Load Reconstruction Based on a Parametric System Identification Approach

The aim of many techniques on the diagnostics of mechanical structures is to determine their technical state and, if found to be unsatisfactory, to localize faults and assess their severity. Conventional diagnostic systems provide techniques that use parameter-based or output-based approaches to detecting structural changes in a system. This paper considers a method for the reconstruction of sine-wave excitation using an inverse data-driven model. Model inversion is feasible if the model structure is known, if its parameters have already been identified, and if the system is time-invariant and linear. Numerical simulations and experimental investigations have been performed to show the feasibility of the proposed approach. Various system identification and inversion scenarios have been carried out for SISO and MIMO models to examine the selected model structures and to reconstruct the load. Examples are provided to illustrate issues that may affect potential application, e.g. the sensitivity of the inversion method to sensor location, model structure, and the system identification method.