Extreme Learning Machine Based-Shear Stress Model of Magnetorheological Fluid for a Valve Design

Magnetorheological fluid (MRF) models are important tools in the design of the material based-valve in a damper. Although Bingham Plastic and polynomial models are quite widely employed, these data-driven models have disadvantages in term of the accuracy and narrow applicable operating conditions, including magnetic field and shear rate. Therefore, this paper aims to utilize an extreme learning machine (ELM) based-shear stress model to design an magnetorheological (MR) valve. Firstly, an MRF model of MRF 132DG is built using rotational rheometer test results. Secondly, the model is employed to model a meandering MR valve drop pressure utilizing the known design parameters and a finite element magnetic method (FEMM) results. The comparison of the steady state pressure between the simulation and experimental results (in literature) has shown a good agreement in term of the pattern and accuracy with error of less than 3. In summary, ELM has shown its potential to model MRF behavior while employing it to an MR device. © 2020, Springer Nature Singapore Pte Ltd.