Plasma Real-time Diagnostics Based on Visual Image Recognition
Ziheng Yu, Shuai Yang, Chuan Li, Yong Yang · 2021 IEEE 4th International Electrical and Energy Conference (CIEEC) · 2021
Real-time diagnostics has always been a key point of plasma research. Optical emission spectroscopy is the most commonly used method to obtain the plasma parameters through fitting of the emission curves. But the process is time-consuming, making it almost impossible to get the data in real-time. In this paper, we propose a method using image recognition for real-time diagnostics. The relationship between visual images and rotational and vibrational temperatures was established. A convolutional neural network model was used for image recognition. After training, the test accuracy was over 0.99. The model was proved to be able to predict the temperature of a new input image, indicating potential applications of image recognition in real-time plasma diagnostics.