基于加权支持向量机与AdaBoost集成的预测模型研究

摘要

Power load forecast is a hot topic in recent year's researches, and is also rather difficult due to the impacts from the factors such as temperature, humidity and nature disaster. Hence, in this paper, in order to overcome the influence of these stochastic factors on power load, we introduce the weights of historical observational data and improve the support vector regression forecasting model and the intelligent selection of model parameters. We also use AdaBoost algorithm to boost the capability of weighted support vector regression forecast and to improve the forecasting precision. By simulative modelling, we carry out the forecasting experiment on actual power load data, the result illustrates that the proposed approach has better forecasting accuracy than the single SVM model and the BP model have with high stability.%电力负荷预测是近年研究的热点话题,因受温度、湿度、自然灾害等因素影响,准确预测相当困难.为此,通过引入历史观察数据的权重、改进支持向量回归预测模型和参数的智能选取,克服影响电力负荷的随机因素的影响,运用AdaBoost算法提升加权支持向量回归预测能力,提高预测精度.通过仿真建模,对真实的电力负荷数据进行预测实验,结果表明所提的方法比单个SVR模型和神经网络BP模型的预测精度高,稳定性好.