A MLP Neural Network-Based Unified Intelligent Model for Predicting Short-Gap Breakdown Voltages in Air

The breakdown voltage characterizes the required voltage at which an electrical insulator will become a conductor of electricity.Measurements of short air gap discharge voltages are essential to guarantee safety, compliance and reliability in electrical energy transmission and transformation networks.After the implementation of three distinct neural models for three short air gap configurations prediction: Electrode-Sphere-Sphere-Earth (ESSE), Electrode-Sphere-plane (ESP), and Electrode-Symmetry-Sphere (ESS); the present research introduces an innovative compact and intelligent architecture to predict the breakdown voltages of any short air gap configuration (without having to resort to reconfiguring the system each time), based on a universal automatic multi-layer neural network (MLP) repressor.The unified architecture receives as input the radius, distance and indicator of the configuration and returns as output the predicted breakdown voltage.Our system has been validated on three configurations (with several internal diameter electrodes) and a fairly rich database (taken from the literature) with a Mean Square Error (MSE) of around 11.84%, which certify the yield of the methodology undertaken.