摘要
为了获得降雨条件下能预测空气间隙击穿电压的数学模型,根据在人工气候室试验得到的降雨条件下空气间隙击穿电压数据,运用神经网络原理,建立了降雨条件下的交流棒-板短空气间隙击穿电压的人工神经网络模型。利用该模型可以对一定降雨条件下的交流棒-板短空气间隙击穿电压进行预测,预测结果满足精度要求,同时,该文根据建立的人工神经网络模型模拟了降雨时单个及多个环境因素对空气间隙击穿电压的影响,并对模拟结果进行了分析,结果表明:大气压强一定时,随着降雨强度、雨水电导率的增加以及环境温度的降低,空气间隙的击穿电压随之降低;当降雨强度、雨水电导率和环境温度其中任一环境因素改变时,另两个因素对空气间隙击穿电压的影响程度也随之改变。人工神经网络模型对训练数据的依赖较大,对训练范围以外的数据预测精度较差。
To obtain a mathematical model capable of predicting the breakdown voltage of air-gap in rain conditions,we adopted results gained in the artificial climate chamber to establish an artificial neural network（ANN） model of AC discharge voltage of rod-plane short air gap and breakdown voltage of air gap under rain conditions.With the help of this model,the AC discharge voltage of rod-plane short air gap under certain rain conditions can be forecasted,and the accuracy of prediction results is under the required range.Simultaneously,based on the ANN model,the effect of single and multiple factors on the breakdown voltage of air gap under rain conditions is simulated and analyzed,showing that,with the increment of rainfall intensity,rain conductivity and the decrement of ambient temperature,the breakdown voltage of air gap will reduce if the atmosphere pressure is constant;once varying one of the three factors,the impact that the other two factors exert on the breakdown voltage of air gap will change as well.Training data play an essential role in the prediction ability of ANN,which leads to an unaccepted error for the data out of the training range.
出处
《高电压技术》
EI
CAS
CSCD
北大核心
2012年第1期102-108,共7页
High Voltage Engineering
基金
国家重点基础研究发展计划(973计划)(2009CB724501/502)~~
