Study Of Compressed Sensing For Mechanical Vibration Signal

As we know, because of the environments which they operating in and the equipment's particular characteristics, generally there is no redundant backup for key components of complex mechanical equipment. For example, the helicopter power transmission chain is the only load channel from the engine to the propeller, so any component fault in the channel would pose a serious threat to the aviation security. Therefore, sampling and analyzing the mechanical vibration signal and then setting up the Health Management System of mechanical equipment, including Condition Monitoring, Fault Diagnoses and Remaining Useful Life Prediction,are very important and necessary.Usually the structures of complex mechanical's key components are complicated and dynamic of operation and damage is every complex. As a result, the health management of these components needs high quality sensors and useful method of signal acquisition and processing. Fortunately, the theory of Compressed Sensing provide a new way to solve the problem, as it can sample signal at a very low sampling rate.This paper supported by the National Natural Science Foundation of China "Research on compressive sensing method of condition signals for remaining useful life prediction of helicopter main gearbox", mainly aim at acquiring signal of mechanical vibration fault to diagnose the fault, researching the method of Compressed Sensing in mechanical vibration signal sampling, by sampling at low sampling rate and accurately reconstructing the signal, while keeping the fault vibration signal features and other important information and data. The main contents include:(1) Studied the technology of Compressed Sensing and the method of Analog to Conversion like the Modulated Wideband Converter. And research the theoretical foundation of how to apply Compressed Sensing to mechanical vibration signal. As its powerful ability of reconstructing spectrum, apply the Modulated Wideband Converter to sample the mechanical vibration signal;(2) Set up the Compressed Sensing simulation model of Modulated Wideband Converter in Simulink. The simulated experiment of Compressed Sensing, accurately reconstructed the spectrum of original signal, obviously reduced the sampling rate and validated the validity of the Modulated Wideband Converter model. Besides, after studied the effect of signal noise and parameters of model, ameliorating the Modulated Wideband Converter model;(3) Established the model of the outer ring of bearing's fault signal, and sampled the bearing's signal at 50 Hz by the Compressed Sensing simulation model of Modulated Wideband Converter. Based on the sparsity of the frequency spectrum of the fault signal, the model identified the 100 Hz and its frequency multiplication, which reducing the sampling rate effectively and identifying the fault feature successfully.(4)Using the model of Modulated Wideband Converter sensed the fault signal of outer ring of bearing, which generated from the teststand of Case Western Reserve University. The model reconstructed the spectrum of original signal at a low sampling rate, which proved that Modulated Wideband Converter model is effective in sampling the practical signal. The experiment explored how to use Compressed Sampling to sample mechanical signal and built a foundation for further research.