A regime-based nested CV machine learning model to forecast the magneto-induced mechanical characteristics of a magnetorheological fluid

Abstract Magnetorheological fluid (MRF) is an intelligent fluid that has a lot of promise in the automation industry. It is well established from earlier research that the magneto-induced yield stress primarily determines MRF's load-bearing capacity. However, the existing experimental and theoretical approaches to measure the magneto-induced yield stress of MRF have limitations. Without making any assumptions, a data-driven machine learning model can learn nonlinear relationships between input parameters and observed attributes from data. This method makes it possible to model with accuracy, interpretability, and uncertainty awareness even in situations when there are few experiments and the system is very non-linear. The objective of this study is to generate a hybrid physics-informed data-driven ML model to predict the magneto-induced yield stress with raw and non-linear input features. The experimental findings for the MRF for different material ratios evaluated over the rotational rheometer (Anton Paar) are gathered for the hybrid physics-informed data-driven analysis.Three distinct machine learning (ML) models-Support Vector Machine (SVM), Random Forest (RF), and Gaussian Process Regression (GPR) -were applied to rheological experimental data of the MRF of different compositions to predict the magneto-induced yield stress of MRF. Using performance analysis, we demonstrated that the ML models can reliably forecast the magneto-induced yield stress of MRF based on material, rheological, and electromagnetic test variable data. It was evident from a comparison of model results that the GPR performed better in predicting yield stress than other ML models. According to a further investigation of permutation feature importance, strong inter-feature correlations and nonlinear interactions seen in magnetorheological systems are reflected in the comparatively large error bars as compared to the mean permutation importance, which shows that the contribution of individual features changes across permutations.

Authors

Institutions

Publication Details

Journal
Physica Scripta
Published
2026-11-17
DOI
https://doi.org/10.1088/1402-4896/aec334
Primary Topic
Vibration Control and Rheological Fluids
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

A regime-based nested CV machine learning model to forecast the magneto-induced mechanical characteristics of a magnetorheological fluid

Danishtah Quamar, Chiranjit Sarkar
1 links
Physica Scripta
Vibration Control and Rheological Fluids
article

A regime-based nested CV machine learning model to forecast the magneto-induced mechanical characteristics of a magnetorheological fluid

Danishtah Quamar, Chiranjit Sarkar
article en

Abstract

Abstract Magnetorheological fluid (MRF) is an intelligent fluid that has a lot of promise in the automation industry. It is well established from earlier research that the magneto-induced yield stress primarily determines MRF's load-bearing capacity. However, the existing experimental and theoretical approaches to measure the magneto-induced yield stress of MRF have limitations. Without making any assumptions, a data-driven machine learning model can learn nonlinear relationships between input parameters and observed attributes from data. This method makes it possible to model with accuracy, interpretability, and uncertainty awareness even in situations when there are few experiments and the system is very non-linear. The objective of this study is to generate a hybrid physics-informed data-driven ML model to predict the magneto-induced yield stress with raw and non-linear input features. The experimental findings for the MRF for different material ratios evaluated over the rotational rheometer (Anton Paar) are gathered for the hybrid physics-informed data-driven analysis.Three distinct machine learning (ML) models-Support Vector Machine (SVM), Random Forest (RF), and Gaussian Process Regression (GPR) -were applied to rheological experimental data of the MRF of different compositions to predict the magneto-induced yield stress of MRF. Using performance analysis, we demonstrated that the ML models can reliably forecast the magneto-induced yield stress of MRF based on material, rheological, and electromagnetic test variable data. It was evident from a comparison of model results that the GPR performed better in predicting yield stress than other ML models. According to a further investigation of permutation feature importance, strong inter-feature correlations and nonlinear interactions seen in magnetorheological systems are reflected in the comparatively large error bars as compared to the mean permutation importance, which shows that the contribution of individual features changes across permutations.

Physica Scripta
Indian Institute of Technology Patna (IN)
Vibration Control and Rheological Fluids
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon