A novel method for optimizing the condition of UV luminescence of microwave light emitting plasma based on neural network

Abstract The microwave light emitting plasma (MLEP) is an important environmentally friendly UV light source. Finding the optimal condition (pressure, ratio and microwave power) of a MLEP spectral line which makes its intensity strongest is the key work to improve its efficiency in applications, but it has not been studied under the condition with multiple variables yet. In this paper, taking Ar+N2 MLEP as an example, a novel method for optimizing the condition of MLEP by employing ANN to model the relationship between the condition and target spectral line intensity using experimental optical emission spectra data is proposed for the first time. The evaluation parameters showed that the ANN has good prediction accuracy. The optimal pressure, Ar concentration and microwave power were simultaneously predicted for UV spectral lines of 281.13 nm, 295.27 nm, 308.91 nm, 313.14 nm, 336.14 nm and 353.00 nm of Ar+N2 MLEP. Their intensities were increased by 53%, 22%, 64%, 17%, 20% and 22% respectively under the predicted optimal condition compared with the maximum intensity in pure experimental method. Therefore, the proposed method is feasible for optimizing the condition of UV luminescence of MLEP, and has great significance in accurately improving the specific spectral line intensity of light sources.