悬垂式玻璃绝缘子覆冰冰凌识别与评估方法研究

Research on Identification and Evaluation Method of Ice-Covered Glacier for Draped Glass Insulator

由于悬垂式玻璃绝缘子覆冰冰凌环境较为复杂、影响因素过多、特征难以提取,导致悬垂式玻璃绝缘子覆冰冰凌识别精准度较低。为此,提出了基于图像分析的悬垂式玻璃绝缘子覆冰冰凌识别与评估方法。采用领域平均法,对采集到的绝缘子覆冰冰凌图像进行平滑去噪处理。按照绝缘子形态参数,建立卷积神经网络层次结构。通过镜面成像原理,捕捉冰凌成像特征。按照现场环境参数,设定成像尺寸。通过换算,识别悬垂式玻璃绝缘子覆冰冰凌。根据历史数据模拟冰凌的风险指标,设置权重参数。通过权重比对,完成评估。试验结果表明,所提方法最低识别误差率为7%、评估均方根误差为0.15。应用该方法的图像更为清晰、细节保留效果更好。该方法能够有效实现悬垂式玻璃绝缘子覆冰冰凌识别,且识别与评估的精准度均较高。

Due to the more complex environment,too many influencing factors,and difficult to extract features of draped glass insulators ice-covered glacier,the ice-covered glacier recognition of draped glass insulators is less accurate.For this reason,a method of identification and evaluation of ice-covered glacier for draped glass insulator based on image analysis is proposed.The domain average method is adopted to smooth and denoise the acquired insulator ice-covered glacier images.A convolutional neural network hierarchy is established according to insulator morphology parameters.The imaging features of glacier are captured through the principle of mirror imaging.The imaging size is set according to the field environment parameters.The ice-covered glacier for draped glass insulator is identified through conversion.The risk indicator of glacier is simulated according to historical data,and the weighting parameters are set.The assessment by weight comparison is completed.The experimental results show that the proposed method has a minimum recognition error rate of 7%and a root-mean-square error of 0.15.The images obtained by applying the method are clearer and the detail preservation effect is better.The method can effectively realize the identification of ice-covered glacier for draped glass insulators,and the accuracy of both identification and assessment is high.