Optimization of Behavioral Model of VO2Switches Using Slime Mould Algorithm

Abstract

A systematic optimization of parameters related with Artificial Neural Network (ANN) is absolutely necessary to extract the best possible optimized ANN, therefore gaining traction for the development of behavioral models for advanced Radio Frequency (RF) and microwave components in the wireless industry. This paper develops and demonstrates a hybrid Slime Mould Algorithm (SMA)-ANN based modelling approach for fully printed Vanadium Dioxide (VO <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> ) RF switches, which constitute a pivotal part for next-generation reconfigurable components. At first, ANN using cascade-forward neural network architecture is exploited to develop behavioral model for VO <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> switch. Thereafter, parameters of ANN are tuned with SMA optimization algorithm. Finally, both ANN and hybrid SMA-ANN approaches are compared with conventional regression-based metrics namely mean squared error, mean absolute error, coefficient of determination, simulation time, parameters' tuning time, complexity of the models and ability of the models to predict on untrained data to establish the pros and cons of each approach.