Improving power supply design for high speed lines and 2 x 25 systems using a genetic algorithm

The railway's electrical system is a complex system the design of which requires an in-depth detailed survey of the main modules of which it consists. Not only is it highly complex, but it is also strongly non-linear. A number of research methods have arisen to secure the best possible design to facilitate the task. CITEF (Railway Technology Research Centre, Technical University of Madrid) has developed a methodology using an optimisation method with a railway simulator (HAMLET), which meets this objective. The methodology takes account of the fact that a good design must meet a number of major requirements. It must meet current electricity regulations, it must meet the budget imposed, it must observe environmental zones as far as possible and contemplate maintenance-sensitive zones, and finally it must identify areas posing particular difficulties in terms of connecting electrical elements to the main grid. Coordination between a genetic optimisation algorithm and a high-performance railway simulator gives the methodology a considerable analysis capacity. The 2 x 25 system requires a high computation capacity, and it was for this reason that the genetic algorithm AMGA-II was selected. Extension of the methodology to 2 x 25 systems called for adaptation of the railway simulator, a new approach to the characteristics of the genetic system, and adaptation of the mathematical model calculating costs by means of the objective functions and restrictions. The methodology was able to adapt to these complex and strongly non-linear systems, and provide a range of optimum solutions with a genuine compromise between the costs of installation and operation (Caper and Opex), a minor impact on critical zones and electrical validation for railways operating AC 2 x 25 systems.