Modelling of MR Damper with Adaptive Neural Network and Particle Swarm Optimisation Technique

摘要

This paper presents a novel method for the non-parametric modelling of a magneto-rheological (MR) damper using adaptive neural network (NN) that incorporates a particle swarm optimization (PSO) method. In this approach, the adaptive NN method using adaptive back-propagation (BP) learning algorithm is used to update the weights in real-time. Initial values of the weights and biases are optimized using PSO in an off-line manner. The experimental data were presented in time histories of the displacement, velocity and force parameters measured both for constant and variable current settings and at selected frequency applied to the damper. The model parameters are determined using a set of experimental measurements corresponding to different current constant values. It has been shown that the MR damper model response via the proposed NN approach is in good agreement with the MR damper test rig counterpart.