Dirk Turschner

Commissioning of a state-controlled high-powered electrical drive using evolutionary algorithms

The subject of this research is the automated startup procedure of a PI state-controlled rolling-mill motor by using evolutionary algorithms. Compared to the conventional PI speed control, applying the method of deliberate pole placement to the state controller design succeeds in improving the transient response of setpoint and disturbance changes. To put the PI state-controlled drive with observer into operation to obtain a controller with a high robustness and dynamics, the precise knowledge of this physical parameter is necessary. An evolution-based system is used to solve the estimation problem. A high degree of reliability respecting multimodal characteristics and robustness against random noise is expected from the identification method. Evulotionary algorithms fulfill this requirement. With genetic operators like mutation, crossover, and selection, evolutionary algorithms mimic the principles of organic evolution in order to solve the optimization problem.

Investigation of the Virtual Synchronous Machine in the island mode

In this paper the investigation results of the Virtual Synchronous Machine (VISMA) in the island mode are presented. The increasing integration of the distributed generation units leads to continual growing of the amount of the inverter interface in the electrical grid. Therefore, the characteristics of inverters will more and more significantly influence the grid operation. The VISMA is a novel control method for the inverter to make it possible that an inverter-interfaced decentralized generation unit can be operated as a real electromechanical synchronous machine, which up to now as the main generation unit in the electrical grid has many advantages in consideration of the grid stability. These advantages can be implemented in the VISMA and are proved experimentally in the grid-connected operation mode. The island-mode is another important operation mode that will be necessary for a stand-alone grid without connecting of the main grid or when the main grid fails due to a disturbance. Therefore this operation mode of the VISMA is focused in this paper.

A New Approach to the Analysis of Network Observability in Medium and Low Voltage Electrical Grids

Medium and low voltage electrical power grids are typically sparsely instrumented, and thus, not observable in a systems’ theory sense. However, this is a requirement to carry out state estimation methods. To this end, many approaches for optimal sensor placement are proposed in the literature. Such methods are typically motivated from a mathematical perspective, not taking the physical properties of the network into account. As a consequence, the dimensionality of the mathematical problem is typically quite large resulting in significant numerical complexity. Therefore, a new approach is proposed here which is based on analyzing the characteristic observable and unobservable nodes by using singular value decomposition (SVD) and the breadth-first search method. The aim of the method is to identify all possibilities for the placement of measuring equipment to achieve observability. The proposed method does render the network observable with a minimal number of sensors. In this way, this reduces the dimensionality for conventional optimal sensor placement algorithms substantially.

Phasen- und amplitudenfehlerfreie Filterung von störbelasteten Messsignalen mit quasistationärer Basiskomponente (Phase- and Amplitude Errorless Filtering of Measurement Signals with Quasi Stationary Fundamental Frequency Component)

Zur Vermeidung von Abtastproblemen und zur Vorbehandlung von störbelasteten Messsignalen werden bei einer vorgesehenen Verarbeitung auf Mikrorechnern oft Vorfilter eingesetzt. Die Vorfilterbehandlung führt zu Phasen- und Amplitudenfehlern am Basisfrequenzanteil des Messsignals, welche mit dem vorgestellten Verfahren kompensiert werden können. To avoid sampling problems and suppress signal disturbances, an analog prefilter is often inserted in front of the ADC input port of the process-control computer. Particularly in the case of noice-disturbed field process signals, prefiltering is obligatory. The disadvantage of phase shift and amplitude attenuation of the fundamental frequency component, e. g. the power grid frequency in such measurement signals, can be prevented with the approach described in this paper.