Classification of Electrical Treeing Images by Machine Learning of Supervised and Unsupervised Learning

In recent years, machine learning has been widely applied in the field of materials such as material informatics. There are several examples its application to the development of insulating materials. The machine learning is good at classification, judgment, and prediction. However, there are few examples of machine learning applied to various discharge experimental data in insulating materials. Therefore, we would like to predict the progress of electrical tree in silicone gel by using our data. In this paper, general machine learning was applied to consecutive images of treeing phenomena. These images can be classified roughly 2 categories. Also, the two types of supervised and unsupervised learning were applied. As a result, classification was possible in both learning. In addition, we found that insight the possibility of predicting the progress of electrical tree.