基于BP神经网络的多参数关联变压器油击穿电压的预测研究

电力变压器的安全、稳定和经济运行是电力输送的重要保证。基于变压器油的日常监督数据,预测变压器油的击穿电压,将为变压器故障的早期诊断、预测防范和及时处理提供重要的科学依据。笔者鉴于BP神经网络具有任意的非线性映射、强大的自学习功能和良好的容错特性,采用BP神经网络的方法进行多参数关联变压器油击穿电压的预测研究。利用变压器油的日常监测数据,用BP算法和改进BP算法训练网络,分别建立了击穿电压与4个影响因素的关联模型。结果表明,基于改进BP算法模型的预测结果精度较高,预测值相对误差在5%以内,具有重要的实际应用价值。

Prediction of Breakdown Voltage of Transformer Oil via Multi-parameter Correlation Based on BP Neural Network

Prediction of breakdown voltage of transformer oil facilitates the early fault diagnosis and fault prevention of transformer.Since back-propagation(BP) neural network has prominent fault-tolerant,non-linear approximation,and self-learning capabilities,this paper presents a BP neural network with BP algorithm and a BP neural network with improved BP algorithm respectively to simulate the correlation between breakdown voltage and four relevant parameters,using the monitoring data of transformer oil.The results show the latter gives more accurate prediction values with a relative error within 5%.