Tordela, Ciro (2022) MODEL-BASED APPROACH FOR MECHANICAL SYSTEM MONITORING. [Tesi di dottorato]

Anteprima

Testo tordela_ciro_35.pdf Download (9MB) | Anteprima

Abstract

The aim of the present Ph.D. thesis is the development of model-based estimators for monitoring mechanical systems. Complex systems have mechanical ones as components, and those latter are intrinsically complex. Indeed, the growth in technology allowed the transformation from purely mechanical systems to mechatronics ones with many advantages in terms of interfacing with other systems, the external environment, and humans. Unfortunately, higher maintenance of components integrated into new mechanical systems, typically subjected to degradation, is required. The possibility of introducing the condition-based approach to maintenance activities is crucial for avoiding early replacements of components in good functioning or late intervention on them in faulty conditions. Different techniques for monitoring mechanical systems in real-time can be employed for realizing the condition-based maintenance. In this work, model-based estimators constituted of Kalman Filters are employed for monitoring three types of mechanical systems: the railway vehicle, the road vehicle, and Curved Surfaces Sliding Isolators. The monitoring through a model-based approach for each class of previously mentioned mechanical systems is described. Anti-yaw suspension components, which constitute a part of the railway secondary suspension, are monitored to identify possible faults that cause stability and safety reduction in railway vehicles. Two different modelling approaches are employed for monitoring the tire-road conditions of road vehicles and for managing their performances by estimating the sideslip angle. The frictional behaviour related to both surfaces of Curved Surfaces Sliding Isolators is characterized through the proposed model-based approach, which is also suitable for monitoring the wear conditions of isolators during their operations. An overview of different possible approaches to the diagnostic and monitoring of mechanical and mechatronic systems, functional for condition-based maintenance, is provided. In particular, a detailed description of Kalman Filters, employed as a model-based monitoring technique in this work, is included. By starting from the linear Kalman Filter, nonlinear formulations of this latter, such as the Extended Kalman Filter and the Constrained Unscented Kalman Filter, are explained. Kalman Filters make estimations based on the mathematical modelling of the system to be monitored. For each mechanical system studied in this work, an estimator design model is developed to include it in a Kalman Filter for activating the estimation process and, therefore, the model-based monitoring. The correct design of the previously mentioned model is crucial for obtaining reliable estimations by Kalman Filters. Formulations of estimator design models able to capture desired dynamical behaviours of mechanical systems to be monitored are provided. Finally, results concerning estimations provided by the proposed monitoring approach for each mechanical system analysed are provided. The estimated quantities are compared with detailed simulation models and with experimental data. The obtained results confirm the suitability of the model-based monitoring approach for mechanical systems, allowing for deepening future research on their applicability in hardware equipment integrated onboard the explored mechanical systems for making real-time condition monitoring.

Downloads

Actions (login required)

Modifica documento