Rotary machinery fault feature extracting method based on local mean decomposition (LMD) and local time-frequency entropy

Abstract

The invention discloses a rotary machinery fault feature extracting method based on local mean decomposition (LMD) and local time-frequency entropy. According to the technical scheme, the rotary machinery fault feature extracting method comprises the following steps of: (1) measuring rotary mechanical equipment by using an acceleration transducer, and acquiring vibration acceleration signals; (2) performing LMD on the vibration acceleration signals to obtain a plurality of pulse frequency (PF) components, and determining the instant amplitude and the instant frequency of each component; (3) making a time-frequency spectrum, dividing a time-frequency plane and calculating the local time-frequency entropy; and (4) extracting fault features by utilizing a local time-frequency entropy value as feature quantity and combining experiments. An analyzing process of a rotary machinery fault diagnosis system based on LMD is realized, difference of vibration signals of the equipment on time-frequency distribution and energy distribution in different states is researched, a local time-frequency entropy theory can be used for diagnosing fault of machinery, the local time-frequency entropy of the vibration signals in the different states is calculated after LMD conversion of the vibration signals, and the local time-frequency entropy value is used as the feature quantity to judge whether the equipment fails or not.

Description

technical field [0001] The invention relates to a method for diagnosing mechanical faults in the field of mechanical engineering. Specifically, the invention is a method for extracting fault features of rotating machinery based on LMD and local time-frequency entropy. Background technique [0002] Nowadays, industrial production is gradually moving towards large-scale, high-speed, automation and intelligentization. Among the main equipment used by production enterprises, rotating equipment accounts for about 80%. Whether these equipment can operate normally is related to the huge economic interests of the enterprise. If a piece of equipment breaks down and fails to detect and eliminate it in time, it may bring huge security risks and even catastrophic consequences. Therefore, the research and application of rotating machinery condition monitoring and fault diagnosis technology is of great significance to ensure production safety, avoid accidents and huge economic losses, and...

AI Technical Summary

Problems solved by technology

At present, there are many methods for analyzing and processing such non-stationary signals. Common time-frequency analysis methods include short-time Fourier transform, Wigner distribution, wavelet transform, Empirical Mode Decomposition (EMD for short), etc., but they Each has its own limitations, such as the size of the time-frequency window of the short-time Fourier transform is fixed; the Wigner distribution will produce cross terms when analyzing multi-component signals; although the wavelet transform has a variable time-frequency window [9] , but like the short-time Fourier transform, it is a mechanical lattice division of the time-frequency plane. In essence, it is not an adaptive signal processing method; EMD is an adaptive signal processing method. It has been applied in many fields such as fault diagnosis, but there are still some problems in theory, such as over-envelope, under-envelope, modal aliasing, end-point effect in the EMD method, and the use of Hilbert transform to calculate the instantaneous frequency cannot be explained Issues such as the negative frequency of the

Images