Non-linear modeling of a twin rotor system using particle swarm optimization

The paper presents a nonlinear modeling technique for an air vehicle system, called Twin Rotor Multi-input Multi-output System (TRMS) using neural networks and particle swarm optimization (PSO). Since the TRMS permits both 1 and 2 degrees of freedom (DOF) motions, it can be considered as a static test rig for an air vehicle. Modeling of TRMS is perceived as a challenging engineering problem due to its nonlinear aerodynamics and cross coupling effects between horizontal and vertical channels. Firstly, a feedforward neural network with conventional training algorithm is designed to capture the dynamics of both channels. Since conventional training algorithm, such as backpropagation is often trapped in local minima, a relatively recent bio-inspired search PSO is employed to overcome the problem. Results show that combination of the feedforward neural network and PSO is very effective in modeling systems with high nonlinearity and complex characteristics. A comparative assessment and a number of validation tests are provided to verify the proposed approach.