Extraction of fan vibration characteristics by cyclostationary method based on complex amplitude modulation model

Abstract

A method for extracting the vibration characteristics of a fan based on a cyclostationary method of a complex amplitude modulation model, comprising the following steps: using an acceleration sensor to collect the vibration acceleration signal of the fan; importing the collected acceleration signal into a program to be run, and using a cycle-based Correlation feature function detection of stationary features to obtain the cyclic density spectrum; using the prior knowledge of the complex amplitude modulation model of the rotating machinery, construct the 3D cyclic density spectrum and the 2D top view of the 3D cyclic density spectrum with reduced amplitude difference; according to the 2D Judge the modulation frequency from the top view, and judge the carrier frequency according to the three-dimensional cyclic density spectrum; use the obtained modulation frequency and carrier frequency to establish a simulation signal according to the complex amplitude modulation model, process the corresponding detection results, and compare them with the actual detection results to verify the correctness of feature extraction ; Utilize the present invention to detect and judge the fault type of the fan in real time, the detection is more accurate, and has strong practicability.

Description

technical field [0001] The invention belongs to the field of signal processing and feature extraction, and in particular relates to a method for extracting fan vibration features by a cyclostationary method based on a complex amplitude modulation model. Background technique [0002] Cyclostationary signal processing is a newly emerging technology of signal processing. A cyclostationary signal is a signal that contains hidden periodic information. The cyclostationary signal is a kind of non-stationary signal, which is closer to the actual signal than the traditional detection method, especially the signal generated by the rotating machinery. [0003] At present, the commonly used rotating machinery fault detection methods in the field of signal processing mainly include Fourier transform, short-time Fourier transform, wavelet transform, second-generation wavelet transform and multi-wavelet transform, etc., which can be said to be based on the principle of inner product. The ...

AI Technical Summary

Problems solved by technology

[0004] However, there are following shortcomings and deficiencies in the prior art: Fault detection methods such as Fourier transform, short-time Fourier transform, wavelet transform, second-generation wavelet transform and multi-wavelet transform are all based on the assumption that the signal is a stationary signal, and In reality, it is often a non-stationary signal, so these detection methods are unreasonable and unrealistic

At the same time, due to theoretical limitations, these traditional detection methods are difficult to detect some important features of rotating machinery, such as blade passing frequency BPF, blade ratio frequency BRF, etc., which have great limitations

Images