< Back to previous page

Project

Development of digital twins of mechatronic systems for condition monitoring

The focus of this PhD track is on the development and deployment of Digital Twins (DTs) in service to condition monitoring and prognostics. In the era of Industrial Internet of Things (IIoT) and Industry 4.0, mechatronics systems such as wind turbines and transmissions will be equipped with a variety of sensors providing new information for the development of novel designs, monitoring systems and controllers, aiming at higher machine performance in terms of durability, noise emissions, energy, maintenance plan, operational costs. Alongside with sensors data, (physics based) models can be employed to infer relevant quantities at locations where a direct sensor cannot be placed, such as bearing and gear loads, system parameters, generation and evolution of defects. By combining models and measurements, novel approaches based on the concept of the DT will be developed, focusing on system level modelling, load and parameter estimation, crack estimation, upscaling (small scale lab tests to predict the behavior of the full size components/systems) and deployment on an edge system or in the cloud or in a hybrid/distributed manner. Several challenges are present to make DT techniques a success for the purposes mentioned before. Firstly, novel modelling approaches should aim at accurate representations of the physical phenomena by taking into account cheap permanent measurements (e.g., strain gauges and MEMS accelerometers). In parallel to this, models should be as simple as possible in order to be deployable online (or real time) on embedded systems. Moreover, efficient mathematical representations of the unknown quantities need to be developed (loads, parameters, operating conditions, defects). This requires to take into account any known time and/or frequency domain information such as cyclostationarity, periodicity, sparsity. Furthermore, the new quantities should become part of new algorithms for condition monitoring, prognostics, design and control. Finally, the new modelling tools and algorithms should run on an embedded system (with the possibility to share some data on a cloud for remote monitoring of fleet monitoring).

Date:19 Oct 2020 →  Today
Keywords:Condition monitoring, Compressive Sensing, Wind turbines
Disciplines:Wind energy, Acoustics, noise and vibration engineering, Dynamics, vibration and vibration control, Sensing, estimation and actuating
Project type:PhD project