摘要
针对电缆导体暂态温度的计算,提出一种粒子群优化的BP神经网络方法。该方法以实测的电缆导体电流、电缆外皮温度为输入,电缆导体温度为输出,构建一个四层神经网络模型。先使用粒子群(PSO)算法对模型进行一次优化训练,再使用BP算法对模型进行二次优化训练。试验研究表明,基于粒子群优化的BP神经网络可以精确地实时计算电缆的导体温度,且不受电缆本身物性参数影响,为电缆导体温度的在线监测提供技术支持。
In order to calculate the transient temperature of cable conductor,a method based on BP neural network optimized by particle swarm was introduced. This method builds a four-layer neural network model by using the load current and outer sheath temperature as inputs,the cable conductor temperature as output. The network is first optimized by particle swarm optimization（ PSO） algorithm and then is further optimized by BP algorithm. The research shows that being not affected by the cable physical parameters,this proposed method can accurately calculate the cable conductor temperature,which gives a reference on online monitoring of cable conductor temperature.
出处
《电器与能效管理技术》
2018年第3期14-19,24,共7页
Electrical & Energy Management Technology
关键词
配电电缆
导体温度
BP神经网络
粒子群优化
暂态计算
在线监测
power distribution cable
conductor temperature
BP neural network
particle swarm optimization （PSO）
transient calculation
online monitoring
