基于输入空间压缩的短期负荷预测

由于影响负荷预测的因素复杂，并且实际获取的历史数据有限，传统的智能预测方法往往达不到工程应用的精度要求。为解决该问题，文中提出一种准确预测电力系统短期负荷的新思路：首先建立负荷输入特征选择模型，其搜索方法采用浮动搜索算法，在去除影响负荷预测的冗余特征之后，利用有限样本学习的统计学习理论(支持向量机)构造负荷预测回归模型，充分发挥其在解决有限样本、非线性中体现出的优势，较好地提高了评估结果的精度和泛化能力。在EUNITE网络中的应用结果证明了该方法对电力系统负荷预测的有效性。

SHORT-TERM LOAD FORECASTING BASED ON INPUT DIMENSION REDUCTION

The traditional methods for load forecasting can not achieve the required accuracy for some engineering application due to the limited history data sets and the complex factors that affect the load forecasting. This paper presents a new framework for the power system short-term load forecasting. It firstly establishes the feature selection model and uses floating search method to find the feature subset. Then it makes use of the support vector machines to forecast the load and takes full advantage of the SVM to solve the problem with small sample and of nonlinear. Hence the accuracy of the estimation result is improved and a better generalization ability is achieved. The EUNITE network is employed to demonstrate the validity of the proposed approach.