Study On The Fault Simulation And Embedded Diagnosis System Of Turbocharger Based On Pattern Recognition Method

Condition monitoring and fault diagnosis of turbocharger is important to improve the reliability of turbocharger, and to meet the requirements of modem predictive maintenance and automatization. However, turbocharger is a typical circumvolving machine working at heigh speed , its monitoring and diagnosis are very difficult because of its complex structure.Under this background , the dissertation has researched the simulation of turbocharger fault and the method of extracting the feature parameters of surface vibration signals generated by turbocharger and the application of pattern recognition theory to fault diagnosis of turbocharger. And a real-time design for inspecting the state of turbocharger has been developped in this dissertation with the technology of embedded-cumputer system and field bus which have been playing an increasingly important role in modern industry.In the dissertation surface vibration signals of turbocharger under 4 different conditions have been simulated and analyzed in time and frequency domain .And non-dimensional feature parameters have been extracted. The characteristics of the tubocharger under the four states have been disclosed and the feature parameters space has been found which is useful to deeper investigation in tubocharger fault diagnosis.And the simulation of the surface vibration signals generated by turbocharger has been discussed on the base of the analysis of the characteristics of these kinds of signals.