Feature extraction method based on blind source separation algorithm of wavelet transform fusion for leakage acoustic wave

Abstract

The invention discloses a feature extraction method based on a blind source separation algorithm of wavelet transform fusion for a leakage acoustic wave. The feature extraction method comprises the following steps: arranging a sensor on a detected pipeline, carrying out signal acquisition on a leakage point through the sensor, and obtaining a leakage acoustic wave acquisition signal; pre-processing the leakage acoustic wave acquisition signal by virtue of wavelet transform to obtain observation signals, and processing the observation signals by virtue of the blind source separation algorithm to obtain a target signal; and evaluating the target signal in step 2, and optimizing the composition of the observation signals. The feature extraction method based on the blind source separation algorithm of wavelet transform fusion for the leakage acoustic wave, which is provided by the invention, has the following beneficial effects: the target processing signal is evaluated through two evaluation parameters, that is, leakage time sampling point deviation and amplitude loss; and the method is capable of accurately locating a leakage time, and obvious in compensation action on the leakage amplitude of the weak signal.

Description

technical field [0001] The invention relates to the technical field of pipeline detection, in particular to a leakage sound wave feature extraction method based on wavelet transform fusion blind source separation algorithm. Background technique [0002] At present, there are many leakage monitoring methods that can be applied to oil and gas pipelines. Among them, the acoustic wave method has many advantages compared with the traditional mass balance method, negative pressure wave method, and transient model method: high sensitivity, high positioning accuracy, and false alarms. Low rate, short detection time, strong adaptability; the measurement is the weak dynamic pressure change in the pipeline fluid, which has nothing to do with the absolute value of the pipeline operating pressure; the response frequency is wider, the detection range is wider, etc. [0003] When a gas pipeline leaks, an acoustic signal is generated, and as the propagation distance increases, the leak sign...

AI Technical Summary

Problems solved by technology

[0009] (1) Wavelet transform can extract low-frequency signal features, which is the most widely used signal processing method, but at the same time, wavelet transform also has obvious defects in signal extraction. In practical applications, the acquisition of low-frequency signal features requires When the original signal is decomposed in depth, it is easy to have a large deviation in the location of the leakage time and the acquisition of the leakage amplitude, which is likely to cause a calculation error of the time difference, resulting in a large positioning error; the loss of the leakage amplitude is likely to cause distortion of the leakage waveform, which is easy to cause Missed and misjudged leaks

[0010] (2) In order to solve this problem, the blind source separation algorithm is used to process the signal. After research, it is found that blind source separation can accurately locate the leakage time, and there is no loss in the leakage amplitude, but it is compensated, especially in the signal The compensation is more obvious when it is relatively weak, but in application, blind source separation also has obvious defects: the order and type of target signals obtained by blind source separation cannot be determined

Images