Applied Continuous-time Threshold AR Model Approach to Short Term Forecasting

• Main Authors: Chen, Chi-Wei, 陳志偉
• Other Authors: Huang Sy-Ruen
• Format: Others
• Language: zh-TW
• Published: 1997
• Online Access: http://ndltd.ncl.edu.tw/handle/05951582849489083822

碩士 === 逢甲大學 === 電機工程研究所 === 86 === An accurate short-term load forecast model is an essential component of any Energy Managment System ( EMS ). This short- term load forecast can be used to forecast of either total MWH requirements during a period or peakMW requirements for that period. Based on the variable data, the operationanaylysts and system dispatchers are able to plan to control power system operations. This thesis presnts an on-line method for modeling and forecastingshort-term hourly load demand. The proposed method combined the Continuous-time Threshold Autoregresion ( CTAR ) model with the clusterrule. According to the proposed CTAR models algorithem, the state varuable form and the method of Kalman filters are applicable to estimatethe loadforecasting parameters. The results, based on Taipower historical(1993) loaddemand, indicate that the proposed algorithm is capable of providing moreaccurate load forecast and on-line forecast.