Inhomogeneous plasma electron density inversion based on Bayesian regularization neural network.

Electron density is one of the most important parameters for characterizing plasma properties, so obtaining accurate electron density is a prerequisite for studying the interaction between plasma and the electromagnetic waves. This paper presents the effects of different electron densities on the electric field distribution of a microstrip antenna with a center frequency of 2.45 GHz. Then, on the basis of the integrated model of plasma and the microstrip antenna, the Bayesian regularization neural network (BRNN) is used to retrieve the electron density of inhomogeneous plasma. Furthermore, the performance of the proposed approach is evaluated and analyzed by comparison with Levenberg–Marquardt (LM) and Scaled Conjugate Gradient (SCG) neural networks. The results show that the BRNN provides better performance than LM and SCG neural networks to retrieve plasma electron density based on the electric field intensity at fewer spatial positions. The accurate distribution of the electron density of inhomogeneous plasma can be obtained using BRNN. In addition, the greater the range variation of electron density, the greater the relative inversion error. This study provides an important theoretical basis for the diagnosis of electron density for inhomogeneous plasma in experiments. [ABSTRACT FROM AUTHOR]