灰色模糊預測之理論應用於電力系統負載預測研究

Grey-Fuzzy Prediction Theory Applied in the Load Forecast of Power System

林家傑(Gia-Jei Lin) ； 張士行(Shih-Hsing Chang)

《中華技術學院學報》 21期 (2000/09) Pp. 46-53

https://doi.org/10.7095/JCIT.200009.0046

摘要

灰色動態預測模式（Grey Dynamic Model；GM）係針對自然界中之物系統多具有能量系統之特徵，故可以用簡單的指數函數來描述其動態規律性，在透過對原始數列的生成運算與簡單建模後，便可建立有效的預測模式。對於預測模式中的未知係數，一般以最小二乘法（Least-Square Method）求得，然而最小二乘法並不適合處理變異性過大的資料，而且當資料數目過少時也容易產生嚴重誤差，導致預測值偏離實際觀測值。本研究以模糊目標迴歸法（Fuzzy Goal Regression Method）取代傳統的最小二乘法，求解預測模式中的未知係數，建立灰色模糊動態預測模式（Grey-Fuzzy Dynamic Prediction Model；GFM），並將其理論與結果應用於品保壽命預測研究。

並列摘要

The previous development of grey dynamic model was based on the similarity characteristics between real world physical systems and energy systems such that all the series of data can be described by an exponential function after the pre-treatment process. Least-squared method was used for the determination of those parameters in the exponential function. However, it is recognized that least-squared method is not suitable for handling series of data with less amount of samples or larger, non-homogeneous variance. This analysis proposes the use of fuzzy goal regression method to replace the least-square method for the prediction in a grey system. In this paper, several kinds of GM models and the linear regression method are proposed to forecast the five-year load demand. From the simulation results, using the GM (1, 1) model predicts the load demand, and then combining the GM (1, 1) and GM (1, 1, t) models. A case study of the load prediction of power system applied by the grey fuzzy dynamic prediction model. Therefore, this method is suitable for the long-term load forecasting.