MSc Thesis Projects

Are you looking for an MSc project in Control Systems? We have compiled the following updated (May 2017) list of topics for you, please contact the intended supervisors if you are interested:



Parameterization of tube-based LPV MPC policies

Tube MPC (TMPC) is a paradigm proposed for the computation of model predictive controllers when the system to be controlled is subject to disturbances or uncertainty.

Contact: Dr. M. Lazar, R. Toth, ir. J. Hanema

On stability and stabilization of impulsive linear systems

Topic description: Impulsive linear systems are a class of hybrid (“jump and flow”) systems which are very relevant to applications. Indeed, interconnecting a digital controller with a physical plant leads to such a hybrid system. Analyzing stability is currently carried out using a combination of Lyapunov methods for discrete-time systems and continuous-time systems. This projects aims to come up with new necessary and sufficient stability and stabilization criteria for impulsive linear hybrid systems based on non-monotone Lyapunov functions and linear matrix inequalities. Applications involving digital control of linear systems will be explored.

Contact: Dr. M. Lazar

Self learning control

Machine learning based control has recently received significant attention in solving control problems where model based control is not applicable due to the uncertainty or the high complexity of the involved phenomena. The project aims at comparing learning control based methods like Gaussian Control, Apprenticeship Control, etc. on laboratory test setups.
Contact: R. Toth

Symbolic model learning for control

The aim of this project is to identify simple models of complex systems that achieve comparable control performance using a recently introduced new paradigm in system identification: symbolic regression. Symbolic regression aims at building models from basic building blocks (time operators, signals, addition etc.) to completely avoid the need of model structure selection or expert decisions, achieving an autonomous logic engine that can accomplish the whole data-driven modeling by itself. The project aims at further improving the current state-of-the-art results, by investigating the possibility of using symbolic regression based methods in order to identify input-output models of complex non-linear systems for the purpose of control. Evidently, a part of the challenge is the formulation of the symbolic regression via a suitable parametrization of the control objectives and their inclusion in the learning process. Additionally, if possible, compare the control performance achieved when using the approximate identified model with that achieved by a more complex/accurate model of the underlying system.
Contact: Dr. ir. R. Toth, ir. D. Khandelwal,

Closed-loop / network identification under non-ideal sampling schemes

Identification methods are typically developed for fixed (equidistant-in-time) sampling schemes. The question that we would like to investigate is: how could identification be done on the basis of nonlinear elements occuring in the loop (e.g. switching elements), or when control actions take place event-triggered, rather than time-triggered, or with non-predictiable delay (e.g. through wireless communication channels). 

Contact: Van den Hof

Identifiability of a dynamic network on the basis of a reduced number of node signals

Given a dynamic network with partly known interconnections and dynamics, and a limited number of node signals that has been measured. We would like to develop a methodology for determining under which conditions the full network can be identified (or which part of the network can be identified) on the basis of the available node signals and (partial) prior information. 

Contact: Van den Hof

LPV identification study

Linear Parameter-Varying (LPV) models allow to represent nonlinear / time-varying systems in terms of a linear structure, allowing the extension of LTI identification and control synthesis methods to handle such dynamical phenomena efficiently. Currently a toolbox is developed to implement the state-of-the-art data-driven LPV modeling methods. This project aims at assisting the Matlab implementation of these methodologies and providing a thorough comparative study  of their performance on laboratory test setups.

supervisor : R. Toth

LPV control study

Linear Parameter-Varying (LPV) models allow to represent nonlinear / time-varying systems in terms of a linear structure, allowing the extension of LTI identification and control synthesis methods to handle such dynamical phenomena efficiently. This project aims at comparing state-of-the-art approaches of control design from the local (gain-scheduling) and global (LFR based control,  polytopic controllers) on laboratory test setups.

supervisor: R. Toth

Frequency domain subspace identification for LPV systems

Linear Parameter-Varying (LPV) models allow to represent nonlinear / time-varying systems in terms of a linear structure, allowing the extension of LTI identification and control synthesis methods to handle such dynamical phenomena efficiently. The concept of handling LPV systems in the frequency domain is underdeveloped, although in an industrial context frequency domain information of the system dynamics have paramount importance in validating the behavior of the model and this setting also allow for better understanding of controller tuning and control oriented identification. This project aims at establishing a frequency domain subspace method based on recent results.

supervisor: R. Toth

Design of stable dynamic networks for simulation purposes

For our research on dynamic networks it is important to being able to create seriously sized dynamic networks, on the basis of which simulations can be performed, and therefore require a stable network. Additionally, for purpose of identification, it is important to run simulation for class of networks (e.g. by varying hte particular parameter values in a model class). In these situations stability of the network is important and needs to be guaranteed. The objective of this project is to develop a methodology for constructing a dynamic network with a growing number of elements, where in each step an additional network element is added, while guaranteeing (robust) stability of the network.

Contact: P.M.J. Van den Hof

Can poor models give good controllers?

An important  application of model approximation amounts to deriving simplified models for control system design. Indeed, if simulation or first-principle models become too complex to allow model-based controller synthesis, then a simplified substitute model is often helpful to enable controller synthesis. This project aims to derive novel model approximation techniques that take the performance objectives and specifications of the controlled system into account in an explicit way. This means that we aim to derive model reduction techniques that provide quantified performance and robustness guarantees on the controlled system that is synthesized on the basis of the reduced model.

Information: Prof. dr. S. Weiland

Computation of finite-step stabilizing ingredients for nonlinear MPC

Topic description: Standard terminal cost and constraint set ingredients for designing stabilizing MPC algorithms are not practical. Recently, a new type of ingredients has been proposed in the reference below. These ingredients are called finite-step terminal cost and finite-step periodically invariant set. This project will aim to develop tractable and systematic methods for computing such ingredients for nonlinear discrete-time systems using optimization and approximation of reachable sets via ellipsoidal sets or interval analysis methods. These methods have been applied successfully to compute invariant sets or to approximate reachable sets for nonlinear systems.

Contact: Dr. M. Lazar

Sampling-driven feedback min-max nonlinear MPC

Feedback min-max model predictive control is one of the most advanced types of robust model predictive control. However, due to the min-max nature of the problem and due to optimizing over feedback policies feedback min-max MPC algorithms are very difficult to implement, even for linear models. This project adopts a combined approach that uses dynamic programming and sampling of the input space to achieve practical feedback min-max MPC algorithms for nonlinear systems.

Contact: Dr. M. Lazar, MSc R.V. Bobiti

Model reduction for parameter varying systems

For many engineering problems the management of complexity of dynamical systems becomes increasingly important. Model reduction amounts to constructing  simplified models that substitute a complex one. This project aims to develop novel model reduction techniques for the special class of linear parameter varying models. Results will prove useful for situations in which fast and reliable prediction of system variables are required.

supervisor: prof.dr. S. Weiland

Controller Synthesis for Switched Systems

Switched and polytopic systems have received an considerable amount of attention in the literature in the last decade and also the controller synthesis problem has been addressed. In particular, controller synthesis procedures have been developed in the form of linear matrix inequalities (LMIs). The main drawback of all these methods is that they introduce conservatism in the synthesis conditions. Moreover, the codesign of a controller and a switching sequence can only be done under certain restrictive assumptions. In this project, we would like to reduce the conservatism and allow for simultaneous synthesis of the controller and the switching sequence by finding a local solution to an optimisation problem with bilinear matrix inequality constraints.

supervisor: M.C.F. Donkers

Game theoretic control for complex networks

This project aims at designing and simulating novel control laws for competitive agents playing network games. Read more.

supervisor: dr. S. Grammatico

The scenario approach optimization for stochastic aggregative games

This project aims at analysing and simulating the scenario approach optimization for agents playing stochastic games. Read more.

supervisor: dr. S. Grammatico

Discretization of PDE’s through Port-Hamiltonian Systems

Systems that evolve over space and time are typically described by partial differential equations (PDE’s) and numerically implemented by finite-element or finite volume techniques. The area of scientific computing has developed many techniques for this purpose. An important disadvantage of numerical meshing techniques is that the underlying physical meaning of the PDE’s (energy conservation, mass conservation, etc) is discarded in the discretization process. It is the (ambitious) purpose of this project to remedy this and develop discretization techniques that provide firm guarantees on energy conservation.
Contact: Prof.dr. S. Weiland

Measuring and analysing impedance spectra for photo-electrochemical cells

Photo-electrochemical cells (PEC) use solar energy to separate water into hydrogen and oxygen. These cells are therefore the perfect generators of energy for a wide scope of applications. This project aims to study the impedance spectrum of PEC cells by a novel measurement system that is available at the DIFFER institute and to use these measurements to model the PEC cell using identification techniques.
Contact: Prof.dr. S. Weiland ; P.M.J. Van den Hof

Frequency domain estimation of parabolic partial differential equations with spatially varying transport coefficients

This project studies transport phenomena in space and time and aims to identify spatially distributed profiles for PDE’s on the basis of frequency domain data. The project is carried out in collaborationwith DIFFER and involves applications of electron heat transport in fusion reactors. Read more...
Contact: Prof.dr. S. Weiland

Charging control for large populations of plug-in electric vehicles

This project aims at designing and simulating innovative coordination schemes the charging schedule of large fleets of plug-in electric vehicles. Read more.

supervisor: dr. S. Grammatico

Semi-decentralized frequency control in power systems

This project aims at designing and simulating novel control laws for regulating the frequency dynamics in future power systems. Read more.

supervisor: dr. S. Grammatico

Closed-loop (and distributed) identification for advanced climate control in buildings (interconnected rooms)

Computational modeling and simulation play a key role for the climate control in buildings. The aim of this project is to identify data-driven thermal and hygric models in a closed-loop settings, i.e., in a conditioned building when the controller (heating/cooling) is on. Furthermore, a building can be seen as an interconnected system of different zones (rooms), a next step is to exploit the structure and perform a distributed (network) identification considering all the interactions between different zones. For a case study, a monumental building, i.e., hermitage museum is used. 

Contact: P.M.J. Van den Hof

Multi-objective optimization for smart power grids

This project aims at analysing and simulating the benefits of multi-objective optimal energy scheduling in future power networks. Read more.

supervisor: dr. S. Grammatico

Model-predictive and distributed climate control in buildings

An accurate indoor temperature and relative humidity control is essential for the preservation of monumental buildings. Traditionally, PID controllers are employed. In this project, we explore the opportunities to use model-predictive control for individual zones of a monumental building, i.e., hermitage museum. The effect of the interaction between different zones will be studied and the question of designing a distributed climate control will be addressed.        

Contact: Van den Hof

Equivalent Consumption Minimisation using Stochastic Dynamic Programming

Improving vehicle energy efficiency is an important topic in automotive research. To optimally exploit the synergy between all the energy consumers in the vehicle, a supervisory control system is needed. Such a supervisory control system is called an energy management system and are typically based on the so-called equivalent consumption minimisation strategy (ECMS). The design of ECMS is typically done using heuristics. In this project, we want to improve and extend the recently proposed approach 1-Step Look-Ahead Stochastic Dynamic Programming to design an ECMS in a systematic way.

Supervisor: Tijs Donkers

Improved State and Parameter Estimation for Li-ion Batteries

Li-ion battery technology is, due to properties such as high energy density, a growing and very promising technology for enabling and accelerating many sustainable and renewable solutions in society, such as energy storage and sustainable transport. For safe and proper operation of Li-ion batteries, battery state and parameter estimation is needed. However, the state-of-the-art of battery state and parameter estimation describes a wide variety of algorithms which all have advantages and disadvantages and in general, are not optimal. For example, some algorithms are quite fast, but not very accurate, whereas other algorithms require meticulous tuning and do not have any guarantees on stability or convergence of the algorithm. Therefore, using an extensive set of data collected from an e-bike experiment, this project aims at, first, investigating the performance of the state-of-the-art of estimation algorithms, and second, improving these algorithms or developing a different and more sophisticated algorithm.

Supervisors: Henrik Beelen, Tijs Donkers

Joint State and Parameter Estimation for Polytopic LPV Systems

Linear parameter-varying systems are very suitable for modelling nonlinear systems, since well-established methods from the linear-systems domain can be applied. Knowledge about the scheduling parameter is an important condition in this modelling framework. In case this parameter is not known, joint state and parameter estimation methods can be employed, e.g., using interacting multiple-model (IMM) estimation methods, or using an Extended Kalman filter (EKF). However, these methods cannot be directly used in case the parameters lie in a polytopic set. Furthermore, these existing methods require tuning in order to have convergence and stability. Therefore, performing joint estimation within the polytopic LPV framework can be seen as an alternative to IMM and EKF. The first results for joint estimation in polytopic systems have been obtained in the form of Dual Estimation (DE). These results have indicated numerous challenges and possibilities for further research. Therefore, the aim of this project is to investigate these challenges and take joint estimation for polytopic systems to the next level.

Supervisors: Henrik Beelen, Tijs Donkers

Joint Voltage-Current-Temperature Modelling and Parameter Estimation in the Doyle-Fuller-Newman framework

Lithium-ion batteries are essential in various applications because of their high specific energy and long service life. Lithium-ion battery models are used for investigating the behaviour of the battery and proper power control in applications. The Doyle-Fuller-Newman (DFN) model is a popular electrochemistry-based model, which characterizes the dynamics in the battery considering mass and charge-transfer limitations in solid and electrolyte and predicts current/voltage relationship. The current model does not takes into account temperature development in the battery during operation. It is known that such temperature development can be considerable, especially in automotive applications where high currents are typical. The topic of MSc project is to extend available DFN model by thermal part, design the experiments necessary for parameters identification and, finally, identify/estimate overall set of parameters. 

Supervisors: Dmitri Danilov, Tijs Donkers

All-Solid-State thin-film Li-ion batteries: Modelling and Estimation (Jülich, Dmitri)

Modern Lithium-ion batteries are fairly considered as a most perspective energy storage device. However they suffer from various drawbacks such as poor safety and still low energy density. Both issues can be resolved when solid-state electrolytes are used instead of conventional liquid ones. That implies replacement of conventional Li-ion production technologies in favor of solid-state deposition techniques. As the result, these batteries are not described by classical Doyle-Fuller-Newman theory, therefore different modeling approach has to be employed. Recently developed model of All-Solid-State Li-ion battery. represents an example of systematic and successful approach to that problem. Developed model characterizes the dynamics in the battery considering mass and charge-transfer limitations in electrode and (solid) electrolyte layers and predicts current/voltage relationship. The topic of MSc project is to test available model against new set off measurements including commercially available and custom-made solid-state cells (experimental data will be provided by Forschungszentrum Jülich, Germany), and extend the model to accommodate mixed electronic-ionic conductivity in electrode(s) and double-layer effects. The updated model should be validated on available set of measurements. If necessary, additional experiments will be designed to identify/estimate total set of parameters.

Supervisor: Dmitri Danilov

Ensemble Kalman Filter for State of Charge Estimation of Li-ion Batteries

The battery management systems (BMS) team in Electrical Engineering Department is working on modelling and control of Li-ion batteries. These type of batteries are extensively used in various applications such as hybrid electric vehicles, laptops, cellphones, etc. It is important for the BMS to have an indication of the State-of-Charge (SoC) of the battery. Since the SoC cannot be measured directly, we need to resort to modelling the battery as a dynamic system and build an observer/estimator that uses measurable data to estimate the SoC [1].
The goal of this project is to design an ensemble Kalman filter (EnKF) [2], using a nonlinear electrochemistry-based model of the Li-ion battery, to estimate the SoC of the battery. We use our BMS experimental setup at Control Systems group to do measurement and provide the required data, such as the voltage and current profile, for designing the observer. The main steps of the projects are as follows: 1) data acquisition from the experimental setup, 2) study the existing nonlinear model of the battery, 3) design the EnKF observer to do SoC estimation, 4) validate the designed EnKF using the experimental measurements.

Supervisor: Esmaeil Najafi

[1] Campestrini, C., Heil, T., Kosch, S. and Jossen, A., 2016. A comparative study and review of different Kalman filters by applying an enhanced validation method. Journal of Energy Storage, 8, pp.142-159.
[2] Evensen, G., 2003. The ensemble Kalman filter: Theoretical formulation and practical implementation. Ocean dynamics, 53(4), pp.343-367.

Nonlinear Model Order Reduction for an Electrochemistry-Based Li-ion Battery Model

Lithium-ion batteries are commonly employed in various applications owing to high energy density and long service life. Lithium-ion battery models are used for analysing batteries and enabling power control in applications. The Doyle-Fuller-Newman (DFN) model is a popular electrochemistry-based lithium-ion battery model which represents solid-state and electrolyte diffusion dynamics and accurately predicts the current/voltage response using a set of nonlinear partial differential equations. However, implementation of the full DFN model requires significant computation time. A computationally efficient implementation of a full and reduced-order DFN model for Li-ion batteries, which is convenient for real-time applications, has been proposed in [1].
The main goal of this project is to enhance the existing DFN model by adding the thermal and aging terms, which leads to an electrochemical-thermal battery model. Then, we use nonlinear model order reduction techniques to provide a reduced order model, convenient for real-time applications.

Supervisor: Esmaeil Najafi

[1] L. Xia, E. Najafi, Z. Li, H.J. Bergveld, and M.C.F. Donkers, “Fast Simulation of a Full and Reduced-Order Electrochemistry-Based Model for Li-Ion Batteries”, submitted to Journal of Power Sources, 2017.

Haptic System Modeling and Control for an Automotive Application

Future automotive displays will have haptic feedback in display surfaces to control appliances in a car.The haptic component in a display will include force sensing and an actuator to move the display so that control operations can be  “felt”. The idea is that such applications will reduce visual distractions while driving. In this system, unwanted vibrations need to be ompensated and haptic information need to be exchanged between driver and system. This project is carried out in collaboration with “Innolux”, a company that develops LCD and OLED displays. Read more...

Contact: Prof.dr. S. Weiland

State and parameter estimation based on extents transformation for Reactive Batch Distillation Columns

Nowadays in the large-scale process industry, high purity of products is not only desired but crucial. Products must meet high-purity standards to conform to market and customers’ requirements. Read more...

Contact: Dr. L. Ozkan

Model-based optimization of chlorine production in conjunction with power grid balancing (AKZO-NOBEL)

The production of chlorine is an electrical-energy-intensive process that can be ramped up or down relatively fast. Because of its high impact on the power grid, the control of the process can be considered not only in relation to the need of chlorine customers, but also in relation to power balancing in the power grid.  When considering the process to be part of the power grid, dynamic operation is a natural and direct consequence, for which model-based control and optimization needs to be adjusted to arrive at the most economic efficient operation.
Contact: Dr. L. Ozkan ;Prof. P. Van den Hof

Robust Integration of MPC and DRTO

The traditional approach to the optimal economic operation of industrial processes has been the use of a hierarchically structured control system Figure 1 shows the hierarchical structure. Read more...

Contact: Dr. L. Ozkan

Thermal model for Generic Substrate Carrier

Sioux CCM has developed a Generic Substrate Carrier (GSC). This is a module which can be integrated in all kinds of high productive printing systems to position the substrate to be printed with high accuracy and speed. Typically sizes range from A3 paper format up to panels of more than 5 m2. The range of applications for the GSC is very wide and it is used at strongly varying environmental and process conditions. In high precision machines, thermal variations negatively affect machine accuracy. Create, implement, validate a real time thermal model for the carries and to use the model for control.
Read more.

supervisor: S. Weiland

Input/Output structure identification in oil reservoirs

This project is based on finding dominant input and output pairings in the modeling of oil reservoirs in order to synthetize efficient models for control of water-flooding based extraction.

supervisor: M.M. Siraj, prof. dr. P.M.J. Van den Hof

Improving convergence velocity of state and parameters estimators through model selection

The real time monitoring of the product quality in a process plant cannot be always achieved by means of hardware analyzers because of technical and economical limitations. Read more...

Contact: Dr. L. Ozkan

Base layer control of the Tennessee Eastman Process

A well-known industrial benchmark for plant-wide monitoring, control, optimization, maintenance and fault diagnosis is the Tennessee Eastman process.Read more...
Contact: Dr. L. Ozkan

Tuning for model based control systems

The goal of this project is to further improve self tuning schemes for model predictive control which can operate the closed-loop system closely to the economic constraints, hence maximizing profit.  

supervisor: dr. L. Ozkan, prof. S. Weiland

Ball Balancing Robot

Alten Mechatronics is de business line van Alten die is gespecialiseerd in Mechatronica en Robotica. Alten Mechatronics werkt voor toonaangevende technisch georiënteerde bedrijven en richt zich op research en development activiteiten waar disciplines als Mechanical, Electrical en Control engineering samensmelten. Lees meer...
Contact: R. Toth


Multicopters zijn de afgelopen jaren behoorlijk populair geworden in de academische wereld. Het complexe dynamische gedrag van dit type drone maakt het geschikt voor een afstudeeropdracht. In dit document zijn vier mogelijke afstudeeropdrachten omschreven die gerelateerd zijn aan een multicopter. Lees meer...
Contact: R. Toth

Mini Automated Guided Vehicles (AGV)

AGV’s worden onder andere gebruikt in warehouses en de logistieke sector om het vervoeren en bezorgen van pakketten en materialen te automatiseren. De oriëntatie en planning voor een AGV kan worden uitgevoerd vanuit een centraal punt (centralised) of vanuit een intelligente AGV als individu (decentralised). De planning voor de uitvoering van een complexe taak voor meerdere samenwerkende AGV’s is een uitdagend onderwerp. Lees meer...
Contact: R. Toth

Flexibele Robotica met ROS-Industrial

Het idee achter ROS (Robotic Operating System) is simpel: Software ontwikkelaars een framework bieden voor het ontwikkelen van robotapplicaties, zodat zij voort kunnen bouwen op het werk van anderen en ideeën en algoritmes uit kunnen wisselen. De flexibele structuur van ROS maakt het een geschikte tool voor de snelle ontwikkeling van intelligente industriële robots. In de academische wereld wordt ROS al veel toegepast. Lees meer...
Contact: R. Toth

Onboard boat classification and recognition using neural networks with sensor fused images

This project is carried out in collaboration with Damen Shipyards in Gorinchem and involves the problem to track the location of ships on the basis of multiple sensor information. Sensor fusion is an important problem in systems and control and amounts to extract the dedicated information from multiple sensors in an efficient way, taking into account noisy and erroneous  measurement information. Read more...
Contact: Prof.dr.S. Weiland

Active Control (commutation) of a Pump which includes a hydro-magnetic bearing

A Hydro-magentic journal bearing is an innovative type of bearing, where especially minimization of shear stress on the fluid is important in medical applications. Read more                     

Contact : S. Weiland

Omron Model Based Design project with Matlab/Simulink

Omron is a company that builds, among other things,  smart inverters and DC optimizers for the use ofenergy conversion in solar panels. This project involves the modeling of a DC/DC converter in Matlab and to  experiment with a model-based design loop in which an optimal control systems for the converter is synthesized. Focus of the project will be on the optimal control architecture and the design of an optimal power stage of the Omron Smart Inverter.
The project is carried out at the ESB department of Omron  in ‘s Hertogenbosch. read more...

Mentor: Hommad el Farissi (Omron)
Contact: Siep Weiland (TUE), Sascha Sanchez (Omron)

Mc projects in control and optimization in hyperthermia treatments

Hyperthermia is among the techniques to enhance the effectiveness of radiotherapy and chemotherapy in cancer treatments For deep tumors, without increasing side effects. A hyperthermia treatment consists of the local heating by EM radiation of a tumor for a pre-defined period of time by applying energy in tissue. The problem of these treatments is to carefully dose the electro-magnetic waves in so that tissue is heated at dedicated locations of the tumor in the tissue while avoiding the heating of nearby healthy tissue.  There are currently two assignments on this optimization problem: one related to control design, one related to observer design. These projects are carried out as a collaboration between Erasmus hospital in Rotterdam, the EM group and the CS group.

MSc project (Care & Cure): Time-varying optimization and control for hyperthermia treatments
MSc project (Care & Cure): Model-based temperature estimation for control during hyperthermia treatments

Contact: Prof.dr. S. Weiland

Non-linear MPC based on LPV embeddings

The aim of this project is to develop systematic and general design procedures that can be used to develop non-linear model predictive controllers based on linear-parameter varying (LPV) embeddings.
Contact: Dr. M. Lazar, R. Toth, J. Hanema

LPV flight controller design for the Parrot AR Drone 2.0

The Linear Parameter-Varying (LPV) framework provides and extension of powerful LTI control synthesis tools w.r.t. nonlinear systems. In case of agile maneuvering, addressing the inherently nonlinear coupling between the dynamical modes of quadcopters has paramount importance to push utilization boundaries of drone based applications. In this project, recently shown theoretical results on LPV flight control of drones are aimed to be implemented on a Parrot AR Drone 2.0 and compared to nonlinear control solutions (using a recently developed Simulink toolbox in co-op with Mathworks).  The aim is to achieve agile maneuvering outdoors using the onboard sensor loadout augmented with a GPS module.
Contact: R. Toth

Learning approaches for the control of a magnetically levitated positioning system

High precision positioning systems play an important role in high-end industry. In semiconductor industry, the demands for the new age lithographic machines mainly involve (sub)nanometer positioning accuracy together with high accelerations. At Eindhoven University of Technology a magnetically levitated planar actuator has been designed and constructed to investigate its performance capabilities and propose new control schemes that push its performance even further.From a control engineering perspective, the main challenges arise due to two main sources. First, the intrinsic nonlinear dynamics of the mechanical system, combined with the limited stiffness of the translator result in position dependent spatial deformations. Secondly, the uncertainty and the resulting disturbances arising from the electromagnetic part, i.e. forces generated by the interaction between stationary coils and the magnets on the translator, add to the complexity of the setup.The main goal of this MSc project is the investigation and proposal of novel control schemes which increase the performance capabilities of the setup. To this end, we aim at developing techniques within the machine learning framework towards the derivation of the dynamic description of the controller by additionally exploiting the multi-actuated nature of the setup.
Contact: R. Toth; I. Proimadis, H. Butler

Model predictive control of distributed generation inverters

Topic description: Inverters play a crucial role in the development of future smart grids and microgrids. They are a key component for connecting distributed generation units with the microgrid. Current control methods used in practice rely on classical PID control. This projects aims at developing model-based predictive control methods that can lead to a real-time feasible controller for distributed generation inverters. Challenges come from the fast sampling rate, switched nature of power electronics and hard constraints on currents and voltages. Several fast MPC methods will be explored and designed, depending on which prediction model is chosen (switched or averaged model).
Contact: Dr. M. Lazar

Control of a planar maglev precision motion system

The technology behind magnetic levitation can be employed to create next generation actuation systems that are not limited by mechanical friction  and that have extreme short response times. When equipped with appropriate control actions, these systems may achieve high levels of performance. The purpose of this project is to control a moving coil stage with log strokes in planar x and y directions and with short strokes in z and 3 rotational movements along the main rotational axes. The project aims to explore multiple control techniques to achieve high precision actuation.
Project is carried out either at TU/e or in Shanghai, China (or both).
Contact: Prof.dr. S. Weiland

Dynamic stabilization of a maglev system

Magnetic levitation systems are unstable by their very nature. This project aims to invent a universally stabilizing periodic excitation for a magnetic levitation system, with the purpose to stabilize such a system. If successful, this implies that one can stabilize an unstable maglev system without employing feedback, simple by applying (open-loop) periodic currents. Project is of fundamental nature, carried out in collaboration with Extreme Motion Technologies.
Contact: Prof.dr. S. Weiland

Multi DoF Piezoelectric Mono Layered Actuator

In a lithographic machine many elements are moved with piezoelectric actuators. Piezo actuators can extend in the length direction if an electric field is applied in the polarization direction. Read more...

Contact: H. Butler

Minimizing disturbance forces in a linear motor

In lithographic machines, linear motors move the stages at high speed and acceleration. Typically, their reaction forces are absorbed by a balance mass or a frame. In the newest generation of linear motors, Read more...

Contact: H. Butler

Model predictive control of a linear motor

Linear motors are used to move the stages in lithographic machines. The conventional control scheme of a linear motor is shown in Figure 1. Read more...

Contact: H. Butler



Kinematics and Kinetics over time/parameter-varying systems

Representations and studies of moving objects in space have been in the interest of human since ancient time. The trend of this particular branch of studies was introduced by Ampère in 1834 and further made known as ‘Kinematics’ nowadays. Read more...

Contact: H. Butler

Application of modal observer on EUV wafer stage

In various types of lithographic machines, a wafer stage holding a wafer is moved with very high precision, its position being measured by a laser interferometer system. In the case of an EUV,  Read more...

Contact: H. Butler

Auto-tuning of wafer stage to reticle stage ‘Feedthrough’ controller

In a lithographic tool, the image produced by a reticle is projected onto a wafer. Both the reticle and the wafer are moveable by a positioning system. The position of the image on the wafer is critical for the imaging process, and must be in the single-nanometer range. Read more...

Contact: H. Butler



Detection of nonlinear dynamics in networks

When modelling (electro)-mechanical systems the use of nonlinar modelling tools can be indispensible for accurate models. In particular joints between different mechanical parts can be the cause of nonlinear effects. In this project the objective is to develop network identification tools that on the basis of measurement data can  detect whether modules in a network have nonlinear behaviour. Experimental data is available of e.g. fighter aircrafts.        
Contact: P.M.J. Van den Hof

Randomized stability analysis for large-scale constrained nonlinear systems

In industrial practice, for example in engine control, it is common to rely on simple controllers, such as PID, to control complex systems. However, PID control is not designed to account for constraints, and the stability of the design is uncertain. It is therefore of interest to analyze the extent to which a control system satisfies safety and stability specifications. Existing strategies for stability analysis typically rely on computing Lyapunov functions, which ultimately resort to solving nonlinear optimization problems. However, for large-scale nonlinear constrained systems, which are often encountered in practice, these optimization problem are non-convex and of a large scale. To address these issues, this project aims at developing an automatic, randomized stability analysis framework and exploring its potential for stability analysis of large scale constrained nonlinear systems.
Contact: MSc R.V. Bobiti, Dr. M. Lazar

Additive Manufacturing (3D Printing) at Ultimaker

This is a proposal for at least 3 graduation projects to be carried out in the Control Systems Group at the Department of Electrical Engineering at Eindhoven University of Technology and in collaboration
with Ultimaker (Geldermalsen). The project involves modeling and control aspects of the 3-D printers that are developed at Ultimaker. read more...

Contact: Prof. S. Weiland



Adaptive Fault diagnosis in metaljet based additive manufacturing

Traditional fabrication methods involve a great deal of e ort, expense and time. In this process, a `3D printer' prints object layer by layer, permitting more creativity and control over the nal shape than any other construction methods.Read more...
Contact: Prof.dr. S. Weiland

Internal Force feedback for isolation systems

In lithographic scanners, the projection optics and critical measurement systems need to be completely vibration-free, even in the presence of the high forces as produced by the scanning stages. In order to achieve this, the projection optics are connected to a frame (“metrology frame”) which is isolated from the base frame (which is connected to the floor). This isolation takes place by pneumatic suspension, to which a 6-DOF control system using Lorentz actuators, accelerometers and position sensors is added. In this assignment, the use of force sensors to reduce the effect of disturbance forces is explored. A control architecture needs to be designed that uses the existing position and acceleration sensor, and the new force sensors to minimize vibration levels at the metrology frame.

supervisor: H. Butler

Compensating for deformation in Selective Laser Melting products

Selective Laser Melting is a field in Additive Manufacturing that is currently evolving fast. It is a process that selectively melts metal powder with a laser. This is realized layerwise and used to make advanced 3D shapes (see Figure 1), that cannot be realized with traditional machining technologies.

The process is however complicated and the success of a build job is dependent on a lot of parameters. The parameters can be process related as laser power, writing speed, layer thickness, melting point material etc.. The parameters can also be more design related for example internal stress, overhang angle and amount of support structures. Not optimal parameters can lead to products that do not meet the requirements or even a build failure. TNO is working on models to understand and optimize the process. Read more.

supervisor: Prof.dr. S. Weiland

Graduation project 3D Food Printing

TNO has been active in the area of 3D printing for over 20 years. Initially 3D printing was used for technical applications (general prototyping, aerospace & automotive engineering) and employed a limited set of materials, typically various types of plastics and metals.
3D printing, however, is finding its way into new areas and over the past 5-6 years TNO has become the world leader in the area of 3D food printing. Food printing not only enables the creation of novel products with interesting shapes, but it also allows the creation of fully personalized nutrition or the creation of novel (micro)structures that can lead to new food textures and food experiences. Read more.

Supervisor: Prof.dr. S. Weiland

Parameter identification and model validation for large-scale multiphysics models of a wafer table

Currently a thermomechanical wafer and wafer table model that is based on first principles modeling has been built using the finite elements method (FEM). As a result, a large-scale model with ~100,000 states has been created, which depends on ~20 material parameters. Read more...
Contact: Prof.dr. S. Weiland ; ir. R. Merks

Novel roll-to-roll web steering concept

Roll-to-roll systems are widely used for transporting web of paper, plastic or other printing media) in professional printing systems. Goal of this project is to develop a model for the transport and steering behavior of the web (medium) using newly developed actuation principles. The project involves modeling, sensitivity analysis of parameter changes due to different wem materials and dimensions and to device an optimal  control strategy for the steering behavior. More info.

supervisor: prof.dr. S. Weiland