基于BP神经网络的SF_(6)在线监测装置压力预测方法研究

Research on Pressure Forecasting Method of SF_(6) Density On-line Monitoring Device Based on BP Neural Network

GIS设备SF_(6)密度监测容易受环境因素影响产生一定的波动,对设备运行状态判定造成干扰,因此提出一种基于BP神经网络的SF_(6)压力预测方法,通过分析环境温度及其变化情况、相对湿度、风速、天气类型和导体电流等内外部因素对SF_(6)压力变化的影响,建立BP神经网络预测模型,对SF_(6)在线监测装置的压力值进行预测,为监测设备正常运行提供数据参考,并给出了分析GIS设备压力降低原因的判别策略。选取某特高压变电站GIS设备的典型SF_(6)在线监测装置气体压力和对应的自然环境及导体电流数据进行MATLAB仿真,结果表明,SF_(6)压力预测结果准确度可达98.5%以上,验证了该方法的可行性。

The SF_(6) pressure value of GIS equipment is easy to fluctuate under the influence of environmental factors,which affects the judgment on equipment operation state.This paper mainly studies the prediction method of SF_(6) pressure based on BP neural network.By analyzing the influence of internal and external factors such as ambient temperature,temperature transition rate,relative humidity,wind speed,weather type and conductor current on the change of SF_(6) pressure,the prediction model of BP neural network is established to predict the pressure value of SF_(6) on-line monitoring device,which provides data reference for the normal operation of monitoring device This paper also puts forward the discrimination strategy for the reason of pressure reduction on GIS equipment.By selecting the gas pressure and the corresponding natural environment and conductor current data of a typical SF_(6) on-line monitoring device of GIS equipment in an UHV substation for simulation calculation,the results show that the accuracy of SF_(6) pressure prediction results can reach more than 98.5%,therefore the correctness of theoretical analysis has been verified.