In this study a closed-loop cruise controller to minimise the running costs of the heavy-haul train is proposed. The running costs of a heavy-haul train are dependent on its travelling time, maintenance costs and energy consumption during the trip. The Coallink train with the new train technologies, Distributed Power (DP) traction and Electronically Controlled Pneumatic (ECP) brake system, is the centre of the study. A literature study on existing train control, both passenger and heavy-haul trains, is carried out to build up a knowledge base. Many different techniques for train handling were observed, their features in relation to heavy-haul ECP trains are discussed. From these backgrounds, a comprehensive longitudinal train model is proposed and successfully validated with real-life data from Spoornet. In the model, both static and dynamic in-train forces are studied, as well as energy consumption. This is possible by modelling each locomotive and wagon as an individual unit. The equations of motion for the train with coupled units and additional non-linearities, such as traction power limits, are considered. An open-loop controller for maintaining equilibrium velocity is designed. During transient velocity changes, a transient controller for calculating the required additional acceleration and deceleration is designed and validated. Because locomotive traction settings are only available in discrete notches, quantisation conversion from force into notches results in input chattering. In addition, during brake to traction transitions, the locomotives receive a sudden traction demand which results in spikes in in-train forces. To avoid these problems, input filtering is performed for these inputs. Closed-loop controllers based on LQR method, optimised for in-train forces, energy consumption and velocity regulation respectively, are designed and compared. To overcome the communication constraints, a fencing concept is introduced whereby the controller is reconfigured adaptively to the current track topology. Different train configurations in terms of availability of additional control channels for both traction and braking are compared, as well as their effects on dynamic and static in-train force. These configurations are unified, distributed and individual traction and brake controls. The results from these different configurations are compared to recorded train data and given in this study. From the results, it is found that the closed-loop controller optimised for in-train force is able to provide the best overall improvement out of the three controllers.
Dissertation (MEng)--University of Pretoria, 2007.