Routenauswahl in Transportnetzen

Automatic route choice in telecommunication networks using external management systems, also known as transport networks, is an unsatisfactorily solved problem. The task can be reduced to weighting and choosing routes in a directed graph starting with weighting edges and vertices using sets of very different parameters, parameters depending on technical and economic conditions. An essential problem of global route weighting is that the consideration of the local edges and vertices parameters cannot be described by a simple function. Therefore, the best route cannot be determined simply by means of a distance measure calculated from the parameters. Both route weighting, which takes the local parameters of edges and vertices into account, and route choice are done by experts. For this task, consideration of the network’s utilisation rate and of economical route provision is essential. Special strategic considerations require temporarily and locally bypassing the usual route choice process. Experts master this detailed fine-tuning of route choice. Existing automatic routing algorithms take into account only parameters from which a distance measure can be calculated. Therefore, the choice of routes does not take into account essential aspects of route choice. This thesis presents an automatic routing process emulating the experts’ method. Using expert knowledge, a fuzzy rule-based system for weighting the graph is constructed. By combining the fuzzy-rule based system with a shortest path algorithm, a flexible and controllable routing process is created. Based on this process, more variants are provided by additionally optimising single, additive edge parameters. The introduction of a special edge parameter and additional fuzzy rules make strategic fine-tuning of route choice possible. In order to use empirical knowledge for the construction of the fuzzy system, the fuzzy rulebased system is reversibly transformed into a neural network. A training algorithm has been especially developed for adapting the fuzzy system to empirical data. The success of all these routing methods is demonstrated by simulating route choice for high priority link connections.