Condition monitoring of a diesel engine by analysing its torsional vibration, using modern information processing tehcnology

摘要

TNO-FEL implemented an experimental prototype of a diesel engine condition monitoring system. The system was developed as a demonstrator within the scope of a project to investigate the use of Modern Information-processing Technology (MIT) for the Royal Netherlands Navy. The purpose of the system is to provide the operator with high-level information on the condition of the engine. The system focuses on the analysis of the torsional vibration of the axis of a diesel engine. This implies that the system only needs a single torsional vibration sensor, which is a very robust sensor. A large data set of torsional vibration data was set up. The system can detect different vibration patterns. This allows the system to detect various faults, e.g. no combustion ofr partial combustion of one or more of the cylinders of the diesel engine. The detection and classification of the torsional vibration patterns is done using advanced signal processing and feature extraction techniques, combined with fuzzy logic and neural network technology. The prototype provides a graphical display of the different stages of the signal processing steps. Partial results as well as the final result are displayed in such a way that the operator can understand how the system reaches a conclusion.