A Line Spectrum Reconstruction Method of Propeller Noise under Small Sample Conditions

Abstract

The invention discloses a propeller noise line spectrum reconstruction method under a small sample condition. The reconstruction method comprises the following steps: (1) collecting a noise signal ofa propeller; (2) detecting the collected noise signal according to a spectrum related function to obtain a cyclic modulation spectrum; (3) performing compressed sensing observation on the cyclic modulation spectrum by using a measurement matrix to obtain a measurement result of the small sample, and then performing reconstruction to obtain a reconstructed cyclic modulation spectrum CMS1; (4) obtaining a corresponding enhanced envelope spectrum EES according to the cyclic modulation spectrum CMS, and obtaining the corresponding enhanced envelope spectrum EES1 according to the reconstructed cyclic modulation spectrum CMS1; and (5) performing line spectrum feature comparison through the feature information and a correlation coefficient. By utilizing a reconstruction method disclosed by the invention, the interference of the liquid noise, the environment noise and like carrier signal can be demodulated, the propeller original signal feature is accurately reconstructed, thereby identifyingthe service performance of the propeller; and the method has strong practicability.

Description

technical field [0001] The invention belongs to the field of signal processing, and in particular relates to a method for reconstructing propeller noise line spectrum under the condition of small samples. Background technique [0002] Propellers are important power components for many vehicles and propulsion equipment. Because it usually works in high-load operation and complex environment conditions, its signal components often contain complex non-stationary signals composed of environmental noise (carrier) and modulation signals of the propeller itself (shaft frequency, blade frequency, etc.) , the complex coupling relationship between various signals makes many traditional signal analysis methods no longer applicable, and the characteristic line spectrum cannot be extracted, which will bury many signal performances in the initial stage of faults. For some new signal feature extraction methods, accurate results It relies on a large number of sampling points under the cond...

AI Technical Summary

Problems solved by technology

However, these methods are all based on the fact that the target signal is a stationary signal, and most of the signals in the real industrial environment are non-stationary signals. For rotating machinery such as propellers, the noise signal is a special non-stationary signal. The implicit periodic characteristics need to be reflected by high-order statistics, so traditional signal processing methods are not applicable; at the same time, the characteristics of the signal at the initial stage of the fault are often relatively weak, and are modulated by various signals, making it difficult to extract features

[0004] Cyclostationary analysis and line spectrum reconstruction usually use second-order or higher-order statistics for feature extraction, because accurate high-order cyclic statistics depend on longer time series and higher sampling rates, and massive data It brings a huge burden to data storage and transmission, and also reduces the efficiency of data processing. Therefore, it is necessary to find a method to obtain signal characteristic line spectrum under small sample conditions to improve the timeliness and accuracy of signal analysis and processing.

Images