中国电机工程学报2016，Vol.36Issue(12)：3334-3342,9.DOI:10.13334/j.0258-8013.pcsee.152083
基于改进BP-SVM-ELM与粒子化SOM-LSF的微电网光伏发电组合预测方法
Combined Forecasting of Photovoltaic Power Generation in Microgrid Based on the Improved BP-SVM-ELM and SOM-LSF With Particlization
摘要
Abstract
ABSTRACT:Aimed at the difficulty of highly-accurate short-term forecasting of photovoltaic power generation in microgrid, a combined method with a new spatial view which only considers the direct relations among micro-sources was proposed. Back propagation neural network with the optimization of mind evolutionary computation, support vector machine with the optimization of particle swarm and extreme learning machine based on the single-hidden layer feed- forward neural network were used to forecast respectively. Variance- covariance combined method solved the problem of dynamic weight distribution. One day prediction and rolling prediction employ another method. This method used Kohonen's self- organizing feature map with particlization and least squares fitting to output results with equal weights. The simulation tests show that the proposed approach has high complementation, better flexibility and accuracy. It can provide a superior technical reference for the optimized operation of microgrid.关键词
分布式能源/微电网/组合预测/思维进化算法/粒子群算法/极限学习机/自组织特征映射网络Key words
distributed energy/microgrid/combined forecasting/mind evolutionary computation/particle swarm optimization/extreme learning machine/self-organizing feature map分类
信息技术与安全科学引用本文复制引用
单英浩,付青,耿炫,朱昌亚..基于改进BP-SVM-ELM与粒子化SOM-LSF的微电网光伏发电组合预测方法[J].中国电机工程学报,2016,36(12):3334-3342,9.基金项目
广东省部产学研项目(2012B091100179,2014B090903009,2016B090918107)；广东省科技计划(2013B010405009)；珠海市战略性新兴产业重大专项(2014D0601990002)。Project Supported by Guangdong province-ministry production-study- research projects (2012B091100179,2014B090903009,2016B090918107) （2012B091100179,2014B090903009,2016B090918107）
Guangdong province science and technology program (2013B010405009) （2013B010405009）
major project of Zhuhai strategic emerging industries (2014D0601990002) （2014D0601990002）
