基于长短期记忆循环神经网络的开关柜设备温度预测

Switchgear equipment temperature prediction based on LSTM recurrent neural network

为了提高开关柜设备温度预测的准确性,提出一种基于长短期记忆(Long Short-Term Memory,LSTM)循环神经网络的开关柜设备温度预测方法.首先获取电力开关柜设备相关数据集,并对原始数据集进行特征变量分析选择、数据预处理等工作;然后将处理后的数据集输入至LSTM循环神经网络中进行训练,得到LSTM温度预测模型;最后以6 kV开关柜母线设备为例,与多种预测算法进行设备温度预测对比实验.实验结果表明:相较于经典的神经网络和循环神经网络(Recurrent Neural Network,RNN)预测方法,本文所提预测方法对开关柜内设备温度的预测具有更高的准确率,为设备的主动预测性维护提供了一种有效途径.

In order to improve the accuracy of temperature prediction of switchgear equipment,a method for temperature prediction of switchgear equipment based on LSTM recurrent neural network is proposed.Firstly,obtain the relevant data set of the power switchgear equipment,and analyze and select the characteristic variables and data preprocessing of the original data set.Secondly,the processed data set is input to the LSTM recurrent neural network for training,and an LSTM temperature prediction model is obtained.Finally,taking the bus equipment of 6 kV switchgear as an example,and a comparison experiment of device temperature prediction is performed with various prediction algorithms.The experimental results show that:By comparing with the traditional neural network and classic recurrent neural network(RNN)temperature prediction model,the method proposed in this paper has higher accuracy for the temperature prediction of the equipment in the switchgear.It provides an effective way for active predictive maintenance.