A new combination forecasting model for ESDD prediction of suspension porcelain insulators

Abstract

Leakage current that is strongly relative to the contamination flashover of outdoor insulators easily monitors online. In the paper, a BP Artificial Neural Network (ANN) model is proposed to predict the equivalent salt deposit density (ESDD) value based on the leakage current characteristics and environmental parameters which are both from artificial pollution tests. The contamination degree of insulators online monitoring is realized by using the proposed model. The data reflecting the relationship between the ESDD and the time is obtained from the natural contamination tests, and then the other ESDD prediction model is established based on the non- linear time series through analyzing ESDD time series properties. The second proposed model has high prediction precision, but in practical engineering, the data of ESDD can be obtained is very limited and need to be measured under a power outage condition. In order to get the better precision and practicality of the prediction, a new combination forecasting model is built, which the ESDD nonlinear time series prediction model is a correction model for the ANN model. The combination forecasting model is verified by part of experimental data, and the results show that higher precise online monitoring of contamination degree of insulators is achieved, the error is within ±4.3% and the prediction precision is 54% higher than that of the ANN model.