“负荷趋势+混沌”预测法的改进  

Improved Method for Short-term Load Forecasting Based on Synthesizing Load Trend and Chaos

作　　者：穆青青[1] 伍永刚[2] 杨纪明[3] 

机构地区：[1]汉江集团信息中心,丹江口442700 [2]华中科技大学水电与数字化工程学院,武汉430074 [3]丹江口水力发电厂,丹江口442700

出　　处：《电力系统及其自动化学报》2013年第6期160-166,共7页Proceedings of the CSU-EPSA

摘　　要："负荷趋势+混沌"预测法,将负荷拆分成"负荷趋势"和"混沌"部分。但其在分离负荷趋势部分时提取过量,混沌预测模型存在累积误差。针对这些问题文中根据周期与混沌的频谱特性,通过快速傅里叶变换确定周期分离的量;根据邻近点是否是预测值,对加权一阶局域法多步预测的邻近点权重增加因子进行修正;同时对剩余负荷为随机噪声的情况进行了处理。通过对海南地区负荷进行预测,验证了改进后的方法在预测精度方面有明显提高。The power load series can be divided into two components by the forecasting method of synthesizing load trend and chaos, which namely load trend and chaotic component. However, the part of periodic frequency is totally extracted, meanwhile, the chaotic prediction model can cause the accumulated error. In this paper, the quantity of seperating periodic part is considered and determined by fast Fourier transform according to the periodic and chaotic spectrum characteristics. The weight of adjacent points which are forecasted by the adding-weight one-rank local-re- gion method are modified by adding a factor according to the adjacent points whether is the predictive value. Mean- while, it makes an analysis on the rest of the load which is the random noise. Moreover, through the Hainan area load forecasting, the results demonstrate that the proposed method is superior to the previous methods in forecasting accuracy.

关 键 词：短期负荷预测 负荷趋势 混沌 加权一阶局域法 邻近点权重 

分 类 号：TM715[电气工程—电力系统及自动化]