IMPROVED LOCAL MEAN DECOMPOSITION FOR VIBRATION-BASED MACHINERY FAULT DIAGNOSIS

The rotor stand vibration signals carry abundant dynamic information of the machinery and are sometimes very useful in the machinery fault diagnosis. Due to the difficulty of recognition for the non-stationery signals, many traditional ways have been discussed and made the comparison to analyze the pros and cons of the methods. Then this paper proposed an improved LMD time-frequency method to solve the shortcuts, so as to obtain the better results during the machinery fault diagnosis. The LMD time-frequency method helps to reduce the appearance of the singular points and glitches and it has better precision in the PF component, so that the characteristics of the original signal can fully be reflected. And the simulated rotor signal along with actual fault signals are used to demonstrate, test, verify the accuracy and the effectiveness of the improved LMD method with the support of the established test rig and NI device. Through the Time-Frequency Analysis Time Spectrum, the newly proposed method has been proved its accuracy, efficiency in the time-frequency analysis and its stableness in the machinery fault diagnosis.