Hybrid Planning for Self-Optimization in Railbound Mechatronic Systems

Abstract

Self-optimizing mechatronic systems with inherent partial intelligence are the research objective of the Collaborative Research Centre Self-optimizing concepts and structures in mechanical engineering (CRC 614, 2010). A mechatronic system is called a self-optimizing system in this context, if it is not only able to adapt the system behavior to reach a set of given objectives or goals but also can adapt the objectives themselfes (or their weighting) on the basis of an analysis of the actual situation. Hence, the self-optimization approach promises to leave degrees of freedom in choosing objectives for the system open until runtime. This means that the system can decide upon internal objectives based on external user input and current environmental conditions while the system is running. This is in contrast to a system where all internal objectives are set before the system is started. Having such fixed parameters leads to complicated and overly pessimistic approximations of the parameters that are needed to be set over the course of action that the system will take. Using self-optimization, leaving that decision open is the key idea of our approach. The external objectives like quality of control, comfort or total energy consumption along with constraints (e.g. maximal peak powers or average power consumption) are still embedded or entered into the system. But settings like distribution of energy usage among subsystems can be determined during runtime based on the actual system conditions. One application of this approach to mechatronic system design is a novel transportation system that is developed in close collaboration with CRC 614. The core of the system consists of railbound vehicles called RailCabs that enable groups of up to 12 passengers to travel directly without intermediate stops. A test track in a scale of 1:2.5 and two RailCabs in the same scale have been built at the University of Paderborn under the name Neue Bahntechnik Paderborn (NBP; see Figure 1). The RailCab is equipped with a doubly-fed linear motor as well as several innovative subsystems that feature inherent intelligence. Among these subsystems are an Air Gap Adjustment System (AGAS) and an active suspension system described more closely below (see Section 3). Operating parameters of these (sub)systems need to be adapted depending on environmental conditions and properties of the track sections that are being travelled. Hybrid Planning for Self-Optimization in Railbound Mechatronic Systems 10