Estimation of Simulated Particles in Magnetron using Image Color Detection

Although Particle-In-Cell (PIC) simulations are widely used to simulate high power microwave (HPM) devices, they have inherent limitations such as long runtimes, lack of optimization features, etc. This paper aims at exploring the possibility of using artificial intelligence and machine learning tools either as an assistance to PIC simulations towards prediction and optimization of results or as an alternate to PIC after proper training and testing with simulated data. In this work, number of simulation particles (electrons) present in a PIC simulation at any time instant are estimated using image processing algorithms by identifying the color percentage in the images generated via PIC simulations. This color percentage is then used to estimate the number of simulation particles in the simulation through linear fitting. Accuracy of different methods for particle estimation is also compared.