You are here: Home / Research areas / Smart system dynamics

Smart system dynamics

The smart system dynamics research focusses on the development, implementation/realization and experimental validation of innovative mechatronic concepts that aim at improving the performance of dynamic systems with respect to vibrations, noise emission, safety, throughput often in a trade-off with energy efficiency. This research combines the development of smart systems for specific applications and their experimental validation and industrial implementation, with the development of design, analysis and modelling methodologies and supporting theoretical contributions.

Core activities involve the development of novel modelling and co-simulation approaches for system level analysis together with appropriate identification strategies; of coupled state-disturbance parameter estimation using moving horizon observers, compressive sensing and extended Kalman filtering with both real-time and online applications for blending simulation data with measurement data; of coupling strategies to embed high-fidelity 3D models in a port-based context; and of design space exploration using variable model complexity representations.
A key research line within the above developments combines high-fidelity, time-efficient models with coupled state, input and parameter estimation methods to develop virtual sensors, allowing to measure quantities otherwise difficult or impossible to measure directly.
Application examples of this approach are a stresscamera
(https://youtu.be/pILsapTeoQU) and a virtual torque sensor (https://youtu.be/9-DbplXC_3M). Other applications are online characterisation of bushing or tyre-parameters and energy flow measurements through machines.
Overall target applications are predominantly in the fields of vehicle mechatronics, industrial machinery and intelligent lightweight structures, where the developed technologies allow the generation of new, more accurate, more reliable and/or timely information to boost decision making, control, monitoring, performance, added-value offered, etc.
A specific research activity concentrates on the design of mechanisms and control strategies to high-accuracy pointing in small spacecraft and CubeSats, where rolling element bearings and reaction wheels which are mounted in spacecraft are a source of disturbance in systems which are designed to operate in microgravity and conditions of extremely low vibrations.

Research topics:
  • Dynamic substructuring

Dynamic substructuring is an engineering topic that generally studies the behaviour of mechanical systems by means of coupling their subcomponents. The opposite, where a component is studied that can only be tested when it is on a test rig, is often also useful when the tested component is either very heavy, fragile or part of a drive-train. Current research focusses on FRF-based substructure decoupling. The aim is to make experimental substructuring more robust to noise and other measurement errors, as the amplification of these inconsistencies is the main bottleneck for these experimental techniques to become more mainstream.

 

  • Model order reduction techniques for system level model simulations

High fidelity, system-level models enable a substantial improvement in the design and monitoring of mechatronic systems. These models are typically network structured with embedded or linked discretization based models. The optimization at system-level, instead of component level, is imperative to converge to the global extrema of performance indices in the case of strongly coupled subsystems. A multitude of simulations is required during the optimization. Therefore, the research focusses on the development of a model reduction technique to lower the computational burden to simulate system level models with a large set of parameters and states with strong nonlinear dependencies.

 

  • Novel nonlinear parametric model order reduction techniques for the dynamic simulation of mechatronic systems

Mechatronic models are developed in the design stage to describe and further improve the performance of systems. These models tend to cause difficulties with respect to numerical modelling and simulation due to (1) a large number of degrees of freedom (DOFs), (2) strongly coupled non-linear equations, (3) a wide frequency difference between the dominant behaviour of the different physical domains and (4) strongly non-linear parametric dependencies. To overcome the computational burden due to the large number of DOFs, the model should be reduced. The current state of the art in model order reduction is not entirely suited to deal in an efficient manner with the described characteristics. Therefore, the objective is to develop a non-linear, parametric model order reduction technique such that the reduced mechatronic model poses less problems in simulation. The research output should lead to parametric reduced models which describe complex systems but are computationally efficient and yet accurate. Models with these properties are particularly well suited for inverse analysis and design space explorations.

 

  • Efficient system level design optimization for the design of mechatronic systems

Modern mechatronic systems exhibit strong (dynamic) interactions between the various (multiphysical) components. This makes is t particularly difficult to effectively design a system in a classical fashion where component level optimization is often performed under the assumption of fixed interface conditions. However, in practice the fully coupled system level design poses challenges due to the large number of design parameters and expensive fully system simulations. This research track focusses on optimally leveraging automatic optimization schemes and reduced order models in order to limit the computational load of these design exercises.

 

  • Application of nonlinear Model Order Reduction and Virtual Sensing to passenger car tire design

In order to meet tire performance requirements, highly nonlinear FE models are used to simulate the effects of design changes and the simulation results are used to update the tire design in an iterative manner. Tire prototypes are built following the new design and the FE simulation results are verified by means of measurements taken from the prototypes. This process is cumbersome and is constrained by the complexity of the nonlinear FE models and the measurement setup. By applying nonlinear Model Order Reduction (nlMOR) techniques to the FE models and Virtual Sensing (VS) techniques to the prototypes, this design process can be made more efficient.

Contacts

Head of the group
Industrial research contact