基于频域介电谱曲线分解的氧化锌避雷器老化状态评估

Method for Evaluating the Aging State of ZnO Arrester Based on Curve Decomposition of Frequency Domain Dielectric Spectrum

氧化锌避雷器是电力系统的重要组成设备,其老化状态直接关系到对过电压的防护效果,开展氧化锌避雷器的老化状态评估具有重要意义。该文针对氧化锌避雷器内部不均匀老化分布,提出一种基于频域介电谱曲线分解的氧化锌避雷器老化状态评估方法。首先,通过构建氧化锌避雷器介电响应等效电路,研究了氧化锌避雷器频域介电谱分解方法,根据整支避雷器频域介电谱分解可以得到避雷器内部各区域氧化锌阀片的频域介电曲线;其次,利用Havriliak-Negami方程对氧化锌阀片频域介电谱特征参数进行了提取,基于极限学习机建立了氧化锌阀片老化状态评估模型,提出氧化锌避雷器老化评估方法流程;最后,通过多重雷击模拟实验,制备了不同老化状态的氧化锌阀片与避雷器实物样本,对所提老化评估方法进行了分析与验证。结果表明,整支避雷器频域介电谱分解与测试结果的平均误差小于3%,所提方法能够有效地对避雷器频域介电谱进行分解并获取各区域的频域介电谱;针对整支避雷器,该文所提方法对各区域老化状态的评估准确率为93.3%,能够有效地对避雷器进行老化评估。

ZnO arrester is an important component of power system equipment,which protects electrical equipment from lightning or overvoltage caused by operation.During the service process of arrester,under the effect of overvoltage such as long-term lightning strike and operation,it gradually ages,leading to the decline of current carrying capacity and affecting its protection effect against overvoltage.Therefore,it is necessary to effectively evaluate the aging state of arrester,so as to guide the maintenance and replacement,and ensure the safe and stable operation of power system.At present,the aging status assessment of ZnO arrester is mainly divided into two ways:online monitoring and offline detection.In the field application of online monitoring method,it is inevitable to be interfered by some on-site objective factors,resulting in certain errors in the evaluation results.And in the offline detection of arresters on-site,most of them need high-voltage equipment,which is bulky and inconvenient to carry,and the on-site experimental conditions are difficult to fully ensure the safety of high-voltage experiments.Therefore,aiming at the uneven aging distribution inside the ZnO arrester,this paper proposes an aging state evaluation method of ZnO arrester based on the decomposition of frequency domain dielectric spectrum curve.Firstly,by constructing the equivalent circuit of the dielectric response of ZnO arrester,the frequency domain dielectric spectrum decomposition method of ZnO arrester is studied.Through the frequency domain dielectric spectrum decomposition of the whole arrester,the frequency domain dielectric curve of ZnO varistor in each region of the arrester can be obtained.Then,the characteristic parameters of frequency domain dielectric spectrum of ZnO varistor are extracted by using Havriliak-Negami equation,the aging state evaluation model of ZnO varistor is established based on extreme learning machine,and the aging evaluation method of ZnO arrester was proposed.The main process is to determine the equivalent circuit parameters of dielectric response based on the frequency domain dielectric spectrum test results of the whole ZnO arrester,obtain the frequency domain dielectric spectrum decomposition of each region,and The aging characteristic parameters of each region are extracted,and the aging status of each region is evaluated based on extreme learning machine.Finally,through multiple lightning simulation experiments,samples of ZnO varistor and arresters in different aging states were prepared,and the proposed aging evaluation method was analyzed and verified.The results show that for the frequency domain dielectric spectrum of the whole arrester,the average error between the decomposition results and the direct test results is less than 3%,which shows that the proposed method can effectively decompose the frequency domain dielectric spectrum of the arrester,and indirectly proves that the frequency domain dielectric spectrum curve of each region of the arrester can be effectively obtained by curve decomposition;The accuracy of the proposed method is 97.5%on the test set of ZnO varistor after the training of ZnO varistor samples in different aging states,and 93.3%on the aging state of each area of the whole arrester,indicating that the proposed method can effectively evaluate the aging of each area of arrester.