Mechanical systems virtual sensing by proportional observer and multi-resolution analysis

Abstract A technique to reconstruct the displacement field throughout structure from pointwise measurements under different noise sources is proposed. The developed estimator a Proportional Observer (PO) that exploits linear, frequency independent, relation between estimated state-space vector and measurements. To improve its accuracy, PO concept then extended definition of sequence proportional observers, each one acting on signal decomposition provided by wavelet multi-resolution analysis, or Multi-Resolution (MR-PO). considered numerical test case straight, uniform beam with an unmodeled stiffness reduction notch, which allows for characterizing analytically this model uncertainty. input data given virtual strain collected top face beam, whereas state variables are time dependent coordinates modal expansion vertical along elastic axis. An optimization solver, minimizes estimation error, employed get gain matrix proposed observers. effect sources, like process measurement unknown excitation, accuracy taken into account sensitivity analysis. obtained results assess effectiveness combination analysis as tool developing digital twin models based experimental data.