摘要
为了解决负荷数据非高斯、高维度影响预测模型精度的问题,提出一种基于局部保持投影-最小二乘支持向量机(LPP-LSSVM)的微电网超短期负荷预测算法。在这一算法中,根据负荷数据的时序相关性,以及天气、温度等因素,选取相似日组成训练数据集,利用局部保持投影进行特征提取,然后利用最小二乘支持向量机训练超短期负荷预测模型。通过对上海某公司微电网示范工程中实际运行负荷进行试验,确认所提出算法的有效性。
In order to solve the issue that the non-Gaussian and high-dimensional load data might affect the accuracy of the prediction model,an algorithm for microgrid ultra-short-term load prediction based on LPP-LSSVM was proposed.In this algorithm,according to the time sequence correlation of load data,as well as weather,temperature and other factors,the similar date is selected to constitute training data set.LPP is utilized for feature extraction,and then LSSVM is adopted to train the ultra-short-term load prediction model.The effectiveness of the proposed algorithm was confirmed by an experiment of practical operating load of a microgrid demonstration project in a certain company in Shanghai.
出处
《上海电气技术》
2019年第1期42-45,53,共5页
Journal of Shanghai Electric Technology
