基于ART2A-E的交叉互联XLPE电缆绝缘在线诊断技术研究  

On-line Insulation Diagnosis Technique of Cross-bonding XLPE Cable Based on ART2A-E

作　　者：张海龙[1] 关伟民[2] 关根志[2] 

机构地区：[1]国网电力科学研究院,湖北武汉430074 [2]武汉大学电气工程学院,湖北武汉430072

出　　处：《高压电器》2010年第9期48-52,共5页High Voltage Apparatus

摘　　要：研究110 kV及以上电压等级的XLPE电缆绝缘的在线监测与诊断技术,确保电缆的安全可靠运行具有重要意义。笔者根据屏蔽层交叉互联接地XLPE电缆在线监测的特点,提出了基于ART2A-E(Adaptive Resonance Theory)神经网络的交叉互联电缆绝缘在线诊断方法。首先,通过分析电缆绝缘老化的特点以及电缆加速老化的实验结果,建立交叉互联XLPE电缆的仿真模型。然后,计算首末端A、B、C三相接地线电流与电缆初始安装时的接地线电流幅值和相位的相对劣化度得到12个特征量,并以此特征量作为ART2A-E的输入样本进行模式识别。仿真实验结果表明,ART2A-E神经网络可以识别发生在交叉互联中的任意一段电缆的绝缘故障,为交叉互联电缆的绝缘在线诊断开辟了新的途径。According to the on-line monitoring character of the cross-bonding XLPE cable,this paper presents a new on-line insulation diagnosis technique based on ART2A-E neural network for the cross-bonding XLPE cable.First,a simulation model of cross-bonding XLPE cable is established by analyzing the aging character and the experimental result of accelerated ageing of the XLPE cable insulation.Then through calculating the original and present current amplitudes and current phases of three XLPE cable grounding lines at initials and terminals,12 characteristic quantities are obtained to serve as the input sample of ART2A-E neural network for pattern recognition.The simulation results show that an insulation fault located at any section of the cross-bonding XLPE cable can be identified through ART2A-E neural network,and the present technique exhibits higher accuracy and efficiency than other methods for on-line insulation diagnosis of cross-bonding XLPE cable.

关 键 词：XLPE电缆 交叉互联 ART2A-E 绝缘 故障诊断 

分 类 号：TM247[一般工业技术—材料科学与工程]