An improved FOA-GRNN integrated in health estimation for RF electronic system

As the core part of the RF circuit, RF amplifier has attracted much attention due to ever increasing demand on the enhancement of safety and reliability. In order to guarantee the RF amplifier safety and implement condition-based maintenance, the health states of RF amplifier circuit should be estimated accurately. This paper illustrates the working principle and then the performance degradation is analyzed by using S-parameter of the RF amplifier. In this paper, a generalized regression neural network with improved fruit fly optimization algorithm (improved FOA-GRNN) is developed to estimate the health state. The traditional FOA-GRNN has an inherently excellent capability of nonlinear mapping, but susceptible to getting stuck in local minima. In order to improve the performance, the width coefficient of FOA-GRNN is redesigned. Finally, the health state estimation result of improved FOA-GRNN is compared with other classic neural networks to validate the performance of this proposed method.