Digital image processing for non-linear system identification

Abstract Emerging digital image processing techniques demonstrate their potential applications in engineering mechanics, particularly in the area of system identification involving non-linear characteristics of mechanical and structural systems. The objective of this study is to demonstrate the proof-of-concept that the techniques permit the identification non-intrusively and remotely. First, the efficacy of the digital image processing method is shown by identifying the friction behavior between two solids that are assumed to be governed by the Coulomb friction model. The inverse analysis is carried out to validate the proposed method of identifying model parameters. Studies further illustrate that the digital imaging procedure and inverse analysis algorithms developed for the friction problem can be extended for identification of non-linear mechanical and structural characteristics. One illustration utilizes the second example, where relative motion between a model structure and a shaking table is measured by digitally processing the analogue image provided by a video tape recorded during a shaking table test of a model structure base isolated by a hybrid isolation device consisting of friction and elastomeric components. Third, constitutive relationship for a non-linear elastomeric membrane is identified by digitally processing images of its deformed states under tension. The relationship is postulated to follow the Mooney–Rivlin function. Schemes developed are verified with iterative non-linear finite-element program that is valid in the finite deformation range. Emerging cost-effective hardware and software systems for high-performance data acquisition and processing are quite promising to the implementation of the techniques by removing most, if not all, of its existing limitations.