Modeling temporal and spatial evolutions of laser-induced plasma characteristics by using machine learning algorithms

Abstract

It is well-demonstrated that laser-induced plasma (LIP) is a transient phenomenon in which plasma characteristics vary with space and time. The study of the spatial and temporal distribution of the temperature and electron number density of LIP can elucidate the process of plasma expansion and plasma interaction. Therefore, establishing an experimental-based LIP model would open a new era in the quantitative explanation of the LIP and its applications. To this end, two popular machine learning algorithms, i.e., ANN, and ANFIS, are proposed in this study for modeling the cadmium plasma characteristics under the local thermodynamic equilibrium (LTE) conditions. The main aim of these models is to predict the plasma temperature and electron number density (representing the LIP characteristics) in terms of the influencing parameters. The accuracy and generalization capability of the developed models are evaluated by measuring performance functions for the training and testing data. The results show that both machine learning algorithms can successfully cope with the complexity of LIP variations over time and space. Moreover, the well-trained models are used to investigate the temporal and spatial evolution of LIP by calculating the temperature and electron number density in ranges of the influencing parameters.