A speed-free extraction method for unsteady-state fault characteristics of bearings

Abstract

The present invention proposes a bearing non-steady-state fault feature non-speed extraction method, comprising the following steps: collecting the original signal of the bearing, and extracting the time-domain impact information of the original signal to obtain an effective impact signal; performing an iterative algorithm on the effective impact signal Optimization, to obtain the target shock signal that removes mis-extraction and complements missing extraction; according to the target shock signal, constructs an equiangular resampling time series to obtain a new resampling signal s(i), and performs corresponding spectrum or envelope analysis on it , get the fault characteristic frequency and fault information of the bearing. A non-rotational speed extraction method for unsteady-state fault characteristics of bearings proposed by the present invention optimizes the proposed shock information through the shock optimization strategy, removes the shocks that are mis-extracted and missed, and then calculates the phase information of the waveform according to the shock position and re-calculates Sampling, analyzing its envelope spectrum to get the fault characteristic frequency of the bearing, that is, the corresponding fault information.

Description

technical field [0001] The invention relates to the technical field of bearings, in particular to a non-rotational speed extraction method for unsteady-state fault characteristics of bearings. Background technique [0002] Rolling bearings are currently widely used in various rotating machinery to reduce friction, such as wind turbines, water pumps, motors, etc. However, because the bearings are generally subject to cyclic loads for a long time, they are in a continuous running state, and sometimes there are poor lubrication. These make the bearings the most easily damaged parts in rotating equipment, and once problems occur, the light ones will lead to equipment failure. The production line is shut down, and even a safety accident occurs in severe cases. [0003] Therefore, it is of great significance to monitor and analyze the early failure of bearings. At present, the commonly used method is to install acceleration sensors on the surface of the bearing seat to collect vi...

AI Technical Summary

Problems solved by technology

[0006] 1. Using the speed information of the equipment for order analysis requires installing a speed sensor on the equipment, and the speed sensor needs to have an absolute fixed point and be able to face the rotating parts directly, which brings difficulties to on-site construction and installation; then rotate the equipment Adding key phases on the surface will have an impact on the site, and it is impossible to find a suitable key phase installation position inside many devices

[0007] 2. At present, in the equipment monitoring system, wireless monitoring is more and more widely used due to its own advantages, but in the wireless monitoring system, the sensor uses battery power and local wireless transmission, which makes it impossible to achieve different Accurate synchronization between sensors, so that it is difficult to ensure the complete synchronization of the speed signal and the vibration signal if the wireless speed sensor is installed in time. Once there is a phase difference, the speed or key phase signal will not be used for order analysis

[0008] 3. Data-driven speed or phase extraction methods are basically based on time-frequency analysis, such as sampling short-time Fourier transform, wavelet transform, etc. This type of time-frequency analysis methods are subject to the Heisenberg uncertainty principle The limitation of time resolution and frequency resolution, that is, the time resolution and frequency resolution are mutually restricted, resulting in the inability to accurately extract the time-frequency distribution ridge, which makes the frequency estimation inaccurate and brings errors to subsequent resampling

[0009] 4. Before time-frequency analysis, it is generally necessary to filter the original signal. Whether it is to select low-frequency band or resonance band envelope signal for time-frequency analysis, it needs to involve the design of filter parameters. Most of the filter The parameters are determined based on experience, which will bring certain uncertainties to the results of frequency extraction

Images