摘要
为了提高短期光伏发电预测的准确性,提出一种改进Semi Boost(Semi-supervised Boosting)天气聚类法和结合混沌纵横交叉的粒子群优化算法(Particle swarm optimization combined with chaos crossover,CC-PSO)优化深度置信网络(Deep Belief Network,DBN)连接权重的组合光伏发电功率预测方法。为了提高预测精度,设计并训练了Semi Boost改进的基于加权K近邻(Weighted K-nearest Neighbor,WKNN)置信度传播(Belief Propagation,BP)分类方法,对各天气类型采用对应的网络进行预测。DBN连接权重采用CC-PSO算法优化,避免出现由随机初始化导致的局部最优解现象,从而提高了DBN网络预测性能。实验结果验证了该模型的有效性。
In order to improve the accuracy of short-term photovoltaic power generation forecasting,we propose an improved Semi Boost weather clustering method and a particle swarm optimization combined with chaos crossover(CC-PSO),which combines photovoltaic power forecasting method for optimizing the connection weight of deep belief network(DBN).In order to improve the prediction accuracy,we designed and trained the Semi Boost improved belief propagation(BP)classification method based on the weighted k-nearest neighbor(WKNN).And the corresponding network was used to predict each weather type.The connection weight of DBN was optimized by CC-PSO,which avoided the phenomenon of local optimal solution caused by random initialization and improved the prediction performance of DBN network.Finally,the experimental results show the effectiveness of the proposed model.
作者
孙辉
冷建伟
Sun Hui;Leng Jianwei(School of Electrial and Electronic Engineering,Tianjin University of Technology,Tianjin 300384,China)
出处
《计算机应用与软件》
北大核心
2020年第8期103-109,共7页
Computer Applications and Software