Drive Systems
The Drive Systems group is part of the Energy Systems Research Domain and conducts research on novel control methods for the broad application area of drive systems. Research focuses on the control and testing of wind turbines, as well as the control of novel combustion processes.
Contact
Wind Energy
The Drive Systems research group develops solutions for control-related issues within the field of wind energy systems. Automatic control of single wind turbines and wind parks is addressed as well as the development and implementation of multi-physical Hardware-in-the-Loop systems that allow full-scale wind turbine testing on ground level test rigs.
Wind Turbine Control
Wind Energy is an interdisciplinary topic that can only be fully addressed in close cooperation with partners of various domains. Therefore, the Institute of Automatic Control is one of the seven founding institutes of the Center for Wind Power Drives, CWD for short. In this center, we collaboratively work on research topics in the field of Wind Energy.
Generally, wind turbine control has two main contradictory goals:
- Maximization of produced power
- Minimization of mechanical and electrical stress.
From the control theory point of view, the wind turbine system can be characterized by the following properties:
- A wind turbine is a nonlinear system.
- The system has multiple inputs and outputs. The inputs are generator torque and pitch angle of the rotor blades, while outputs depend on the control problem at hand. From an energetic perspective, the most important output is generator current and power. On a more detailed level, mechanical loads and stresses are outputs that gain increasing attention due to high demands on reliability.
- The system is mainly influenced by disturbance that is not well-measureable, for example altering wind speeds. Therefore, we can interpret the control task as disturbance rejection control. Hereby, the impact that the low-power, high-frequency portion of the incoming wind has on power output and mechanical stress is to be rejected. At the same time, the low-frequency amount provides the energy that is to be harvested.
New control concepts for wind turbines are developed and prepared for testing in real-world applications in close cooperation with industrial and academic partners. To accomplish this aim, the following methods are of main interest:
- Development of adaptive Model Predictive Controllers in order to reduce loads at rotor blades, tower and drive train.
- Utilization of wind predictions in both feedback and in feedforward control
- Observer design and validation for estimation of wind characteristics and turbine states
- Development of reduced-order, real-time capable white box turbine models
- Validation of innovative control concepts using Co-Simulations of well-respected, high fidelity simulation tools such as FAST, Bladed, Simpack or alaskaWind and implementation on industrial controller hardware
Hardware-in-the-Loop Systems for Wind Turbine Test Benches
Another topic within our research focus of wind energy are multi-physical Hardware-in-the-Loop (HiL) systems and their application to system test benches for operating full-scale wind turbines with rated power of multiple megawatt. In order to reproduce the wind turbine dynamics on ground-based test benches, an innovative HiL-system emulates missing components and allows realistic load application despite the absence of mechanical components such as the turbine’s rotor and tower. To do so, an electromechanical drive machine and a hydraulically powered wind load unit, which apply calculated loads in all degrees of freedom, actuate the system. A realistically reproduction of the electrical grid behavior in fault conditions is guaranteed by using a grid emulator. We actively transfer our achieved research results into the standardization process for the electrical certification of wind turbines.
Control-oriented challenges arising when transferring a wind turbine from field- to ground-testing and applying multi-physical HiL-systems are:
- The overall system exhibits electromechanical interactions originated in the system’s dynamical coupling.
- To effectively use the developed HiL-concept for testing and certification purposes, the system must be variably adapted to different wind turbine and drivetrain types in order to optimally apply realistic loads and compensate for inertia discrepancies.
- The emulation of missing components, such as the rotor and the tower, with the aim of reproducing the real plant dynamics on ground level.
- Existing parametric and dynamic uncertainties mainly affect the control problem at hand. Time delays, induced by the communication of a multitude of subsystems, are unavoidable and contribute to the controller synthesis problem. The additionally unknown dynamics of the controller the wind turbine is delivered with and the dynamics of the electrical system must be taken into consideration as well.
So far, the developed HiL procedure has been successfully validated with different state-of-the-art wind turbines of well-known industrial partners as well as an independently automated research wind turbine of the multi-MW class. The following topics and methods are constantly being developed and form the framework for research:
- Development of Model Predictive Control algorithms and Robust Control approaches to compensate for test rig related eigendynamics in order to enable the exact application of calculated loads to the system
- Guarantee of closed loop stability, despite model plant mismatches, uncertain dynamics and communication induced time delays
- Mathematical description of uncertainties and their consideration in controller synthesis
- Construction of reduced-order, real-time capable white and grey box models for the implementation on real-time simulators
- Control-oriented description of multi-physical couplings of the electrical and mechanical subsystem interacting in the overall system and taking into account different time scales
Combusion Processes
Another research focus of the Drive Systems group is on combustion processes and combustion engines. The common features of the different types of combustion processes result in overarching challenges for control engineering:
- Combustion processes are non-linear.
- Combustion processes can be modelled physically by considering the different physical-chemical processes such as combustion chemistry, fluid mechanics and thermodynamics. However, the resulting models are too complex for an approach to application in control engineering. Therefore, these models have to be reduced for control purposes or other suitable data-driven models have to be identified.
- The dynamics of combustion processes covers several time scales. The process of chemical reactions up to the cycle of an engine covers several time scales.
Together with industry and university partners, we are conducting research on modern, model predictive control methods for combustion processes that go far beyond the methods commonly used in industrial practice. A further focus is on the real-time-capable implementation of the algorithms on rapid-control-prototyping hardware and experimental testing on test benches or in prototype vehicles.
- Application of real-time capable nonlinear model predictive controls. Here, the nonlinear behavior can be systematically taken into account and at the same time, in addition to the follow-up control, the compliance with limit values can also be taken into account.
- Combination of model predictive and iterative learning controls for cyclic processes.
- Nonlinear black box identification for process components that are difficult to model
- Creation of control-oriented white-box models.
- Nonlinear observer concepts to estimate the non-measurable quantities quickly and reliably.
Theses and HiWi activities
We are always looking for motivated students who want to work in the above mentioned areas within their project-, bachelor- or master-thesis or are looking for a student job. You are very welcome to send us your application.