Please use this identifier to cite or link to this item: https://idr.l2.nitk.ac.in/jspui/handle/123456789/17112
Title: Theoretical and Experimental Investigation of Intelligent Non-Linear Controls for Magnetorheological Elastomer based Vibration Systems
Authors: Kumar, Susheel.
Supervisors: Murigendrappa, S M.
Gangadharan, K V.
Keywords: Department of Mechanical Engineering;Magnetorheological elastomer;Vibration isolation;Vibration absorber;sliding mode control;Model-free;Neural network;Fuzzy system;Adaptive control
Issue Date: 2021
Publisher: National Institute of Technology Karnataka, Surathkal
Abstract: Magnetorheological elastomer (MRE) based semi-active isolators and absorbers are prominently used to reduce undesirable vibration for a wide range of operating frequencies. The real-time implementation of these systems requires the controllers to vary the electric current of the electromagnet. Previous researchers have focused on the model-based and fuzzy controllers to control input current. However, due to the viscoelastic behavior of MRE, it exhibits nonlinearities and time-varying properties. This behavior makes the real-time implementation of the MRE-based vibration isolator and absorber less effective with existing controllers. The present work concentrates on the theoretical and experimental investigation of the intelligent adaptive nonlinear controls on the MRE-based vibration isolation and adaptive tuned vibration absorber (ATVA). For the implementation of the controllers, a thorough knowledge of the field-dependent properties of MRE is studied using an in-house custom-made dynamic characterization setup. The dynamic characterization of the MRE is carried out for variable input frequency, displacement and magnetic field. Further, the Bouc-Wen model is employed to comprehend constitutive the relationship between the individual parameters. The characterized MRE is used in the MRE vibration isolator. The properties of the MRE vibration isolator are extracted from shift frequency data. Furthermore, the performance of the MRE vibration isolation is investigated for the model-based PID and LQR controllers. MRE exhibits nonlinearity and time-varying properties that limit the application of linear controllers. To overcome the limitations of the linear controllers, the nonlinear and intelligent controls based on neural networks and fuzzy systems are designed. The designed boundary sliding mode control (BSMC) and neural network-based adaptive observer neural network fuzzy sliding mode control (NNAONFSMC) are implemented on the MRE vibration isolation system. The Lyapunov theorem assesses the asymptotical stability of the designed observer and controls. The controllers' effect is compared without and with parameter uncertainties of MRE vibration isolation at the single frequency excitation. Further, The NNAONFSMC has been analyzed for variable excitation frequency, and the maximum percentage reduction of the measured ii acceleration is 34%. From these outcomes, it is evident that the NNAONFSMC is more effective with time-varying parameter uncertainties of MRE than the BSMC at single and variable frequency excitation. Furthermore, the performance of model-free adaptive fuzzy sliding mode control for the magnetorheological elastomer-based adaptive tuned vibration absorber (MRE ATVA) has been investigated. MRE ATVA is fabricated with anisotropic MRE. Sliding mode and adaptive fuzzy sliding mode controls have been developed. The boundary layer is applied to the sliding surface to reduce the chattering effect in the sliding mode control. In the adaptive fuzzy sliding mode control, two fuzzy systems approximate the equivalent control and switching control. The Lyapunov theorem evaluates the asymptotical stability of the developed adaptive control based on fuzzy systems. The performance is compared for both the controls subjected to single-frequency excitation. Further, the adaptive fuzzy sliding mode control has been investigated for variable frequency excitation. The maximum reduction of transmissibility of primary mass is 38.14%. Based on the results, the model-free adaptive fuzzy sliding mode control is more effective in tuning the natural frequency of MRE ATVA by 0.5 s with parameter uncertainties and under variable frequency excitation compared to the boundary layer sliding mode control.
URI: http://idr.nitk.ac.in/jspui/handle/123456789/17112
Appears in Collections:1. Ph.D Theses

Files in This Item:
File Description SizeFormat 
Susheelkumar_145057ME14F15_Thesis(Final -05-08-2021).pdf14.82 MBAdobe PDFThumbnail
View/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.