摘要
为了实现高精度的电力系统短期负荷预测,提出了基于粒子群优化(particle swarm optimization,PSO)算法优化门控循环单元(gated recurrent unit,GRU)神经网络的电力系统短期负荷预测模型。首先建立GRU神经网络,GRU神经网络采用了门控循环单元,与采用传统循环单元的传统循环神经网络相比,克服了传统循环神经网络中可能出现的梯度爆炸和梯度消失问题;继而采用具有较强全局优化能力的改进粒子群算法对GRU神经网络参数进行优化,有效提高模型的预测精度。通过实际算例仿真分析,并与传统的GRU神经网络预测模型以及反向传播(back propagation,BP)神经网络预测模型进行对比,验证了所提电力系统短期负荷预测模型具有较好的精度和稳定性。
In order to achieve short-term load forecasting of power system with high accuracy,a short-term load forecasting model of power system based on PSO algorithm to optimize the gated recurrent unit(GRU)neural network which adopts the gated recurrent unit is proposed in this paper.Compared with the traditional recurrent neural network(RNN)using traditional recurrent unit,The GRU neural network overcomes the problems of gradient explosion and gradient disappearance that may occur in the RNN.It uses the improved particle swarm optimization(PSO)algorithm with strong global optimization ability to optimize the parameters of GRU neural network,which can effectively improve the prediction accuracy of the model.Compared with traditional GRU neural network forecasting model and back propagation(BP)neural network forecasting model,the short-term load forecasting model proposed in this paper is proved to be more accurate and stable by simulation analysis of practical examples.
作者
王康
龚文杰
段晓燕
张智晟
WANG Kang;GONG Wenjie;DUAN Xiaoyan;ZHANG Zhisheng(College of Electrical Engineering,Qingdao University,Qingdao,Shandong 266071,China;State Grid Qingdao Electric Power Company,Qingdao,Shandong 266002,China)
出处
《广东电力》
2020年第4期90-96,共7页
Guangdong Electric Power
基金
山东电力科技项目(2019)
2016智慧青岛建设计划重点项目(强化重点领域智慧企业服务类-11)。
关键词
短期负荷预测
门控循环单元
GRU神经网络
粒子群优化
预测精度
short term load forecasting
gated recurrent unit(GRU)
GRU neural network
Particle Swarm Optimization
prediction accuracy
