摘要
为准确评估城市轨道供电环网XLPE绝缘的老化状态,提出了一种基于科尔模型及多分类支持向量机的老化评估方法。首先,设计加速老化实验并制备了不同老化阶段的XLPE样品,基于介电测试和拉伸测试获取原始数据,以断裂伸长率为依据提出了用于表征老化状态的评估因子;然后,基于科尔模型提取介电特征参数,使用“一对其余”的多分类SVM方法进行老化状态分类;最终,通过训练集和验证集的分类结果论证了所提方法的准确性,并经现场应用测试证明了其有效性。该方法不需对实测样品进行破坏性加工或取样,就可实现无损的老化状态评估,适用于城轨供电环网XLPE绝缘的老化评估,能为日常运维及现场检修提供可靠参考。
In order to accurately evaluate the aging state of XLPE insulation in urban rail power supply ring network,an aging evaluation method based on Cole model and multi classification support vector machine is proposed.Firstly,the accelerated aging experiment was designed and XLPE samples at different aging stages were prepared.Based on the original data obtained from dielectric test and tensile test,the evaluation factors used to characterize the aging state were proposed based on elongation at break.Then,the dielectric characteristic parameters are extracted based on Cole model,and the aging state is classified by"one to the rest"multi classification SVM method.Finally,the accuracy of the method is demonstrated by the classification results of training set and verification set,and the effectiveness of the method is proved by field application test.The method proposed in this paper does not need to destructively process or sample the measured samples,and can realize the nondestructive aging state evaluation.It is suitable for the aging evaluation of XLPE insulation of urban rail power supply ring network,and can provide a reliable reference for daily operation and maintenance and on-site maintenance.
作者
李海岳
LI Haiyue(China Construction Third Engineering Bureau Installation Engineering Co.,Ltd.,Wuhan 430000,China)
出处
《电工技术》
2022年第21期193-195,共3页
Electric Engineering