Hybrid Adaptive Filters for Dynamic Mass Measurement of Moving Objects on Flexible Platforms

Abstract

Dynamic mass measurement of moving objects on flexible platforms remains a significant challenge due to mechanical vibrations, structural flexibility, and the inherently low Signal-to-Noise Ratio (SNR) of strain gauge sensors. Traditional analog filtering approaches, such as Tow– Thomas active filters, exhibit limitations since gain stages propagate and may even amplify intrinsic noise, compromising metrological accuracy. This work proposes a hybrid adaptive filtering methodology that integrates a passive input stage, a four-Degree-Of-Freedom (4-DOF) dynamic platform model, and a digitally implemented adaptive Tow–Thomas IIR filter with iterative SNR-based prediction. The passive stage operates as an initial low-pass filter, attenuating high-frequency electromagnetic interference (≈ 724 kHz) and ensuring proper conditioning of the signal for subsequent A/D conversion. The 4-DOF model describes the coupled effects of translational and rotational dynamics of the platform, enabling the use of the Random Decrement technique for modal frequency identification. Finally, the adaptive digital Tow–Thomas filter iteratively updates its coefficients until the specified SNR is achieved, guaranteeing unitary gain and preventing spectral distortions. Simulation results and industrial checkweigher measurement datasets demonstrate that the proposed structure enhances system robustness and improves noise rejection while preserving accuracy in the dynamic quantification of moving masses.