Industry4: Energy Optimal Train Control with Artificial Intelligence
This topic can be conducted both in English and Hungarian ! / A téma magyarul és angolul is felvehető !
A driver advisory system (DAS) is developed, which focuses on energy optimal control of the trains. The system is based on an optimization algorithm which considers the constraints coming from the timetable and constraints given by the track. The system also considers information coming from the central traffic control (TMS). But the system mainly focuses on a single train, network level optimization is currently not in the scope of our system. However, we see a high energy saving potential in introducing network level optimization or by defining network level constraints.
The optimization algorithm must consider more and more parameters which in combination with the run duration requirements makes the development very complex and effort intensive. Efforts for adding new features are increasing exponentially, which could be kept under control if at least the run duration requirement could be ignored.
In order to expand the capabilities of our system and to find solutions for the above-mentioned problems, we can outline three individual, yet related stream of activities:
Research the possibilities to mimic our current optimizer using AI-powered technologies and develop a solution which uses the current algorithm to generate learning/input data for the AI-powered solution. Compare the achievable accuracy and run duration with the existing algorithm.
Research and presumably develop a novel AI-based optimization engine that is self-learning and uses as input energy measurement data and telemetric data available from the trains.
Research AI-based fleet-wide energy optimization methods.