TU München - Fakultät für
Es sprechen Studenten über ihre abgeschlossenen Diplomarbeiten und Systementwicklungsprojekte.
Am Dienstag, 16.02.16, ab 10:00 Uhr, im Raum 01.11.018 (Konrad Zuse):
A Distributed Approach to Approximate Numerical Solution of Discrete-Time Optimal Control Problems
During early phases of systems engineering it is important to obtain an estimation regarding the risks of engineering projects as early as possible. Here, specification and solution of underlying problems as optimal control problems can help to obtain an early problem understanding. However, in many cases problem solution strategies for optimal control problems are either not applicable to a large number of problems or do not scale well for highly complex problems. In this work, we present a concise formal model specification technique for optimal control problems. Furthermore, we propose an approach, which employs modular approximate solution techniques for the solution of specified problems. Moreover, we present a distributed architecture, which supports former solution techniques. Then, we conduct a demonstration of our overall approach using a set of different example scenarios. Finally, we discuss the strengths and weaknesses of our approach with regard to observed results and different criteria.
Representation of AUTOSAR Application Interfaces reference Architecture in Chromosome Modeling Tool XMT
The automotive industry uses different standards that are created by industry consortia in order to ensure a high quality level and facilitate the cooperation between automotive manufacturers and suppliers. The AUTOSAR Consortium is the most important development partnership of leading companies in the Field of automotive Electrics and Electronics architectures. The AUTOSAR open industry standard affects and defines in particular software and topology of automotive control units (ECUs) and buses. The AUTOSAR AI describes a layered application architecture that can be used in the passenger car. It describes logical components, data dependencies between components and defines data types. This architectural model includes five application domains: Body, Powertrain, Chassis, Safety and Telematics. Below this there is a hierarchical architecture of interconnected components which in turn can be partitioned into further subcomponents. Currently, various projects adopt different centralized approaches to vehicle architectures. One of them is the RACE project, which is based on the Chromosome Middleware and Modeling Tool. The aim of this thesis is to analyze this architecture, which is defined by the AUTOSAR Application Interfaces (AI), and make it accessible in the Chromosome Modeling Tool (XMT). During this transformation the main focus is put on different modeling approaches, which represent all the relevant properties of the structure with the minimum losses of information. Finally, the structure of AUTOSAR is visualized in a dynamic environment.