基于改进PSO-LSSVM模型的变压器绕组热点温度预测Prediction of Hot-Spot Temperature in Transformer Winding Based on Improved PSO-LSSVM Model
刘闯,卢银均,刘红云,向晓,王梁伟
LIU Chuang,LU Yinjun,LIU Hongyun,XIANG Xiao,WANG Liangwei
摘要(Abstract)：
针对现有变压器绕组热点温度预测方法中存在的不足,采用收缩因子对粒子速度更新方式进行改进,保证PSO算法前期的全局搜索能力和后期的局部寻优能力,提高了算法的收敛性能;利用改进PSO对LSSVM参数进行寻优,建立基于改进PSO-LSSVM的变压器绕组热点温度预测模型。利用实际监测数据进行仿真分析,改进PSO-LSSVM的变压器绕组热点温度预测模型的预测效果优于其他方法,验证了本方法的正确性和实用性。
In view of the shortcomings of the existing hot spot temperature prediction methods of transformer windings, the shrinkage factor is used to update the particle velocity to ensure the global search ability in the early stage and the local optimization ability in the late stage of PSO algorithm, so as to improve the convergence performance of the algorithm. The parameters of LSSVM are optimized, and the prediction model of transformer winding hot spot temperature based on improved PSO-LSSVM is established. Through the simulation analysis of the actual monitoring data, the prediction effect of the improved PSO-LSSVM transformer winding hot spot temperature prediction model is better than other methods, which verifies the correctness and practicability of this method.
关键词(KeyWords)：
粒子群算法;最小二乘支持向量机;变压器;绕组热点温度;收缩因子;粒子速度更新方式
particle swarm optimization;least squares support vector machine;transformer;hot-spot temperature of winding;shrinkage factor;particle velocity update method
基金项目(Foundation):
作者(Author):
刘闯,卢银均,刘红云,向晓,王梁伟
LIU Chuang,LU Yinjun,LIU Hongyun,XIANG Xiao,WANG Liangwei
参考文献(References)：
- [1]徐永明，刘飞，齐玉麟.基于流体网络的电力变压器绕组温度预测[J].高电压技术，2017,43(5):1509-1517.
- [2]刘辉，刘强，张立，等.基于改进参数辨识方法的变压器绕组热点温度计算[J].变压器，2018,55(4):68-74.
- [3] Chen W G, Pan C, Yun Y X. Power transformer top-oil temperature model based on thermal-electric analogy theory[J]. European Transactions on Electrical Power, 2009, 19(3):341-354.
- [4]李树卿，陈鼎，仇群辉，等.基于灰色神经网络的变压器绕组热点温度预测[J].自动化与仪器仪表，2017(4):116-118.
- [5]李孟励.基于BP神经网络的变压器绕组热点温度预测方法研究[D].重庆：重庆大学，2010.
- [6]陈伟根，滕黎，刘军，等.基于遗传优化支持向量机的变压器绕组热点温度预测模型[J].电工技术学报，2014,29(1):44-51.
- [7] Suykens J A K,Vanddewalle J.Least Squares Support Vector Machines Classifiers[J].Neural Network Letters,1999,19(3):293-300.
- [8]刘闯，何沁鸿，卢银均，等.输电线路PSOEM-LSSVM覆冰预测模型[J].电力科学与技术学报，2020,35(6):131-137.
- [9]罗涛.粒子群算法在电量组合预测中的应用[J].内蒙古电力技术，2017,35(1):43-46.
- [10]杨凌云.改进粒子群算法在车辆路径问题中的应用研究[D].开封：河南大学，2011.
- [11]滕黎.油浸式变压器热路计算及热点温度遗传支持向量机预测研究[D].重庆：重庆大学，2012.