting.