Detector parasitic oscillation identification method and system

Abstract

The embodiment of the present application provides a geophone parasitic oscillation identification method and system, and belongs to the field of geophone performance detection. The method comprises the following steps: based on the seismic signal received by the geophone, calling a first arrival picking algorithm to determine the first arrival time of the geophone, and establishing the corresponding relationship between the first arrival time and the occurrence time of the interference signal; performing short-time Fourier transform on the seismic signal, and combining the optimized Hamming window to track the local frequency dynamics to obtain the frequency distribution characteristics corresponding to the parasitic oscillation; monitoring the energy change in the multiple frequency band of the natural frequency of the geophone, and extracting the energy anomaly characteristics corresponding to the parasitic oscillation; based on the corresponding relationship, the frequency distribution characteristics and the energy anomaly characteristics, the seismic signal is jointly discriminated, and the identification result of the geophone parasitic oscillation is output. The present application can accurately distinguish parasitic oscillation from other interference signals, thereby improving the reliability of geophone performance evaluation and seismic data acquisition.

Description

Technical Field

[0001] This invention relates to the field of detector performance testing, and more specifically to a method and system for identifying parasitic oscillations in detectors. Background Technology

[0002] In seismic exploration, the operational status of the geophone directly determines the authenticity and reliability of the acquired data. However, when receiving external excitation, the geophone often exhibits parasitic oscillations. These oscillations are not caused by signals reflected from geological bodies, but rather by additional responses triggered by the geophone's own mechanical and electrical characteristics. Once these parasitic oscillations are superimposed on the valid seismic signal, they manifest as high-frequency periodic components, leading to waveform distortion and spectral anomalies in the seismic record. For the interpretation of seismic data that relies on high-fidelity signals, this distortion can obscure information about target layers, and in severe cases, may even mislead the determination of reservoir characteristics.

[0003] Existing detection methods have significant limitations in identifying parasitic oscillations. Common methods mainly rely on manual interpretation or simple spectral analysis, but the frequency range of parasitic oscillations often overlaps with atmospheric interference or industrial noise, making accurate differentiation difficult solely from a frequency domain perspective. Furthermore, in the time domain, parasitic oscillations typically exhibit spiky waveforms, which are easily confused with random noise. Due to the lack of systematic identification standards and discrimination procedures, parasitic oscillations are often not confirmed in a timely manner, thus continuously impacting data quality during fieldwork and subsequent data processing.

[0004] Parasitic oscillations are particularly prominent for low-frequency geophones. For example, in the application of a 5Hz geophone, its parasitic oscillation frequency range may fall within the target exploration frequency band. If this range is not identified and distinguished, it will directly interfere with the energy distribution of the effective reflected wave. This problem not only reduces the accuracy of geophone performance evaluation but also weakens the reliability of seismic data acquisition.

[0005] Therefore, there is currently a lack of a method that can effectively distinguish parasitic oscillations from other interference signals. This deficiency severely restricts the screening and application of detector quality and has become a key technical bottleneck restricting the improvement of seismic exploration data quality. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for identifying parasitic oscillations in detectors, so as to at least solve the problem that existing solutions lack effective means to distinguish parasitic oscillations from other interference signals, which leads to distortion of seismic data.

[0007] To achieve the above objectives, a first aspect of the present invention provides a method for identifying parasitic oscillations in a geophone. The method includes: based on a seismic signal received by the geophone, calling an first-arrival picking algorithm to determine the first-arrival time of the geophone and establishing a correspondence between the first-arrival time and the occurrence time of the interference signal; performing a short-time Fourier transform on the seismic signal and combining it with an optimized Hamming window to track local frequency dynamics and obtain frequency distribution characteristics corresponding to the parasitic oscillations; monitoring energy changes within the octave band of the geophone's natural frequency and extracting energy anomaly characteristics corresponding to the parasitic oscillations; and jointly discriminating the seismic signal based on the correspondence, the frequency distribution characteristics, and the energy anomaly characteristics to output the identification result of the geophone's parasitic oscillations.

[0008] Optionally, based on the seismic signal received by the geophone, the first arrival time of the geophone is determined by calling the first arrival picking algorithm, including: preprocessing the seismic signal including denoising filtering and baseline correction; calculating the energy accumulation curve of a short time window in the preprocessed seismic signal, and extracting candidate first arrival points based on the energy accumulation curve; performing waveform correlation calculation within a sliding window on each candidate first arrival point, and selecting the target point with the greatest difference from the background noise waveform as the first arrival time.

[0009] Optionally, establishing the correspondence between the arrival time and the occurrence time of the interference signal includes: based on the determined arrival time, extracting an analysis time window containing a preset duration from the seismic signal, and using the analysis time window as a candidate interval for interference detection; calculating the instantaneous energy envelope curve within the candidate interval, and locating energy abrupt change points on the energy envelope curve, the energy abrupt change points corresponding to the occurrence time of the interference signal; calculating the difference between the occurrence time of the interference signal and the arrival time to generate a time-domain correspondence parameter, which serves as the correspondence.

[0010] Optionally, the rules for performing short-time Fourier transform on the seismic signal are as follows: the seismic signal is divided into multiple consecutive frames, the length of each frame is determined according to a preset time step, and some data overlap is maintained between adjacent frames; the short-time Fourier transform operation function is called one by one for each frame of data to generate the local spectrum of the corresponding frame; the local spectra are arranged and combined in chronological order to form a complete time spectrum result.

[0011] Optionally, by combining optimized Hamming window tracking of local frequency dynamics to obtain frequency distribution characteristics corresponding to parasitic oscillations, the following steps are taken: applying a Hamming window function during the generation of each local spectrum and setting the window length parameter based on an integer multiple of the detector's natural frequency; performing amplitude envelope analysis on each local spectrum after applying the Hamming window function to extract the dominant frequency component of each local spectrum; arranging the dominant frequency components in chronological order to form a frequency dynamic sequence, and locating frequency abrupt change points and stable intervals in the frequency dynamic sequence as frequency distribution characteristics to characterize the triggering position and duration of parasitic oscillations.

[0012] Optionally, monitoring energy changes within the octave band of the detector's natural frequency and extracting energy anomaly features corresponding to parasitic oscillations includes: extracting preset octave band spectrum data covering integer multiples of the detector's natural frequency from the time-spectrum results; performing power spectral density calculation on the octave band spectrum data to obtain an energy distribution curve that changes over time; identifying energy spikes and energy decay intervals on the energy distribution curve, using the energy spikes as trigger markers for parasitic oscillations and the energy decay intervals as persistent features of parasitic oscillations; and combining the trigger markers and persistent features to form energy anomaly features.

[0013] Optionally, power spectral density calculation is performed on the octave band spectrum data to obtain an energy distribution curve that changes over time. This includes: organizing the octave band spectrum data into segments according to time sequence to generate a continuous octave band time-series spectrum; calling the power spectral density calculation function on the octave band time-series spectrum to calculate the squared amplitude of each frequency component in each time period and perform normalization processing to obtain the energy value of the corresponding time period; and arranging the energy values ​​of each time period in sequence to form an energy distribution curve covering the entire analysis period.

[0014] Optionally, the seismic signal is jointly discriminated based on the correspondence, the frequency distribution characteristics, and the energy anomaly characteristics to output the identification result of the detector parasitic oscillation, including: inputting the correspondence parameters, the frequency distribution characteristics, and the energy anomaly characteristics into a joint discrimination model to establish a one-to-one correspondence between the three types of features; performing feature matching operations in the joint discrimination model to determine whether the correspondence parameters and the frequency distribution characteristics are consistent in the time dimension, and simultaneously verifying the changing trend of the energy anomaly characteristics within the corresponding time period; when the three types of features simultaneously meet the preset consistency conditions in the discrimination model, outputting a parasitic oscillation identification identifier, and associating the parasitic oscillation identification identifier with the corresponding time period; combining the parasitic oscillation identification identifier with the associated time period to form the detector parasitic oscillation identification result.

[0015] Optionally, feature matching is performed in the joint discrimination model to determine whether the correspondence parameters and frequency distribution features are consistent in the time dimension, and to simultaneously verify the changing trend of the energy anomaly features within the corresponding time period. This includes: mapping the correspondence parameters to a time difference sequence and synchronizing the dominant frequency dynamic sequence in the frequency distribution features to the same time coordinate; calculating the intersection interval of the correspondence time difference sequence and the dominant frequency dynamic sequence under the time coordinate to determine their consistency in the time dimension; extracting the abrupt increase point and attenuation interval of the energy anomaly features within the intersection interval, and comparing the abrupt change point and stable interval of the dominant frequency dynamic sequence to determine whether the energy change trend corresponds to the frequency change trend; when the time difference sequence, the dominant frequency dynamic sequence, and the energy anomaly features all meet the preset consistency condition within the intersection interval, the determination result of feature matching passing is output.

[0016] Optionally, the method further includes: acquiring a reference detector signal deployed adjacent to the detector, and extracting the first arrival time, frequency distribution characteristics, and energy anomaly characteristics of the reference signal under the same processing flow; comparing the characteristic parameters of the reference signal with the characteristic parameters of the target detector item by item, and determining that the target detector has parasitic oscillation interference when the reference signal does not show parasitic oscillation characteristics but the target detector does; verifying the consistency between the comparison result and the determination result of feature matching, and confirming the validity of the parasitic oscillation identification conclusion when they are consistent. A second aspect of the present invention provides a detector parasitic oscillation identification system, the system comprising: an acquisition unit, configured to determine the first arrival time of the detector based on a seismic signal received by the detector by calling an first arrival picking algorithm, and establishing a correspondence between the first arrival time and the occurrence time of the interference signal; a processing unit, configured to perform a short-time Fourier transform on the seismic signal, and combine an optimized Hamming window to track local frequency dynamics, and obtain frequency distribution characteristics corresponding to the parasitic oscillation; a monitoring unit, configured to monitor energy changes within the octave band of the detector's natural frequency, and extract energy anomaly characteristics corresponding to the parasitic oscillation; and an output unit, configured to jointly discriminate the seismic signal based on the correspondence, the frequency distribution characteristics, and the energy anomaly characteristics, and output the identification result of the detector parasitic oscillation. A third aspect of the present invention provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the above-described detector parasitic oscillation identification method.

[0017] A fourth aspect of the present invention provides an electronic device, the electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described detector parasitic oscillation identification method.

[0018] The fifth aspect of the present invention provides a computer program product, including a computer program that, when executed by a processor, implements the above-described detector parasitic oscillation identification method.

[0019] Through the above technical solution, this invention utilizes first-arrival pickup to establish a time reference for the occurrence of interference, avoiding temporal confusion between parasitic oscillations and valid signals. Secondly, by combining short-time Fourier transform with an optimized window function, it dynamically tracks local frequency changes, highlighting the existence of parasitic oscillations from spectral characteristics. Furthermore, by combining energy monitoring in the natural frequency octave band, it extracts energy mutations as a typical manifestation of parasitic oscillations. Finally, in the joint discrimination stage, the three characteristics of time, frequency, and energy are comprehensively verified, effectively eliminating interference factors such as atmospheric noise or industrial interference, ensuring that parasitic oscillations can be reliably distinguished. Therefore, this not only improves the accuracy of detector performance evaluation but also provides assurance for the authenticity of seismic data and subsequent interpretation.

[0020] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0021] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of the steps of a detector parasitic oscillation identification method provided by one embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the manifestation of parasitic oscillations in seismic records according to one embodiment of the present invention; Figure 3 This is a system structure diagram of a detector parasitic oscillation identification system provided in one embodiment of the present invention. Detailed Implementation

[0022] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0023] Figure 1 This is a flowchart illustrating the steps of a detector parasitic oscillation identification method according to one embodiment of the present invention. Figure 1 As shown, an embodiment of the present invention provides a method for identifying parasitic oscillations in a detector, the method comprising: Step S10: Based on the seismic signal received by the detector, call the first arrival picking algorithm to determine the first arrival time of the detector, and establish the correspondence between the first arrival time and the time of occurrence of the interference signal.

[0024] Specifically, based on the seismic signal received by the geophone, the first arrival time of the geophone is determined by calling the first arrival picking algorithm, including: preprocessing the seismic signal including noise reduction filtering and baseline correction; calculating the short-time-window energy accumulation curve in the preprocessed seismic signal, and extracting candidate first arrival points based on the energy accumulation curve; performing waveform correlation calculation within a sliding window on each candidate first arrival point, and selecting the target point with the greatest difference from the background noise waveform as the first arrival time.

[0025] Furthermore, establishing the correspondence between the arrival time and the occurrence time of the interference signal includes: based on the determined arrival time, extracting an analysis time window containing a preset duration from the seismic signal, and using the analysis time window as a candidate interval for interference detection; calculating the instantaneous energy envelope curve within the candidate interval, and locating energy abrupt change points on the energy envelope curve, the energy abrupt change points corresponding to the occurrence time of the interference signal; calculating the difference between the occurrence time of the interference signal and the arrival time to generate a time-domain correspondence parameter, which serves as the correspondence.

[0026] In this embodiment of the invention, based on the seismic signal received by the geophone, the first arrival time of the geophone is determined by calling the first arrival picking algorithm, and a correspondence is established between the first arrival time and the time of occurrence of the interference signal. Here, the so-called first arrival time essentially refers to the time point when the geophone first receives the effective energy of the seismic wave, which usually reflects the earliest arriving effective wavefront in the signal. To ensure the accuracy of the first arrival picking, the original seismic signal received by the geophone needs to be preprocessed. Specifically, firstly, a noise reduction filtering operation is performed to suppress the high-frequency random components introduced by environmental noise, industrial interference, etc., to ensure that the main waveform is not masked by abnormal energy. The filtering method can be a bandpass filter to filter out invalid components far from the target frequency band. Secondly, a baseline correction operation needs to be performed on the seismic signal, that is, to remove the DC offset in the signal caused by the electronic characteristics of the geophone or the external electromagnetic environment, so that the signal can be presented in the form of zero mean, so that baseline drift will not interfere with the extraction of candidate points during subsequent energy analysis.

[0027] After preprocessing, candidate first arrival points need to be extracted. To this end, a short-time-window energy accumulation curve is constructed. Specifically, the preprocessed signal is segmented into continuously sliding small windows. The instantaneous energy of the signal is calculated and accumulated within each window, and these energies are then arranged sequentially over time to obtain a curve showing the energy evolution over time. On this energy accumulation curve, significant energy abrupt changes often occur near the first arrival points, reflecting the physical characteristics of the arrival of the effective wavefront of the seismic wave. Therefore, based on the energy accumulation curve, thresholds or abrupt change criteria can be set to automatically extract several candidate first arrival points.

[0028] However, relying solely on the energy curve cannot accurately determine the first arrival time, as sudden noise or localized interference can also cause an energy increase. To further improve accuracy, waveform correlation analysis needs to be performed near the candidate first arrival point. The rule is as follows: within a sliding window of the candidate point, the waveform of the signal segment is correlated with the waveform of the background noise segment, quantifying the difference between them. If the correlation is extremely low, it indicates that the signal waveform at that candidate point differs most from the background noise, and is therefore more likely to be the actual first arrival point. In this way, the target point with the greatest difference from the background noise waveform can be selected from multiple candidate points, and ultimately determined as the detector's first arrival time. This result will serve as an important time reference for all subsequent parasitic oscillation identification processes.

[0029] After obtaining the first arrival time, it is necessary to establish a correspondence between the first arrival time and the occurrence time of interference signals in order to accurately distinguish between the occurrence of effective wavefronts and parasitic oscillation interference in the time domain. Specifically, based on the determined first arrival time, an analysis time window containing a preset duration is extracted from the seismic signal. This time window should cover a key interval after the first arrival point, typically selecting a time range that includes the potential occurrence of parasitic oscillations. This time window is used as a candidate interval for interference detection, ensuring that the analysis focuses on signal segments where interference may exist.

[0030] Within the candidate interval, the instantaneous energy envelope curve of the signal is calculated. This energy envelope is obtained by extracting the envelope curve after squaring the signal amplitude. This transforms the rapidly changing waveform into a smooth energy profile curve, facilitating the observation of abrupt changes in energy over time. After obtaining the energy envelope curve, abrupt change points are further located. An abrupt change point is the location where the energy envelope curve shows a rapid rise or fall, which usually indicates the appearance of new high-frequency or periodic components in the signal. Since parasitic oscillations typically manifest in the time domain as sudden appearances of periodic high-frequency components, this abrupt change point can be directly identified as the occurrence time of the interference signal.

[0031] Based on this, the time difference between the occurrence time of the interference signal and the previously acquired first arrival time is calculated to obtain the time interval between the two. This time interval reflects the delay of the parasitic oscillation relative to the arrival of the effective wavefront. It can be used to characterize the relative position of the parasitic oscillation and can also be used as a time-domain feature parameter input into the subsequent discrimination model. Therefore, through this difference calculation, a time-domain correspondence parameter is generated and defined as the "correspondence". This parameter not only indicates the temporal location of the parasitic oscillation interference but also establishes a direct link between the effective signal and the interference signal.

[0032] Overall, this implementation method completes a comprehensive chain from determining the first arrival time to identifying the time of interference signal appearance, and generating the corresponding parameters between the two, through preprocessing of seismic signals, extraction of candidate first arrival points, waveform correlation screening, location of energy envelope abrupt change points, and difference calculation. This ensures the accuracy of first arrival picking and avoids misjudging interference signals as valid wavefronts. Furthermore, by establishing a correspondence with the time points of interference signals, parasitic oscillations can be separated from valid signals in the time domain, providing a reliable reference point for subsequent frequency domain analysis and energy anomaly detection. This method provides a clear time reference for the identification of parasitic oscillations, thereby improving the reliability and stability of parasitic oscillation discrimination.

[0033] Step S20: Perform a short-time Fourier transform on the seismic signal and combine it with an optimized Hamming window to track local frequency dynamics and obtain the frequency distribution characteristics corresponding to the parasitic oscillations.

[0034] Specifically, the rules for performing short-time Fourier transform on the seismic signal are as follows: the seismic signal is divided into multiple consecutive frames, the length of each frame is determined according to a preset time step, and some data overlap is maintained between adjacent frames; the short-time Fourier transform operation function is called one by one for each frame of data to generate the local spectrum of the corresponding frame; the local spectra are arranged and combined in chronological order to form a complete time spectrum result.

[0035] Furthermore, by combining optimized Hamming windows to track local frequency dynamics, the frequency distribution characteristics corresponding to parasitic oscillations are obtained, including: applying a Hamming window function during the generation of each local spectrum and setting the window length parameter based on an integer multiple of the detector's natural frequency; performing amplitude envelope analysis on each local spectrum after applying the Hamming window function to extract the dominant frequency component of each local spectrum; arranging the dominant frequency components in chronological order to form a frequency dynamic sequence, and locating frequency abrupt change points and stable intervals in the frequency dynamic sequence as frequency distribution characteristics to characterize the triggering position and duration of parasitic oscillations.

[0036] In this embodiment of the invention, a short-time Fourier transform is performed on the seismic signal, and an optimized Hamming window is used to track local frequency dynamics to obtain the frequency distribution characteristics corresponding to parasitic oscillations. In the identification of parasitic oscillations, frequency characteristics are often the most direct criterion, because parasitic oscillations generally manifest as additional energy components located within integer multiples of the detector's natural frequency. These components exhibit significant abrupt changes and clustering characteristics in the frequency domain. Therefore, it is necessary to combine time and frequency information through time-frequency analysis to accurately capture the occurrence and evolution of parasitic oscillations.

[0037] Specifically, the rules for performing a short-time Fourier transform on the seismic signal are as follows: First, the seismic signal is divided into continuous frames, with the length of each frame determined according to a preset time step. The selection of this time step needs to consider both time-domain and frequency-domain resolution, generally setting a length that can cover several periods of signal based on the sampling rate to ensure that local frequency characteristics are fully represented. During frame division, a certain proportion of overlap is maintained between adjacent frames, typically around 50%, to avoid frequency leakage or loss of time information due to inter-frame breaks. In this way, the entire continuous signal is decomposed into multiple local segments, laying the foundation for subsequent frame-by-frame spectral calculations.

[0038] Next, the short-time Fourier transform function is called on each frame of data. Operationally, the signal sequence of each frame is multiplied by a complex exponential basis function and integrated to obtain the spectral distribution of the corresponding frame. This process essentially expands the frequency components within that frame. Since seismic signals contain both low-frequency reflected wave energy and potential high-frequency parasitic oscillations, it is necessary to retain as complete a frequency band information as possible during the calculation. After processing each frame, a set of local spectra for that frame is obtained. By repeating this process, the local spectra of all frames can be obtained sequentially.

[0039] Subsequently, the local spectra are arranged and combined chronologically to form a complete time-spectrum result covering the entire analysis period. The time-spectrum is a two-dimensional representation where the horizontal axis represents time, the vertical axis represents frequency, and the amplitude represents the energy level of the frequency component at that moment. This graphical representation allows for a direct observation of the frequency distribution of signal energy across different time periods. Parasitic oscillations often appear suddenly within a specific time period and accumulate energy within a certain frequency range; these phenomena can be directly reflected in the time-spectrum.

[0040] Furthermore, in the aforementioned time-frequency spectrum generation process, an optimized Hamming window function is needed to improve the accuracy of the analysis. Specifically, a Hamming window function is applied before performing a Fourier transform on each frame of signal. The Hamming window is a commonly used weighted window function that can effectively reduce spectral leakage while maintaining frequency resolution, making the boundaries of frequency components clearer. To match the detector characteristics, this implementation sets the window length parameter to an integer multiple of the detector's natural frequency, ensuring that the window function's frequency domain coverage matches the frequency bands where parasitic oscillations may occur. Through this targeted optimization, time-frequency analysis can more sensitively capture the local frequency dynamics of parasitic oscillations.

[0041] After applying the window function and generating the local spectra, further processing of the spectrum is required, namely, performing amplitude envelope analysis on each local spectrum. Specifically, this involves calculating the square of the amplitude of each frequency component and extracting its envelope to obtain the variation of the dominant frequency component. The introduction of the amplitude envelope effectively smooths out random fluctuations, highlighting the stable variation trend of the main frequency components. After extracting the dominant frequency components from the local spectrum of each frame, these components are arranged in chronological order to form a dynamic frequency sequence. This sequence visually reflects the evolution of the dominant frequency of the signal over time.

[0042] Next, within the aforementioned frequency dynamic sequence, it is necessary to locate frequency abrupt change points and stable intervals. A frequency abrupt change point refers to the location where the dominant frequency experiences a sharp jump within a short period. This often corresponds to the triggering moment of parasitic oscillations, as the addition of parasitic oscillations superimposes new frequency components onto the existing frequency structure, causing a significant abrupt change in the dominant frequency. A stable interval, on the other hand, refers to a state where the dominant frequency remains relatively constant over a period of time, typically corresponding to the sustained phase of parasitic oscillations. In analysis, abrupt change points can be identified by setting a threshold, or stable intervals can be defined by variance calculation.

[0043] Ultimately, combining the frequency abrupt change points and stable intervals constitutes the frequency distribution characteristics of parasitic oscillations. This characteristic contains two dimensions of information: first, the temporal location of the parasitic oscillation, i.e., when it is triggered and how long it lasts; second, the frequency location of the parasitic oscillation, i.e., in which frequency bands it is concentrated. By combining these two dimensions, the frequency behavior of parasitic oscillations can be comprehensively characterized.

[0044] Step S30: Monitor energy changes within the octave band of the detector's natural frequency and extract energy anomaly features corresponding to parasitic oscillations.

[0045] Specifically, a preset octave band spectrum data covering integer multiples of the detector's natural frequency is extracted from the time-frequency spectrum results; power spectral density calculation is performed on the octave band spectrum data to obtain an energy distribution curve that changes over time; energy spikes and energy decay intervals are identified on the energy distribution curve, with energy spikes serving as trigger markers for parasitic oscillations and energy decay intervals serving as persistent characteristics of parasitic oscillations; the trigger markers and persistent characteristics are combined to form an energy anomaly characteristic.

[0046] Furthermore, power spectral density calculation is performed on the octave band spectrum data to obtain an energy distribution curve that changes over time. This includes: organizing the octave band spectrum data into segments according to time sequence to generate a continuous octave band time-series spectrum; calling the power spectral density calculation function on the octave band time-series spectrum to calculate the square amplitude of each frequency component in each time period and perform normalization processing to obtain the energy value of the corresponding time period; and arranging the energy values ​​of each time period in sequence to form an energy distribution curve covering the entire analysis period.

[0047] In this embodiment of the invention, energy changes are monitored within the octave band of the detector's natural frequency to extract energy anomaly features corresponding to parasitic oscillations. A significant characteristic of parasitic oscillations is that their frequencies often appear within integer multiples of the detector's natural frequency, and they produce prominent energy peaks or anomalies in this frequency band. Therefore, if a mechanism for discriminating energy changes over time can be established within these specific octave bands, parasitic oscillations can be distinguished from general noise or normal seismic wave energy, thereby achieving more reliable identification.

[0048] Specifically, it is necessary to first extract a preset octave band of spectrum data covering integer multiples of the detector's natural frequency from the time-spectrum results. The time-spectrum results already provide the energy distribution map of the signal at different times and frequencies. Based on this, a target frequency band is selected; for example, for a 5Hz detector, the spectrum data between 50–75Hz can be extracted. The advantage of doing this is that it avoids redundant data from full-band analysis and allows subsequent calculations to focus on the region where parasitic oscillations are most likely to occur. The extracted octave band spectrum data will serve as input for subsequent energy analysis.

[0049] Next, power spectral density calculations are performed on the octave band spectrum data to quantify the energy changes over time within that band. Power spectral density projects the energy distribution of a signal into the frequency domain, and its calculation results accurately reflect the dynamic changes of energy within a specific frequency band over time. To achieve this, the octave band spectrum data needs to be segmented in chronological order to generate continuous octave band time-series spectra. In other words, the entire analysis period is divided into multiple adjacent small segments, ensuring that each segment contains frequency information within the octave range. After this processing, energy calculations can be performed sequentially on each time slice.

[0050] The power spectral density calculation function is invoked on the time-series spectrum of the aforementioned octave band to calculate the squared amplitude of each frequency component within each time period, and then normalize it. The calculation of the squared amplitude converts the signal strength into an energy quantization value, while the normalization process eliminates potential differences in sampling intensity between different time periods, allowing for direct comparison of energy values ​​across different time segments. Ultimately, a numerical value representing the energy level of each time period is obtained.

[0051] The energy values ​​for each time period are arranged sequentially to form an energy distribution curve covering the entire analysis period. This curve is essentially a continuous function of energy over time, clearly demonstrating the dynamic changes in octave band energy. Under normal circumstances, the octave band energy curve should be relatively stable without drastic abrupt changes; however, when parasitic oscillations occur, one or more sharp increases in energy will suddenly appear on the curve, which are typical manifestations of energy anomalies.

[0052] Identifying energy spikes and decay intervals on the energy distribution curve is crucial for extracting parasitic oscillation characteristics. Energy spikes often signify the trigger point of parasitic oscillations because the detector generates high-frequency additional vibrations unrelated to external excitation at this moment, causing a sudden energy jump within the octave band. Energy decay intervals reflect the duration of the parasitic oscillation, during which the energy gradually decreases but remains above the background energy level. By using energy spikes as trigger markers for parasitic oscillations and energy decay intervals as characteristics of their persistence, a more complete description of the energy dimension of parasitic oscillations can be obtained.

[0053] Finally, the trigger identifier and the persistence feature are combined to form an energy anomaly feature. This feature contains not only the initiation information of the parasitic oscillation, but also its duration and decay trend, providing strong energy dimension support for subsequent joint discrimination. By combining it with time-domain correspondence and frequency distribution characteristics, the energy anomaly feature can help distinguish genuine parasitic oscillations from external random interference. For example, atmospheric noise may also cause energy fluctuations, but it often lacks a correspondence with the first arrival time and does not conform to the typical energy pattern of parasitic oscillations in the octave band.

[0054] This implementation method locks the parasitic oscillation phenomenon into its most likely frequency range by monitoring energy within the octave band of the detector's natural frequency. Furthermore, it quantifies its sudden and persistent characteristics through power spectral density calculation and energy curve analysis. In this way, parasitic oscillations can be reliably identified in terms of energy, eliminating reliance on manual experience and thus improving the objectivity and accuracy of parasitic oscillation detection.

[0055] Step S40: Based on the correspondence, the frequency distribution characteristics, and the energy anomaly characteristics, the seismic signal is jointly discriminated, and the identification result of the detector parasitic oscillation is output.

[0056] Specifically, the correspondence parameters, frequency distribution features, and energy anomaly features are input into a joint discrimination model to establish a one-to-one correspondence between the three types of features. Feature matching is performed in the joint discrimination model to determine whether the correspondence parameters and frequency distribution features are consistent in the time dimension, and the changing trend of the energy anomaly features within the corresponding time period is simultaneously verified. When the three types of features simultaneously meet the preset consistency conditions in the discrimination model, a parasitic oscillation identification identifier is output, and the parasitic oscillation identification identifier is associated with the corresponding time period. The parasitic oscillation identification identifier is combined with the associated time period to form the detector parasitic oscillation identification result.

[0057] Furthermore, feature matching operations are performed in the joint discrimination model to determine whether the correspondence parameters and frequency distribution features are consistent in the time dimension, and to simultaneously verify the changing trend of the energy anomaly features within the corresponding time period. This includes: mapping the correspondence parameters to a time difference sequence and synchronizing the dominant frequency dynamic sequence in the frequency distribution features to the same time coordinate; calculating the intersection interval of the correspondence time difference sequence and the dominant frequency dynamic sequence under the time coordinate to determine their consistency in the time dimension; extracting the abrupt increase point and attenuation interval of the energy anomaly features within the intersection interval, and comparing the abrupt change point and stable interval of the dominant frequency dynamic sequence to determine whether the energy change trend maintains a corresponding relationship with the frequency change trend; when the time difference sequence, the dominant frequency dynamic sequence, and the energy anomaly features all meet the preset consistency condition within the intersection interval, the determination result of feature matching passing is output.

[0058] In this embodiment of the invention, the seismic signal is jointly discriminated based on the correspondence, the frequency distribution characteristics, and the energy anomaly characteristics to output the identification result of the detector's parasitic oscillations. Joint discrimination is necessary because parasitic oscillations are easily confused with other interferences under a single feature dimension. For example, relying solely on time-domain spike characteristics can easily mistake random noise for oscillations; relying solely on frequency clustering phenomena in the frequency domain can easily confuse them with atmospheric noise; and relying solely on abrupt changes in the energy curve may overlap with industrial interference. Therefore, it is necessary to integrate the features of the three dimensions, forming a mutually corroborating judgment chain in terms of time, frequency, and energy, in order to output a reliable identification result.

[0059] Specifically, the correspondence parameters, frequency distribution features, and energy anomaly features are input into the joint discrimination model to establish a one-to-one correspondence between the three types of features. The correspondence parameters are derived from the difference between the arrival time and the time of interference occurrence, reflecting the relative temporal position of the parasitic oscillation; the frequency distribution features are derived from the frequency dynamic sequence optimized by short-time Fourier transform and Hamming window, reflecting the triggering frequency and duration range of the parasitic oscillation; and the energy anomaly features are derived from the energy distribution curve calculated by power spectral density, reflecting the energy bursts and attenuation of the parasitic oscillation within the octave band. When the three types of features enter the same discrimination framework, a mapping relationship needs to be established to ensure that they can be compared and matched under a unified time coordinate.

[0060] In the joint discrimination model, feature matching is performed to determine whether the correspondence parameters and frequency distribution features are consistent in the time dimension, and simultaneously verify the changing trend of the energy anomaly features within the corresponding time period. This feature matching operation is not an abstract "judgment," but rather has explicit execution steps. First, the correspondence parameters need to be mapped to a time difference sequence, which indicates the delay of the parasitic oscillation relative to its first arrival time. Simultaneously, the dominant frequency dynamic sequence in the frequency distribution features needs to be synchronized to the same time coordinate. In other words, the frequency dynamics and time delay must be unified under the same reference system to determine the consistency of the parasitic oscillation in the time dimension.

[0061] Under a unified time coordinate, the intersection interval between the corresponding time difference sequence and the dominant frequency dynamic sequence is calculated. The intersection interval refers to the time period during which both types of features are simultaneously effective. If parasitic oscillations do exist, then within a certain time delay after the first arrival time, both frequency abrupt changes and significant time-domain differences in the correspondence should be observable. By taking the intersection interval, the analytical focus of the three types of features is effectively locked onto the same time period, creating conditions for further energy trend comparison.

[0062] Within the intersection interval, the points of sudden increases and decay intervals of energy anomalies are extracted, and compared with the abrupt change points and stable intervals of the dominant frequency dynamic sequence. In other words, within this intersection interval, it is necessary to observe not only whether the frequency jumps or enters a stable state, but also whether the energy correspondingly increases and subsequently decays. A typical characteristic of parasitic oscillations is that at the trigger point, the energy suddenly rises while the frequency abruptly changes; during the sustained phase, the energy gradually decays while the frequency remains stable. If these two types of characteristics—energy and frequency—form this correspondence within the intersection interval, it indicates that the occurrence of parasitic oscillations is credible.

[0063] When the time difference sequence, the dominant frequency dynamic sequence, and the energy anomaly feature all meet preset consistency conditions within their intersection interval, the output result indicating successful feature matching is determined. The consistency conditions can include the following categories: first, a time condition, meaning the energy surge point must fall within a preset delay range after the initial arrival time; second, a frequency condition, meaning the dominant frequency must be within the octave band of the detector's natural frequency; third, an energy condition, meaning the energy surge magnitude must be higher than the statistical threshold of the background energy; and fourth, a trend condition, meaning the energy decay process and the frequency stability interval must overlap. Only when all these conditions are simultaneously met will the discrimination model consider that parasitic oscillations have occurred.

[0064] When a feature match is successful, a parasitic oscillation identification marker is generated and associated with the corresponding time period. This marker is not merely a "yes" or "no" conclusion, but contains complete information such as the triggering time, duration, frequency range, and energy variation of the parasitic oscillation. This information, associated with the corresponding time period, can be used to locate the parasitic oscillation in the seismic record. Finally, the parasitic oscillation identification marker is combined with the associated time period to form the geophone parasitic oscillation identification result.

[0065] This implementation introduces a cross-validation mechanism using three features—time, frequency, and energy—through a joint discrimination model, effectively avoiding misjudgments caused by single-dimensional identification. The correspondence in the time domain ensures that the oscillation signal is not confused with the first arrival wave; the dynamic features in the frequency domain ensure that the oscillation signal is distinguishable from background noise; and the sudden increases and decreases in energy provide quantitative evidence of triggering and persistence. When all three simultaneously meet the consistency condition, the reliability of the identification result is significantly improved, effectively distinguishing parasitic oscillations from atmospheric interference, industrial noise, and even random spikes.

[0066] Preferably, the method further includes: acquiring a reference detector signal deployed adjacent to the detector, and extracting the first arrival time, frequency distribution characteristics, and energy anomaly characteristics of the reference signal under the same processing flow; comparing the characteristic parameters of the reference signal with the characteristic parameters of the target detector item by item, and determining that the target detector has parasitic oscillation interference when the reference signal does not show parasitic oscillation characteristics but the target detector does; verifying the consistency between the comparison result and the determination result of feature matching, and confirming the validity of the parasitic oscillation identification conclusion when they are consistent.

[0067] In this embodiment of the invention, the method further includes: acquiring a reference detector signal deployed adjacent to the target detector, and extracting the first arrival time, frequency distribution characteristics, and energy anomaly characteristics of the reference signal under the same processing procedure. The so-called adjacent reference detector refers to another detector located near the target detector on the same survey line or working surface, under similar geological environment and excitation conditions. By performing the same processing steps on the reference detector signal as on the target detector, including noise reduction filtering, baseline correction, first arrival acquisition, short-time Fourier transform, and octave band energy monitoring, its corresponding first arrival time, dominant frequency dynamics, and energy distribution curves can be obtained.

[0068] Based on this, the characteristic parameters of the reference signal are compared item by item with the characteristic parameters of the target detector. The comparison includes: comparing the difference in their first arrival times to determine if they are ahead of the same seismic wave; comparing frequency distribution characteristics to check if the reference detector experiences frequency abrupt changes within the same octave band; and comparing energy anomaly characteristics to observe whether the energy distribution curve of the reference detector has an energy surge point or attenuation range consistent with that of the target detector. When the reference detector does not exhibit parasitic oscillation characteristics, but the target detector exhibits obvious parasitic oscillation characteristics within the same time period and frequency range, it can be determined that the target detector does indeed have parasitic oscillation interference.

[0069] Furthermore, the comparison results are verified for consistency with the judgment results of feature matching. Specifically, if the aforementioned joint discrimination model has output an identification indicator that the target detector has parasitic oscillations, and the reference detector signal, after comparison verification, does not exhibit the same type of interference, then the conclusions of the two are consistent, and the validity of the parasitic oscillation identification result can be confirmed. In this way, misjudgments caused by external environment or regional interference can be eliminated, and the credibility of the identification results can be enhanced.

[0070] In one specific implementation, such as Figure 2 The seismic signal received by the geophone was first-arrival picked up, and waveform intervals before and after the first arrival were extracted. The waveform before the first arrival showed low overall signal energy and fluctuations close to background noise levels. However, the waveform after the first arrival exhibited significant high-frequency components and increased periodicity, indicating that parasitic oscillations were triggered during this period. A short-time Fourier transform was further performed on the entire signal to generate an amplitude spectrum. The amplitude spectrum clearly showed a characteristic curve with abrupt amplitude changes within the octave band of the geophone's natural frequency, exhibiting an abnormal "down-up-down" trend. Combined analysis of the time-domain first-arrival correspondence and the frequency-domain amplitude anomaly confirmed that the geophone exhibited parasitic oscillations during actual operation, providing direct evidence for performance evaluation and subsequent screening.

[0071] Figure 3 This is a system structure diagram of a detector parasitic oscillation identification system provided in one embodiment of the present invention. Figure 3 As shown, this invention provides a detector parasitic oscillation identification system. The system includes: an acquisition unit, used to determine the first arrival time of the detector based on the seismic signal received by the detector by calling an first arrival picking algorithm, and establishing a correspondence between the first arrival time and the occurrence time of the interference signal; a processing unit, used to perform a short-time Fourier transform on the seismic signal, and combine an optimized Hamming window to track local frequency dynamics to obtain the frequency distribution characteristics corresponding to the parasitic oscillation; a monitoring unit, used to monitor energy changes within the octave band of the detector's natural frequency, and extract energy anomaly characteristics corresponding to the parasitic oscillation; and an output unit, used to jointly discriminate the seismic signal based on the correspondence, the frequency distribution characteristics, and the energy anomaly characteristics, and output the identification result of the detector parasitic oscillation. A third aspect of the present invention provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the above-described detector parasitic oscillation identification method.

[0072] A fourth aspect of the present invention provides an electronic device, the electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described detector parasitic oscillation identification method.

[0073] The fifth aspect of the present invention provides a computer program product, including a computer program that, when executed by a processor, implements the above-described detector parasitic oscillation identification method.

[0074] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a microcontroller, chip, or processor to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0075] The optional embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details described above. Within the scope of the technical concept of the embodiments of the present invention, various simple modifications can be made to the technical solutions of the embodiments of the present invention, and these simple modifications all fall within the protection scope of the embodiments of the present invention. It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, the embodiments of the present invention will not further describe the various possible combinations.

[0076] Furthermore, various different embodiments of the present invention can be combined in any way, as long as they do not violate the spirit of the embodiments of the present invention, they should also be regarded as the content disclosed by the embodiments of the present invention.

Claims

1. A method for identifying parasitic oscillations in a detector, characterized in that, The method includes: Based on the seismic signal received by the detector, the first arrival time of the detector is determined by calling the first arrival picking algorithm, and a correspondence is established between the first arrival time and the time of occurrence of the interference signal. A short-time Fourier transform is performed on the seismic signal, and an optimized Hamming window is used to track local frequency dynamics to obtain the frequency distribution characteristics corresponding to the parasitic oscillations. Energy changes are monitored within the octave band of the detector's natural frequency to extract energy anomaly features corresponding to parasitic oscillations; Based on the correspondence, frequency distribution characteristics, and energy anomaly characteristics, the seismic signal is jointly discriminated, and the identification result of the detector parasitic oscillation is output.

2. The method according to claim 1, characterized in that, Based on the seismic signal received by the geophone, the first arrival time of the geophone is determined by calling the first arrival picking algorithm, including: The seismic signal is preprocessed, including noise reduction filtering and baseline correction. Calculate the short-time-window energy accumulation curve in the preprocessed seismic signal, and extract candidate first arrival points based on the energy accumulation curve; For each candidate first arrival point, waveform correlation calculation is performed within a sliding window, and the target point with the greatest difference from the background noise waveform is selected as the first arrival time.

3. The method according to claim 1, characterized in that, Establishing the correspondence between the initial arrival time and the occurrence time of the interference signal includes: Based on the determined first arrival time, an analysis time window containing a preset duration is extracted from the seismic signal, and the analysis time window is used as a candidate interval for interference detection. Calculate the instantaneous energy envelope curve within the candidate interval, and locate the energy mutation point on the energy envelope curve, the energy mutation point corresponding to the time of occurrence of the interference signal; The difference between the time of occurrence of the interference signal and the time of its initial arrival is calculated to generate a time-domain correspondence parameter, which serves as the correspondence.

4. The method according to claim 1, characterized in that, The rules for performing a short-time Fourier transform on the seismic signal are as follows: The seismic signal is divided into multiple consecutive frames, the length of each frame is determined according to a preset time step, and some data overlap is maintained between adjacent frames; The short-time Fourier transform function is called one by one for each frame of data to generate the local spectrum of the corresponding frame; The local spectra are arranged and combined in chronological order to form a complete time-spectrum result.

5. The method according to claim 4, characterized in that, By combining an optimized Hamming window to track local frequency dynamics, the frequency distribution characteristics corresponding to parasitic oscillations are obtained, including: A Hamming window function is applied during the generation of each local spectrum, and the window length parameter is set based on an integer multiple of the detector's natural frequency. Amplitude envelope analysis is performed on each local spectrum after applying the Hamming window function to extract the dominant frequency component of each local spectrum; The dominant frequency components are arranged in chronological order to form a dynamic frequency sequence, and frequency abrupt change points and stable intervals are located in the dynamic frequency sequence as frequency distribution characteristics to characterize the triggering position and duration of parasitic oscillations.

6. The method according to claim 4, characterized in that, Energy changes are monitored within the octave band of the detector's natural frequency to extract energy anomaly features corresponding to parasitic oscillations, including: Extract a preset octave band spectrum data covering integer multiples of the detector's natural frequency from the time-spectrum results; Power spectral density calculations are performed on the octave band spectrum data to obtain the energy distribution curve that varies with time; On the energy distribution curve, identify the energy surge point and the energy decay interval, use the energy surge point as the trigger mark of parasitic oscillation, and use the energy decay interval as the continuous feature of parasitic oscillation. The trigger identifier is combined with the persistence feature to form an energy anomaly feature.

7. The method according to claim 6, characterized in that, Perform power spectral density calculations on the octave band spectrum data to obtain time-varying energy distribution curves, including: The octave band spectrum data is segmented and organized in chronological order to generate continuous octave band time-series spectra. The power spectral density calculation function is called on the time-series spectrum of the octave band to calculate the square amplitude of the frequency components in each time period and perform normalization to obtain the energy value of the corresponding time period. The energy values ​​for each time period are arranged sequentially to form an energy distribution curve covering the entire analysis period.

8. The method according to claim 1, characterized in that, Based on the aforementioned correspondence, frequency distribution characteristics, and energy anomaly characteristics, the seismic signal is jointly discriminated, and the identification result of the detector parasitic oscillation is output, including: The correspondence parameters, the frequency distribution features, and the energy anomaly features are input into the joint discrimination model to establish a one-to-one correspondence between the three types of features. In the joint discrimination model, feature matching operation is performed to determine whether the correspondence parameters and frequency distribution features are consistent in the time dimension, and the change trend of the energy anomaly features in the corresponding time period is verified simultaneously. When the three types of features simultaneously meet the preset consistency condition in the discrimination model, a parasitic oscillation identification identifier is output, and the parasitic oscillation identification identifier is associated with the corresponding time period; The parasitic oscillation identification identifier is combined with the associated time period to form the detector parasitic oscillation identification result.

9. The method according to claim 8, characterized in that, In the joint discrimination model, feature matching operations are performed to determine whether the correspondence parameters and frequency distribution features are consistent in the time dimension, and the changing trend of the energy anomaly features within the corresponding time period is verified simultaneously, including: The corresponding relationship parameters are mapped to a time difference sequence, and the dominant frequency dynamic sequence in the frequency distribution characteristics is synchronized to the same time coordinate. The intersection interval of the corresponding time difference sequence and the dominant frequency dynamic sequence is calculated under the time coordinate to determine the consistency between the two in the time dimension. Within the intersection interval, extract the burst points and decay intervals of energy anomaly features, and compare them with the abrupt change points and stable intervals of the dominant frequency dynamic sequence to determine whether the energy change trend corresponds to the frequency change trend. When the time difference sequence, the dominant frequency dynamic sequence, and the energy anomaly feature all meet the preset consistency condition within the intersection interval, the result of the feature matching is output as a judgment result.

10. The method according to claim 9, characterized in that, The method further includes: Acquire the reference detector signal deployed adjacent to the detector, and extract the first arrival time, frequency distribution characteristics and energy anomaly characteristics of the reference signal under the same processing flow; The characteristic parameters of the reference signal are compared with the characteristic parameters of the target detector item by item. When the reference signal does not show parasitic oscillation characteristics but the target detector does show parasitic oscillation characteristics, it is determined that the target detector has parasitic oscillation interference. The comparison results are verified for consistency with the judgment results of feature matching, and the validity of the parasitic oscillation identification conclusion is confirmed when they are consistent.

11. A detector parasitic oscillation identification system, characterized in that, The system includes: The acquisition unit is used to determine the first arrival time of the geophone by calling the first arrival picking algorithm based on the seismic signal received by the geophone, and to establish the correspondence between the first arrival time and the time of occurrence of the interference signal. The processing unit is used to perform a short-time Fourier transform on the seismic signal and combine it with an optimized Hamming window to track local frequency dynamics and obtain the frequency distribution characteristics corresponding to the parasitic oscillations. The monitoring unit is used to monitor energy changes within the octave band of the detector's natural frequency and extract energy anomaly features corresponding to parasitic oscillations. The output unit is used to jointly discriminate the seismic signal based on the correspondence, the frequency distribution characteristics, and the energy anomaly characteristics, and output the identification result of the detector parasitic oscillation.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the detector parasitic oscillation identification method according to any one of claims 1-10.

13. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the detector parasitic oscillation identification method according to any one of claims 1-10.

14. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the detector parasitic oscillation identification method according to any one of claims 1-10.

Images