【作者】 毛李帆；

【导师】 江岳春；

【作者基本信息】 湖南大学 ， 电力系统及其自动化， 2008， 硕士

【摘要】 中长期负荷预测是制定电力发展规划的基础,也是规划的重要组成部分。提高中长期负荷预测水平,有利于计划用电管理,有利于制定合理的电源建设规划,有利于提高电力系统的经济效益和社会效益。针对中长期负荷预测,本文介绍了偏最小二乘回归分析方法(PLS)的原理并详细推导了该算法的简化建模步骤。该方法能在最大限度地保留原有数据信息的前提下,将数据信息集中在几个互不相关的主成分上,因而能有效解决建立负荷预测模型时,样本个数较少及自变量存在严重多重相关性时,难以有效地通过多元回归分析建立预测模型的问题,并以衡阳地区经济发展指标和全社会用电量为基础数据,通过算例将偏最小二乘回归分析方法与最小二乘法和逐步回归进行了详细的比较说明。为解决偏最小二乘法中成分解释能力不均衡的情况,给出了正交信号修正法(OSC)的基本思想并详细推导了该算法的实现步骤,并将一种改进后的正交信号修正法与偏最小二乘法相结合,先对原始数据通过OSC消除正交分量,再利用PLS建立中长期负荷预测模型。该方法能有效地去除自变量系统中与因变量无关的正交数据信息,增强自变量因变量之间的相关性,在有限的成分中最大限度的提高成分解释能力。并通过算例将PLS与OSC-PLS进行比较分析,结果表明,运用OSC-PLS进行中长期负荷预测,尽管预测模型提取的成分个数变少了,但模型成分的解释性却大幅度增强,预测精度明显提高,具有较强的实用性。更多还原

【Abstract】 The medium and long-term load forecasting is the foundation on laying down the electric power system developing scheme, also the very important part of electric power system programming. Its improvement benefits not only the electric power plan administration, but also the suitable frame of electrical sources construction programming and the advance of economic and social profits in electric power system.This paper introduces the fundamental tenets and detailed calculating steps of partial least square method(PLS). The method can highly include the information of original data and concentrate it into some irrelevant primary component, so PLS can effective solve less samples problem and multiple correlation problems when making model of load forecasting. Comparing with least square method and step regression method, the results of calculation example show that in medium and long term load forecasting, the PLS, proposed in this paper, has high modeling speed and more forecasting accuracy.In order to eliminate unbalance among the explanation analysis of model’s components, the paper gives the fundamental tenets and detailed calculating steps of orthogonal signal correction(OSC) and combines partial least square method(PLS) with a improved OSC. The OSC-PLS method which eliminates orthogonal component before makes model of load forecasting can remove the useless information between the independent variable(X) and dependent variable(Y) effectively; strengthen the correlation of X and Y; improve the explanation of model’s component highly. Comparing with PLS method, the results of calculation example show that in medium and long term load forecasting, the model of OSC-PLS, proposed in this paper, with better explanation and forecasting accuracy, is more useful.更多还原