Study On The Fault Diagnosis For Scroll Compressor Based On Information Entropy

In recent years, The infrastructure of platform based on the scroll compressortest system and the analysis of the vibration noise demonstrated a very broadapplication prospects, but still difficult to meet the requirements of the scrollcompressor running condition monitoring. For the application of domestic scrollcompressor is not long, the failure analysis used in the process is few. still in theprimary stage of building a test platform and using a traditional and unitary spectrumanalysis to describe the work status, And lots of vibration excitation source caused theshell surface of scroll compressor showing a non-stationary and nonlinear signal, sothe failure diagnose is complicated, based on the analysis of the conventionalspectrum, this paper provides a fault diagnosis discrimination of non-stationary signalthrough an information fusion of multiple perspectives.In view of the notion of vibration signal analysis, combined with the theory ofentropy and grey incidence degrees in information theory, the paper sets up a faultdiagnosis method of information entropy which based on singular spectrum entropy intime domain, power spectrum entropy in frequency domain, wavelet power spectrumentropy and wavelet space feature spectrum entropy in time-frequency domain, Whichis used as the comprehensive appraisal index of quantitative feature for the scrollcompressor's vibrational status. The difficulties in several key analysis,one is aboutsingular spectral entropy theory of embedded delay parameter selection problem,thebear directly on a fantastic decomposition of the effective information for signal andnoise and distinguish the effect.the other it is to the traditional probability entropyjoined the sliding window was improved,make the power spectral entropy and smallwave energy spectrum entropy reflects the local signal characteristics of thedistribution difference and change.In the process of signal sampling, After debugging of the hardware connection,combining with software operating platform and constituted a complete experimentalplatform, then obtained experimental data from this platform, achieved mathematicmodel of the four entropy: singular spectral entropy, power spectrum entropy, waveletenergy spectrum entropy and wavelet feature spectral entropy by several informationentropy algorithm of the MATLAB toolbox.Compared the test sample signal and the reference sample of the entropy value with different speed of scroll compressor, canachieve good recognition of the scroll compressor several failures and prove theeffectiveness of the fault diagnosis method.