Flexible Neural Tree Based Identification of the Top Dead Center from Large Gas Engine Vibration Signals

The handling of structure-borne sound (vibration) sensors is simple compared to combustion chamber pressure sensors, and the transducers can be manufactured in a very robust form. For this reason, the Department of Mechatronics at Óbuda University, Hungary has developed a novel method (algorithm) that leads to an identification of the Top dead center (TDC) from large gas engine structure borne sound vibration signals and allows a very trivial adaptation to different engine types. In this paper, we investigate how the seemingly chaotic behavior of vibration signals could be well represented using Flexible Neural Tree (FNT) technique. Since the TDC is determined by FNT in the working cycle, it could be described how to calculate the engine rotational speed directly from structure-borne sound signals, thus also without pickup. (Clock recovery).