Bouc-Wen model parameter identification for a new magneto-rheologigal fluid damper using particle swarm optimization

• Main Author: Mohd Azraai, Mohd Razman
• Format: Thesis
• Language: English
• Published: 2014
• Subjects: Q Science (General); TJ Mechanical engineering and machinery
• Online Access: http://umpir.ump.edu.my/id/eprint/13165/1/Bouc-Wen%20model%20parameter%20identification%20for%20a%20new%20magneto-rheologigal%20fluid%20damper%20using%20particle%20swarm%20optimization%20.pdf; http://umpir.ump.edu.my/id/eprint/13165/

In constructing a reliable semi-active suspension system, the modelling of the damper is imperative as it produces the controllability on the suspension system. The modelling of Magneto-rheological (MR) fluid damper for the control device has been major focuses throughout the decades as semi-active systems are deems to be efficient in vibration suppression for various applications. MR fluid damper is abided by the behaviour of hysteresis model that not just predict the subsequent impact, but has the ability to retract the motion by the model internal memory. Acquiring a suitable model comes a setback from the natural existence of non-linearity from the MR fluid damper as the parameters of the hysteresis model may requires tuning as the response time for the absorber to response are in milliseconds. Hence, Particle Swarm Optimization (PSO) was introduced for altering significant parameters for Bouc-Wen hysteresis model to replicate the MR fluid damper performance in real-time. The objectives are succinct in three main criteria starting with the development of MR fluid damper, then a representation of hysteresis model and lastly optimizing these parameters by inducing PSO algorithm. Validations by physical experiment and simulation were conducted to enhance the justification of the present model. These performances are measured in force against displacement and force against displacement for the hysteresis model to depict MR fluid damper behaviour. The average marginal error was presented to strengthen the model along with analysis and discussion in deliberating the outcome. Approximation of the model demonstrates dependable fitting compared to the experimental data with the average marginal error ranging from 6.0 % to 8.3 %. The findings suggest that several parameters of the hysteresis systems requires boundary and by imposing the known sensitive variables to the model can emulated into near perfect model.