摘要
电能质量直接影响电力系统的稳定性和设备运行的可靠性,电流总谐波畸变(Total Harmonic Current Distortion, THDi)是衡量电能质量的重要指标。为了解和掌握电力系统中的电流谐波畸变情况,预测THDi变得非常必要。鉴于现有THDi预测模型的预测精度不足问题,基于门控循环单元(Gated Recurrent Unit, GRU)神经网络、粒子群(Particle Swarm Optimization, PSO)算法和模拟退火(Simulated Annealing, SA)算法,提出了一种SA-PSO算法优化GRU的THDi预测模型,并采用均方根误差(RMSE)和平均绝对误差(MAPE)对模型的预测精度进行了分析。仿真与试验结果均表明:与PSO-GRU预测模型相比,所提模型的RMSE和MAPE分别减小了16.9%和18.6%。相较于GRU预测模型,所提模型的RMSE和MAE分别减小了39.1%和36.2%,有效提高了THDi的预测精度。
Power quality directly affects the stability of the power system and the reliability of equipment opera-tion.Total Harmonic Current Distortion(THDi)is a crucial metric for assessing power quality.It is essential to pre-dict THDi in order to comprehend and control the current harmonic distortion in the power system.Considering the in-adequate prediction accuracy of existing THDi prediction models,this paper introduces an SA-PSO algorithm to opti-mize the THDi prediction model of the Gated Recurrent Unit(GRU)neural network.The algorithm combines Particle Swarm Optimization(PSO)and Simulated Annealing(SA)to optimize the model and analyzes its prediction accura-cy using Root Mean Square Error(RMSE)and Mean Absolute Percentage Error(MAPE).Both simulation and ex-perimental results demonstrate that compared with the PSO-GRU prediction model,the RMSE and MAPE of the pro-posed model are reduced by 16.9%and 18.6%respectively.Compared with the CRU prediction model,the RMSE and MAE of the proposed model are reduced by 39.1%and 36.2%respectively,effectively improving the prediction accuracy of THDi.
作者
邱孟夏
刘保国
张会广
QIU Meng-xia;LIU Bao-guo;ZHANG Hui-guang(School of Electrical Mechanical Engineering,Henan University of Technology,Zhengzhou Henan 450001,China;Henan Key Laboratory for Super Abrasive Grinding Equipment,Zhengzhou Henan 450001,China)
出处
《计算机仿真》
2025年第9期503-508,共6页
Computer Simulation
基金
国家自然科学基金资助项目(12072106)。
关键词
总谐波畸变预测
门控循环单元
模拟退火算法
粒子群优化
Total harmonic distortion prediction
Gated recurrent unit
Simulated annealing
Particle swarm optimi-zation
