Earthquake data power frequency interference automatic identification and suppression method based on principal component analysis

Abstract

The invention relates to an earthquake data power frequency interference automatic identification and suppression method based on principal component analysis. The method comprises the steps of through calculating of average energy values of all paths of earthquake records, automatically identifying power frequency interference paths according to difference of the average energy values, extracting several adjacent paths including interference paths to form an earthquake interference path set, and using a principal component analysis technology to reconstruct feature vectors according to strong correlation of power frequency interference noises so that the purpose of suppressing the power frequency interference can be achieved. Since the principal component analysis is introduced, denoising is achieved, and the energy of effective signals is weakened at the same time, for this purpose, the amplitude-keeping processing is conducted on reconstructed data, and the path consistency of the earthquake records is guaranteed. As is proved, through the earthquake data power frequency interference automatic identification and suppression method based on principal component analysis, automatic identification and suppression on power frequency interference in earthquake data can be achieved, the quality of the earthquake data is effectively improved, and the earthquake data power frequency interference automatic identification and suppression method has important significance particularly in repairing destroyed paths of original earthquake records caused by the power frequency interference.

Description

Technical field: [0001] The invention relates to a seismic data processing and denoising method, in particular to a seismic record denoising method with severe power frequency interference, a method for identifying interfering channels by using the average energy value of each channel of seismic recording and suppressing power frequency interference by using principal component analysis technology . Background technique: [0002] The seismic exploration data acquisition process is often affected by power frequency interference. The collected seismic records are often mixed with power frequency noise. When the power frequency interference is very strong and runs through the whole of a seismic data, the seismic trace often appears bad. Channel, causing degradation of seismic data quality. Through the analysis of a large amount of bad track data, many seismic tracks do not originate from strong interference sources near the ground, but often come from the problem of the geophone it...

AI Technical Summary

Problems solved by technology

The article "Automatic Identification and Suppression of Power Frequency Interference in Seismic Data" proposes the notch method to filter out industrial interference, but this method will also filter out seismic signals with a frequency of 50 Hz; the article "Extraction and Elimination of Strong Power Frequency Interference Waves" It is proposed to use the conjugate gradient algorithm in the time domain to extract single-frequency interference, but this method ignores the variability of power frequency in actual seismic data. The research shows that in different regions and at different times, the frequency of power frequency interference changes; the article "Suppression of Earthquakes The improvement and application of the cosine approximation method for industrial electrical interference in the data" proposes to filter out industrial interference based on the cosine approximation method, but this method cannot realize automatic identification of interference channels; In addition to industrial interference, but this method is inefficient and cannot automatically identify interference channels; the article "50Hz Industrial Interference Denoising Method and Application Based on Wiener Filter" proposes a Wiener filter to adjust the amplitude and phase matching of the reference track to predict single frequency noise, but this method is not suitable for power frequency interference with poor stability; the article "Power Frequency Interference Elimination Technology Based on Independent Component Analysis" proposes to separate the independent signal components from the mixed signal to eliminate industrial noise, but This method relies on experience and lacks strict theoretical basis;

[0003] CN104570118A discloses "a method for automatic identification and removal of industrial interference based on two factors", which proposes a sin-cosine weighted approximation method to deal with industrial interference, but this method ignores the frequency, phase and amplitude of the power frequency.

[0004] The method disclosed in CN103630935A "Method for Removing AC Interference Signals in Seismic Data" proposes a method for extracting power frequency in the frequency domain, but this method does not consider the existence of power frequency harmonics;

Images