An Aging Evaluation Method of Composite Insulators Based on Deep Learning and Decision Tree Model

Composite insulators have attracted more and more attention all over the world because of their excellent insulation performance. With the increase of operating life, long-term exposure to the daily environment will often lead to the gradual decline of its insulation performance. In this paper, the samples of composite insulators with service life of 0-35 years were tested by FTIR, shore hardness and hydrophobicity classification. The results of FTIR show that the peak of Si-O-Si and Si-CH3 are negatively correlated with the aging degree, and the peak of -OH is positively correlated with the aging degree. It can be seen that the three peaks of Si-O-Si, Si-CH3 and -OH can be used as the basis for judging the degree of aging. The shore hardness test results show that the difference between the actual and the standard hardness can be used as the basis for judging aging. In addition, a hydrophobicity classification evaluation method of composite insulators based on deep learning is proposed in this paper. The recognition accuracy can reach 97.6%. Finally, a decision tree model for aging evaluation of composite insulation sheds is established in this paper. The model can evaluate insulators into three grades: A, B, C and the accuracy of the evaluation is 99.8% and 95.6%, which is of great engineering value.