Power management strategy study for a multiple unit train

The use of multiple power sources in railway applications is common on modern rolling stock. Prime movers are usually distributed along the length of the train, and are able to provide distributed traction. If the primary power for such vehicles comes from individual diesel engines, or if there are non-linear efficiency characteristics within the drive train, then the supervisory controllers for each motive unit may be decoupled and operated independently. This can potentially save energy. This paper investigates two global optimisation algorithms. These are the dynamic programming method, and the greedy algorithm. Both of these two graphic searching algorithms help to find the optimum power distribution between engines for a typical Diesel Multiple Unit (DMU) train. It is found that the application of Dijkstra's Algorithm, in association with a Fibonacci heap (F-heap) based advanced data structure, can significantly improve the computing efficiency while maintaining the same optimisation precision as the dynamic programming method. In an optimisation process with M power requirements and n possible engine power states for each requirement, the time complexity will be reduced from O(M*n 2 ) to O(M*(n log(n) + n)).