A prediction for breakdown voltages in supercritical CO2 using artificial neural network

The measurements for breakdown voltages have been made with point-plane electrodes for high-pressure carbon dioxide up to supercritical conditions at different temperatures. The breakdown voltages depend on electrode gap, temperature and pressure of gas while preserving the intrinsic nonlinear combination of these characteristics. Artificial neural network was used to model the complex nonlinear relationship. The predicted breakdown voltages using neural network have been compared with the measured ones.