Evaluation of Flashover Voltage Mechanisms on Hydrophobic Polymer Insulators with Artificial Neural Network Approach

This paper presents an experimental measurement of ac 50 Hz flashover voltage (kV) of hydrophobic polymer insulators. Hundred thirty five different testing conditions were used to evaluate the electrical performance of hydrophobic surfaces of composite polymer insulators. The study of flashover voltages depend on the silicone rubber (SiR) content (%) in Ethylene propylene diene monomer (EPDM) rubber, water conductivity (μS/cm), volume of water droplet (ml) and number of water droplets on the surface of polymer insulators. Artificial neural network (ANN) is used successfully to model nonlinear functions which are difficult to model using classical methods. ANN can estimate the values of flashover voltage (kV) for different polymer insulators. The proposed network is trained using different environmental wet condition such as; water conductivity, volume of water droplet and number of water droplets on the surfaces of composite different polymer. After training, the network can estimate the flashover voltage for different inputs. A comparison between the laboratory measurements of flashover voltages and computational results of ANN were convergent. The results obtained from applying ANN show that it can be used to model the data with accuracy of 96%. These results prove that ANN can be considered a successful model to evaluate the electrical performance of hydrophobic polymer insulators and predicts the best hydrophobic composite surface that withstands higher flashover voltage under wet contaminated weather condition.