Paper Number
PO84                         My Program 

Session
Poster Session

Title
Challenges in rheometric signal denoising: Limitations of traditional filters and the potential of latent space modelling

Presentation Date and Time
October 22, 2025 (Wednesday) 6:30

Track / Room
Poster Session / Sweeney Ballroom E+F

Authors (Click on name to view author profile)

  1. Das, Mohua (Massachusetts Institute of Technology)
  2. Vadillo, Damien C. (Corporate Research Analytical Laboratory, 3M)
  3. Perego, Alessandro (Corporate Research Analytical Laboratory, 3M)
  4. McKinley, Gareth H. (Massachusetts Institute of Technology, Mechanical Engineering)

Author and Affiliation Lines (in printed abstract book)
Mohua Das1, Damien C. Vadillo2, Alessandro Perego2 and Gareth H. McKinley1
1Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA 02139; 2Corporate Research Analytical Laboratory, 3M, St Paul, MN, MN 55144

Speaker / Presenter 
Das, Mohua

Keywords
artificial intelligence; machine learning; rheometry

Text of Abstract
Accurate rheometric analysis relies on capturing clean time-domain response signals—such as strain (or angular displacement) and stress (or torque)—while preserving their amplitude and phase characteristics. However, in practice, these signals are often contaminated by substantial noise originating from instrumentation or environmental sources, especially when measuring very soft solids or mobile liquids. Traditional denoising methods, including moving average, Savitzky–Golay, Wiener, and wavelet filters, all require extensive parameter tuning and risk altering critical signal features, which can undermine the fidelity of subsequent frequency-domain analysis. These challenges motivate the exploration of data-driven denoising alternatives that can remove artifacts while preserving both temporal and spectral signal integrity. In this poster, we will demonstrate that latent space modeling outperforms traditional denoising methods for rheometric signal denoising. By comparing reconstructed signals from our framework with those produced by conventional filters, we show that this new approach not only reduces noise effectively but also preserves critical signal characteristics in both the time and frequency domains. It delivers good performance in frequency-domain analysis, improving the accuracy of material property evaluations while eliminating the need for manual parameter tuning. This robust approach provides a scalable and efficient solution for establishing a unique signal baseline for any rheometer, as well as for subsequent denoising and processing large rheometric datasets, enabling accurate, high-throughput automated material characterization.