Dies sind die archivierten Webseiten des Lehrstuhls für Programmierung und Softwaretechnik (PST).
Die Seiten des Software and Computational Systems Lab (SoSy) finden Sie auf https://www.sosy-lab.org/.

Oberseminar 12.4.2016

Studentenvorträge zu Projekt- und Abschlussarbeiten

14:15h - 15:45h

Raum U127, Oet. 67


Christian Lacek - Demand Response Management with a Fleet of Electric Vehicles

Masterarbeit, betreut von Marianne Busch und BMW

On late summer afternoons, when air conditioning systems are running both in homes and companies, California is experiencing power peaks up to 45 Gigawatts. To be able to cope with loads like that, peaking power plants are built, which have the drawback that they are usually cost-intensive and environmentally unfriendly. To keep the number of these facilities low, one approach to balance the load peaks are demand response programs, where participants receive incentives to reduce their power draw at peak hours and shift it to off peak hours instead. Considering electric vehicles as the transportation mean of the future, it appears reasonable to include them in demand response processes.

In this thesis a fleet of approximately 100 electric vehicles (BMW i3) is used to be able to release 100 kW over one hour one time per day when high peaks occur. To achieve that, all cars that are drawing power interrupt their charging and delay it to a later point of time. As the load reduction achieved by the vehicles might be too low, a battery system serves as a backup to deliver the remaining delta if needed. In this thesis a smart charging algorithm for the battery is developed to both ensure that the battery's capacity is always high enough to serve a load reduction request as mentioned above and also to use the excessive capacity of the battery to balance the load of a facility including several car charging stations. The algorithm is improved by using a weather forecast and real time data of the vehicle pool collected by the BMW i Charge Forward project in the San Francisco bay area. As the capacity needed strongly depends on the amount of charging cars, the same should be predicted one hour in advance. Therefore, seven different machine learning approaches are introduced, evaluated and compared using real data of the vehicle pool. The most accurate results are achieved by a combination of classification and linear regression. The accuracy of the models is likely to improve even more in the future with growing amount of usable historical data. 


Mathias Wendrich - Lean Product Development for Small Businesses

Bachelorarbeit, betreut von Prof. Rolf Hennicker

Lean Product Development ist die Umsetzung der Lean Prinzipien in der Produktentwicklung. Diese dienen der Optimierung von Gesamtsystemen und der Eliminierung allen Überflüssigen. Eric Ries hat in seinem Buch The Lean Startup die Verwirklichung dieser Prinzipien für Startups beschrieben.
Nachdem ein theoretischer Ansatz für Small Businesses beschrieben wurde, wird dessen Durchführbarkeit und Nutzen an einem Beispiel gezeigt. Dabei wird insbesondere der Umgang mit begrenzter Arbeitskraft berücksichtigt.