Machine Learning-Based Buck-Boost Converter Design for Traction Power Centers in Rail Systems

Abstract

Abstract Since the traction power energy consumption is at high levels in Rail Systems, the design of the traction power center and minimizing the losses and increasing the system efficiency are important issues. While different designs are used while creating the traction power electrification system, the basic principle is to effectively transfer energy from one point to another. The power transmission system from vehicles to vehicles, from vehicle to traction center or from vehicle to another consumption center develops depending on the developments in power electronics technology, and the efficiency of the system changes according to different topological structures. While the developments in converter, inverter, semiconductor element and control technology constitute important inputs for system performance, it is necessary to create design conditions in harmony with the rail system operation for efficiency. In this study, machine learning-based Buck-Boost Converter circuit design was carried out with real data measurements performed in the field for the power supply system located in the traction centers. With the proposed topology, the efficiency of the power supply system has been increased, gains of the system are compared with the situation where different control methods are used, and the results are presented comparatively.