Prediction Model for Dissolved Gas in Transformer Oil Based on Non-Parametric Regression

Abstract

The non-parametric regression prediction model for dissolved gases in power transformer and its application are studied. As the intervals between two analytic experiments of transformer dissolved gas are unfixed,the data sequence sampled with unequal intervals is converted into the data sequences with equal intervals,which is smoothed to form a new sequence. And then use the historical samples data to establish non-parametric regression model for prediction. Compared with the grey model,the non-parametric regression model has better prediction accuracy. The case verifies the correctness and feasibility of the method.