Marine formation activity real-time monitoring method

Abstract

The invention relates to the technical field of ocean formation activity monitoring, in particular to an ocean formation activity real-time monitoring method, which mainly comprises the following steps of: connecting a communication optical cable existing at the ocean bottom with a land demodulator to form a distributed optical fiber sound wave sensing monitoring system, and carrying out time-space uninterrupted continuous observation on ocean formation activity signals; a deep learning algorithm is trained, and efficient and accurate pickup of the marine stratum activity DAS monitoring signals is achieved; a DAS positioning algorithm is developed, and the space accurate position of the underwater vibration signal is accurately obtained; and in combination with a time-frequency analysis technology, a time-space-frequency-strong association relationship of the marine formation activity is constructed, and an evolution law of the marine formation activity is explored. According to the method, the reliability and decision-making capability of marine geological disaster risk assessment are improved, and important support is provided for sustainable development and utilization of future marine resources.

Description

Technical Field

[0001] The present invention relates to the technical field of marine stratum activity monitoring, and in particular to a method for real-time monitoring of marine stratum activity. Background Art

[0002] Monitoring marine strata activities and ensuring the healthy development of economy and personnel is one of the important tasks facing China in the 21st century. However, it is difficult for traditional marine strata activity monitoring methods to obtain the status of seabed strata activities in real time: reflection seismic, seafloor seismometer, well logging, gravity inversion and other means can only obtain the structure of marine strata at different spatial scales, but cannot monitor the strata activities in real time. The distributed fiber acoustic sensing (DAS) monitoring technology developed in recent years provides a new idea for monitoring the activities of the bottom of the ocean. DAS technology is a new vibration signal detection method developed in recent years. It has the advantages of low comprehensive cost, good real-time transmission, continuous spatial acquisition signal, strong environmental adaptability, wide spectrum, high sensitivity, etc. It is suitable for various extreme environments such as deep sea, deep earth and deep space. Based on DAS technology, the signal acquisition, detection and positioning of marine strata activities can timely understand the activities of marine strata and prevent the impact of submarine natural disasters on the economy and personnel health of coastal cities. Summary of the invention

[0003] In view of at least one shortcoming in the prior art, the present invention provides a method for real-time monitoring of marine stratum activity, aiming to solve the problem of low timeliness of real-time monitoring and positioning of marine stratum activity.

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] A method for real-time monitoring of marine stratum activity comprises the following steps:

[0006] Connect the communication optical cable at the bottom of the ocean and the land demodulator to form a distributed optical fiber acoustic wave sensing ocean stratum activity monitoring system;

[0007] A vibration signal acquisition unit is constructed in the distributed optical fiber acoustic wave sensing ocean stratum activity monitoring system, and the vibration signal acquisition unit trains a pre-constructed neural network model based on public data; the trained neural network model is used to perform spatiotemporal continuous observation of vibration signals generated by ocean bottom stratum activity to obtain vibration signal detection data, wherein the observation data includes vibration signals generated by ocean bottom stratum activity;

[0008] A vibration signal positioning unit is constructed in the distributed optical fiber acoustic wave sensing marine stratum activity monitoring system, and the vibration signal positioning unit obtains the spatiotemporal distribution data of stratum activity based on the observation data and the pre-constructed signal positioning algorithm, and the spatiotemporal distribution data of stratum activity includes the time, spatial position and corresponding probability of stratum activity;

[0009] A time-frequency characteristic analysis unit is constructed in the distributed optical fiber acoustic wave sensing marine stratum activity monitoring system. The time-frequency characteristic analysis unit performs time-frequency analysis based on the observation data and stratum activity data to obtain the temporal and spatial evolution law of the vibration signal.

[0010] The above-mentioned real-time monitoring method for marine strata activity further includes training the pre-built neural network model, specifically including:

[0011] Acquire a training data set, wherein the training data set includes vibration signals from multiple public data sources, and preprocess the training data set to obtain a training database;

[0012] Build a neural network model based on PhaseNet-DAS, and train the neural network model using the training database;

[0013] The trained neural network model is used together with a traditional signal detection algorithm to perform a water pool signal test to obtain a comparison result. Based on the comparison result, the network structure of the neural network model is iteratively optimized and the parameters of the neural network model are updated.

[0014] The above-mentioned real-time monitoring method for marine stratum activity further constructs a signal positioning algorithm, which specifically includes:

[0015] The Bayesian probability inversion theory is used to construct the objective function, and then the DAS monitoring vibration event location algorithm is constructed;

[0016] The most likely position of the predicted vibration event is taken as the initial position, and the DAS monitoring vibration event location algorithm is inverted and solved through the Markov chain Monte Carlo algorithm to obtain the precise location of the vibration event.

[0017] The above-mentioned real-time monitoring method for marine stratum activity further uses the Bayesian probability inversion theory to construct an objective function, which specifically includes:

[0018] The Bayesian formula is expressed as:

[0019]

[0020] Wherein, d is the observed vibration signal arrival data, m is the vibration signal position parameter, p(m) represents the prior probability distribution of the vibration signal spatial position, p(d|m) represents the conditional probability of the observed vibration signal arrival data d obtained when the vibration event spatial position m is given, p(d) is the probability of obtaining the observed data d, which is set as a normalization constant, and p(m|d) is the posterior probability, which is the possible position of the vibration event spatial position m under the condition of the observed vibration signal arrival data d;

[0021] Since the spatial position of the vibration event and the probability of various situations in the observed data are the same, the prior probability p(m) and the observed data probability p(d) are set as constants;

[0022] Then the maximum a posteriori probability p max (m|d) is only related to the value of the likelihood function p(d|m):

[0023] p max (m|d)∝p(d|m),(2)

[0024] Assuming that the time-pickup error conforms to the Gaussian distribution, the likelihood function is expressed in the form of Gaussian probability:

[0025]

[0026] In the formula, is the real arrival time data of the vibration signal, d is the arrival time data calculated according to the vibration event position, σ is the standard deviation, represents the travel time mismatch function.

[0027] The real-time monitoring method for marine strata activity as described above further constructs a mismatch function, which specifically includes:

[0028] Considering the spatial continuity of the seismic records of the distributed optical fiber acoustic wave sensing marine stratum activity monitoring system, its mismatch function Ψ1 is expressed as:

[0029]

[0030] nr represents the number of detection channels, ns1 represents the number of vibration events; t = O + T is the recorded arrival time, O is the start time of the event, and T is the travel time from the source point to the detection channel; and denote the observed travel times of events i and j to detection channel r respectively; and indicates the calculated P-wave and S-wave travel times from events i and j to detection channel r; obs and cal represent the observed and calculated values;

[0031] The mismatch function Ψ1 uses the arc shape of the arrival time connection curve of the distributed optical fiber acoustic wave sensing marine stratum activity monitoring system to constrain the absolute spatial position of the microseismic event. The spatial likelihood function of the vibration event is expressed as

[0032]

[0033] The spatial position corresponding to the maximum likelihood function is the most likely position of the shock event.

[0034] The real-time monitoring method for marine stratum activity as described above further uses the most likely position of the predicted vibration event as the initial position, and accurately locates the vibration event through the Markov Chain Monte Carlo algorithm:

[0035] ① Initialization: Input the initial model m containing the spatial position of noise and vibration signals 0，0 =[σ0,x0,y0,z0], calculate the posterior probability size p(d|m 0，0 );

[0036] ② Generate a Markov Chain, the outer loop is performed N times, each time only the arrival error σ changes; the inner loop is performed M times, each time only the earthquake location parameters are updated;

[0037] Assume that the i-th arrival error of the outer loop is σ i At this time, update the j-th position parameter [x j ,y j ,z j ] Steps:

[0038] [1] In the previous step, the model parameter m i,j-1 =[σ i ,x j-1 ,y j-1 ,z j-1 ], random perturbations generate new models m i,j =[σ i ,x j ,y j ,z j ];

[0039] [2] Calculation model m i,j The posterior probability p(d|m i,j );

[0040] [3] Calculate the reception probability

[0041] [4] Compare the received probability α with the random number u between [0,1]; if α≥u, receive the new model parameters; if α<u, then m i,j =m i,j-1 ;

[0042] [5] Return to [1] for the next iteration;

[0043] ③ After the internal and external cycles are completed, the final posterior probability distribution is generated by the subsequent stable part statistics, and its maximum posterior solution is the most likely location of the vibration event.

[0044] The real-time monitoring method for marine stratum activity as described above further includes performing time-frequency analysis based on the observation data and stratum activity data to obtain the evolution law of the vibration signal, which specifically includes:

[0045] The observation data and formation activity data are subjected to time-frequency analysis by wavelet transform, Hilbert-Huang transform or generalized S transform.

[0046] The wavelet transform formula is:

[0047]

[0048] x(t) is the original signal, ψ(t) is the mother wavelet function, a and b are the scaling and translation amounts respectively;

[0049] The Hilbert-Huang transform formula is:

[0050]

[0051] x(t) is the original signal;

[0052] The generalized S transform formula is:

[0053]

[0054] x(t) is the original signal, λ and p are the control parameters of the Gaussian window function;

[0055] The vibration signal evolution law includes the evolution law of the occurrence time, spatial position, signal time-frequency parameters and waveform intensity of the formation activity.

[0056] The above-mentioned real-time monitoring method for marine stratum activity further connects the communication optical cable at the bottom of the ocean and the land demodulator to form a distributed optical fiber acoustic wave sensing marine stratum activity monitoring system, which specifically includes:

[0057] Select the communication optical cables suitable for laying on the bottom of the ocean, accurately measure their lengths, and determine their spatial locations;

[0058] Select a suitable location on land to install the fiber optic vibration signal demodulator, ensure that it is placed in a waterproof and moisture-proof equipment room, and provide continuous 24-hour power supply;

[0059] Connect the demodulator on land to the submarine communication optical cable to monitor the vibration signals of different parts of the cable in real time.

[0060] The above-mentioned real-time monitoring method for marine stratum activity further includes performing spatiotemporal continuous observation of vibration signals generated by marine bottom stratum activity, specifically comprising:

[0061] The spatial sampling interval is set to 100 meters and the temporal sampling rate exceeds 500 Hz, ensuring that the vibration signals around the communication optical cable at the bottom of the ocean are measured continuously every day.

[0062] Compared with the prior art, the present invention has the following beneficial effects: the present invention connects the submarine communication optical cable through the land demodulator, constructs a distributed fiber acoustic sensing (DAS) monitoring system, and realizes the spatiotemporal continuous observation of the activity signal of the ocean bottom strata; by training the deep learning algorithm, the vibration signal in the DAS monitoring of the ocean strata is efficiently and accurately picked up; the DAS positioning algorithm of the ocean vibration signal is developed to accurately determine the spatial position of the submarine vibration signal; combined with the time-frequency analysis technology, the spatiotemporal-frequency-intensity correlation model between the vibration signal of the ocean strata and the stratum activity is established, revealing the evolution law of the ocean stratum activity. The advantage of this method is that it can record the continuous vibration signal of the ocean strata with high spatiotemporal resolution, and can accurately locate the spatial position of the vibration signal, and deeply explore the relationship between the DAS monitoring vibration signal and the intensity of the stratum activity. The results of this project are expected to improve the accuracy and decision-making efficiency of marine geological disaster risk assessment, and provide key technical support for the sustainable development and utilization of the ocean bottom. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0064] Figure 1 The figure is a flow chart of the method for real-time monitoring of marine stratum activity of the present invention.

[0065] Figure 2 This is a schematic diagram of a DAS real-time monitoring signal of marine stratum activity according to the present invention.

[0066] Figure 3 Schematic diagram of using deep learning algorithm to efficiently pick up DAS vibration signals.

[0067] Figure 4Flow chart for locating the location of seabed stratum activity based on DAS monitoring signals. DETAILED DESCRIPTION

[0068] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0069] Example:

[0070] The terms "including" and "having" and any variations thereof in the embodiments of the present invention are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products or apparatus.

[0071] Marine strata activity has a significant impact on the economic development of coastal areas and the safety of life and property of urban residents. Traditional strata activity monitoring technology makes it difficult to ensure real-time acquisition of accurate marine strata activity data. Even with seismic stations widely distributed on land, it is difficult to accurately capture the full wavefield information of underwater vibration signals. Traditional methods lack timeliness in collecting and processing marine strata activity data and often face the problem of insufficient spatial resolution. Therefore, developing a technology that can monitor marine strata vibration signals in real time is of great significance for preventing marine geological disasters.

[0072] The present invention utilizes a land demodulator connected to a submarine communication optical cable to construct a distributed fiber acoustic sensing (DAS) monitoring system, which realizes the spatiotemporal continuous observation of the ocean bottom stratum activity signals; by training a deep learning algorithm, the efficiency and accuracy of picking up the ocean stratum DAS monitoring vibration signal are improved; a DAS positioning algorithm for ocean vibration signals is developed to accurately determine the spatial position of the seabed vibration signal; combined with time-frequency analysis technology, a time-space-frequency-strength correlation model between the ocean stratum vibration signal and the stratum activity is established to reveal the evolution law of the ocean stratum activity.

[0073] See also Figures 1 to 4 The embodiment of the present invention provides a method for real-time monitoring of marine stratum activity, which may specifically include the following steps:

[0074] Step 101: Connect the communication optical cable at the bottom of the ocean and the land demodulator to form a distributed optical fiber acoustic wave sensing ocean stratum activity monitoring system.

[0075] In this step, when implementing it, select a communication optical cable suitable for laying on the bottom of the ocean, accurately measure its length, and determine its spatial position. Select a suitable location on land to install the optical fiber vibration signal demodulator, ensure that it is placed in a machine room with waterproof and moisture-proof functions, and provide continuous 24-hour power supply. Connect the demodulator on land to the submarine communication optical cable to monitor the vibration signals of different parts of the optical cable in real time. The spatial sampling interval is set to 100 meters, and the time sampling rate exceeds 500Hz. Ensure that the vibration signals around the submarine optical cable are measured continuously every day.

[0076] When it comes to specific implementation, Figure 2 This is a schematic diagram of the distributed optical fiber sensor (DAS) monitoring of marine stratum activity signals. It mainly consists of the following parts:

[0077] Fiber optic demodulators are laid on land near the coastline and connected to the communication optical cables at the bottom of the ocean. For both land optical cables and submarine cables, it is necessary to ensure close coupling contact with the stratum and to ensure that the optical fiber does not break or have abnormal conditions in the transmission path. The optical cable signal demodulator is placed in a land machine room or other safe location to ensure that the demodulator has a continuous 24-hour stable power supply. The demodulator is connected to the submarine cable in series. The signal demodulator is used to analyze the submarine vibration signal detected by the submarine cable. The demodulator sets the interval of optical cable space monitoring to 100 meters and the time sampling rate to exceed 500Hz, so as to realize continuous monitoring of submarine stratum vibration signals and ensure uninterrupted 24 hours a day.

[0078] Step 102: construct a vibration signal acquisition unit in the distributed optical fiber acoustic wave sensing ocean stratum activity monitoring system, and train the pre-constructed neural network model based on public data; use the trained neural network model to perform spatiotemporal continuous detection of vibration signals generated by ocean bottom stratum activity to obtain vibration signal detection data, wherein the observation data includes vibration signals generated by ocean bottom stratum activity.

[0079] In this step, the pre-constructed neural network model is trained, specifically including: obtaining a training data set, the training data set includes vibration signals from multiple public data sources, preprocessing the training data set to obtain a training database; constructing a neural network model based on PhaseNet-DAS, and training the neural network model using the training database; using the trained neural network model and the traditional neural network model to perform a water pool signal test to obtain comparison data, and based on the comparison data, iteratively optimizing the network structure of the neural network model and updating the parameters of the neural network model.

[0080] In the specific implementation, ① build a training database. Collect internationally disclosed marine and land distributed acoustic sensing (DAS) seismic signals, as well as theoretical DAS seismic signals generated based on forward simulation. Improve the diversity and generalization ability of the training data set by performing preprocessing steps such as filtering, smoothing, normalization, and adding noise. ② Train the neural network model. Using the PhaseNet-DAS deep learning architecture, select appropriate neural network parameters and number of layers to train a neural network model suitable for DAS technology to monitor marine stratum vibration activities. ③ Conduct a water pool signal test. Apply the trained neural network model to underwater DAS monitoring data and compare it with traditional detection technology to further optimize the network structure. By quantitatively analyzing the prediction accuracy and efficiency of the model under different vibration positions, different magnitudes, and different signal-to-noise ratios, determine the scope of application and applicable conditions of the algorithm.

[0081] When it comes to specific implementation, Figure 3 In order to efficiently pick up DAS vibration signals using deep learning algorithms, the basic principles are:

[0082] ① It is necessary to provide a variety of training samples. Combine a variety of data to form training samples, including: public ocean and land DAS data; DAS data obtained by numerical simulation methods; increase the richness and generalization of training data through filtering, smoothing, normalization, adding noise and other operations. ② It is necessary to train to obtain a suitable neural network structure. Based on the U-shaped neural network structure of the PhaseNet-DAS two-dimensional convolution layer, the network model parameters (such as the number of layers, the number of neurons, the filter size, the activation function, and the learning rate) are tested and optimized by adding training samples. During the optimization process, the embodiment will use the Gaussian mixture model associator GaMMA to filter erroneous phase information, and increase the sample label types by means of superposition, flipping, stretching, and adding noise. Through multiple cycles of iteration, the training obtains a suitable neural network structure. ③ Conduct a water pool experiment. After the necessary preprocessing of the original data (filtering, smoothing, and normalization), the trained deep learning algorithm is applied to the DAS signal of the actual water pool experiment that has not been tested. The detected DAS vibration signal is compared with the seismograph result, and the unidentified DAS vibration signal is separated to retrain the neural network structure. At the same time, the implementation example will change the signal-to-noise ratio of the actual DAS monitoring seabed vibration signal, detect vibration events based on the trained neural network, analyze the impact of different noises on the prediction results, and obtain the effective range and applicability conditions of the deep learning algorithm.

[0083] Step 103: A vibration signal positioning unit is constructed in the distributed optical fiber acoustic wave sensing marine stratum activity monitoring system. The vibration signal positioning unit obtains the spatiotemporal distribution data of stratum activity based on the observation data and a pre-constructed signal positioning algorithm. The spatiotemporal distribution data of stratum activity includes the time, location and corresponding probability of stratum activity.

[0084] In this step, the Bayesian probability inversion theory is used to construct the objective function, and then the DAS monitoring vibration event positioning algorithm is constructed; the most likely position of the predicted vibration event is taken as the initial position, and the Markov chain Monte Carlo algorithm is used to invert and solve the DAS monitoring vibration event positioning algorithm to obtain the precise positioning of the vibration event.

[0085] In specific implementation, by constructing an accurate objective function and applying the Markov Chain Monte Carlo (MCMC) algorithm, the potential location of stratum activity and its probability distribution can be inferred, thereby quantitatively evaluating the possibility of stratum activity with different spatial distributions.

[0086] Figure 4 This is a flow chart for locating the position of submarine stratum activity based on DAS monitoring signals. The basic principle is:

[0087] (1) Constructing the objective function. DAS (distributed acoustic sensing) can collect spatially continuous vibration wave fields, but its signal-to-noise ratio is low, and the pick-up delay accuracy of different channels varies. In order to avoid the interference of different noises on the positioning results, this embodiment uses Bayesian probability inversion theory to construct the objective function.

[0088] The Bayesian formula can be expressed as:

[0089]

[0090] In the formula, d is the observed vibration signal arrival data, and m is the vibration signal position parameter. p(m) represents the prior probability distribution of the vibration signal spatial position, and p(d|m) represents the conditional probability of the observed data d obtained when the vibration event position m is given, which is called the likelihood function. p(d) is the probability of obtaining the observed data d, which can be regarded as a normalization constant. p(m|d) is the posterior probability, which is the possible position of the vibration event spatial position m given the observed data d.

[0091] Since the spatial location of the vibration event and the probability of each occurrence in the observed data are the same, the prior probability p(m) and the observed data probability p(d) are often calculated as constants. max (m|d) is only related to the value of the likelihood function p(d|m):

[0092] p max(m|d)∝p(d|m),(2)

[0093] When the picking error generally conforms to the Gaussian distribution, the likelihood function can be expressed in the form of Gaussian probability:

[0094]

[0095] In the embodiment is the real arrival time data of the vibration signal, d is the arrival time data calculated according to the vibration event position, and σ is the standard deviation. Different standard deviations are related to the signal-to-noise ratio and represent different picking accuracies. represents the travel time mismatch function.

[0096] Considering the spatial continuity of DAS seismic records, its mismatch function Ψ1 can be expressed as:

[0097]

[0098] nr represents the number of detection channels, ns1 represents the number of vibration events. t = O + T is the recorded arrival time, O is the start time of the event, and T is the travel time from the source point to the detection channel. and They represent the observed travel times of events i and j to detection channel r respectively. and Indicates the calculated P-wave and S-wave travel times from events i and j to detector channel r. obs and cal represent the observed and calculated values.

[0099] Ψ1 uses the arc shape of the DAS arrival time curve to constrain the absolute spatial position of the microseismic event. The spatial likelihood function of the vibration event can be expressed as

[0100]

[0101] The spatial position corresponding to the maximum likelihood function is the most likely position of the shock event.

[0102] (2) Selecting an inversion algorithm. In this embodiment, a grid search algorithm is first used to construct an initial vibration signal source spatial location dictionary model based on a broad search grid, and the possible geographical location of the vibration event is predicted by comparative analysis with the actual data. The predicted geographical location is used as the starting point of the inversion analysis, and then the vibration event is accurately located through the inversion process.

[0103] In order to evaluate the impact of noise error on positioning accuracy, this study case also simulated the positioning results under different noise standard deviations σ. In addition, this embodiment uses the classic MH MCMC (Metropolis-Hastings MarkovChainMonte Carlo) algorithm to generate a posterior probability model to invert and infer the exact location of the microseismic event. The Markov Chain in the inversion step can be expressed as:

[0104] ① Initialization: Input the initial model m containing the spatial position of noise and vibration signals 0，0 =[σ0,x0,y0,z0], calculate the posterior probability size p(d|m 0，0 );

[0105] ② Generate a Markov Chain, the outer loop is performed N times, each time only the arrival error σ changes; the inner loop is performed M times, each time only the earthquake position parameters are updated.

[0106] Assume that the i-th arrival error of the outer loop is σ i At this time, update the j-th position parameter [x j ,y j ,z j ] Steps:

[0107] [1] In the previous step, the model parameter m i,j-1 =[σ i ,x j-1 ,y j-1 ,z j-1 ], random perturbations generate new models m i,j =[σ i ,x j ,y j ,z j ];

[0108] [2] Calculation model m i,j The posterior probability p(d|m i,j );

[0109] [3] Calculate the reception probability

[0110] [4] Compare the received probability α with the random number u between [0,1]. If α≥u, receive the new model parameters; if α<u, then m i,j =m i,j-1 ;

[0111] [5] Return to [1] for the next iteration;

[0112] ③ After the internal and external cycles are completed, the initial “warm-up” part of the Markov Chain is discarded, and the final posterior probability distribution is generated by the statistics of the subsequent stable part. The maximum posterior solution is the most likely location of the microseismic event.

[0113] The idea of ​​Bayesian posterior probability evaluation will help to quantitatively evaluate the impact of vibration signal arrival-time picking errors on positioning results and infer the accurate spatial position of vibration events.

[0114] Step 104: constructing a time-frequency characteristic analysis unit in the distributed optical fiber acoustic wave sensing marine stratum activity monitoring system, and performing time-frequency characteristic analysis based on the observation data and stratum activity data to obtain the spatiotemporal evolution law of the vibration signal.

[0115] In this step, the frequency and spatial density of marine strata activity can be determined by analyzing the temporal and spatial distribution characteristics of ocean bottom vibration events. In addition, this embodiment will also use the noise and vibration waveform data obtained by distributed acoustic sensing (DAS) to perform fk transform, wavelet transform, Hilbert-Huang transform, generalized S transform and other methods to extract the time-frequency characteristics of the data and analyze the change patterns of these time-frequency parameters over time during the strata activity process. Among them,

[0116] The wavelet transform formula is:

[0117]

[0118] x(t) is the original signal, ψ(t) is the mother wavelet function, a and b are the scaling and translation respectively.

[0119] The Hilbert-Huang transform is to perform Hilbert transform on a single frequency function based on the empirical mode decomposition of the signal. The formula can be expressed as:

[0120]

[0121] x(t) is the original signal.

[0122] The generalized S transform uses a Gaussian window function to perform time-frequency analysis on the original signal, and its formula can be expressed as:

[0123]

[0124] x(t) is the original signal, λ and p are the control parameters of the Gaussian window function.

[0125] By analyzing the time and location of the detected vibration events, statistical methods can be used to determine the focal depth, direction, occurrence time, duration, sequence interval, frequency distribution and energy of the vibration signal. This helps to build a time-space-frequency-energy relationship model to identify the evolution of stratum activity and quantitatively evaluate the activity and evolution characteristics of the submarine strata.

[0126] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0127] The above embodiments are only for illustrating the technical concept and features of the present invention, and their purpose is to enable ordinary technicians in the field to understand the content of the present invention and implement it accordingly, and they cannot be used to limit the protection scope of the present invention. Any equivalent changes or modifications made based on the essence of the content of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for real-time monitoring of marine stratum activity, characterized in that: Includes steps: Connect the communication optical cable at the bottom of the ocean and the demodulator on land to form a distributed optical fiber acoustic wave sensing ocean stratum activity monitoring system; A vibration signal acquisition unit is constructed in the distributed optical fiber acoustic wave sensing ocean stratum activity monitoring system, and the vibration signal acquisition unit trains a pre-constructed neural network model based on public data; the trained neural network model is used to perform spatiotemporal continuous detection of vibration signals generated by ocean bottom stratum activity to obtain effective abnormal vibration observation signals, wherein the observation data includes vibration signals generated by ocean bottom stratum activity; A vibration signal positioning unit is constructed according to the distributed optical fiber acoustic wave sensing marine stratum activity monitoring system. The vibration signal positioning unit obtains the spatiotemporal distribution data of stratum activity according to the observation data and the pre-constructed signal positioning algorithm. The spatiotemporal distribution data of stratum activity includes the time, location and corresponding probability of stratum activity. A time-frequency characteristic analysis unit is constructed in the distributed optical fiber acoustic wave sensing marine stratum activity monitoring system. The time-frequency characteristic analysis unit performs time-frequency analysis based on the observation data and stratum activity data to obtain the temporal and spatial evolution law of the vibration signal.

2. The method for real-time monitoring of marine strata activity according to claim 1, characterized in that: Training a pre-built neural network model includes: Acquire a training data set, wherein the training data set includes vibration signals from multiple public data sources, and preprocess the training data set to obtain a training database; Build a neural network model based on PhaseNet-DAS, and train the neural network model using the training database; The trained neural network model is used together with a traditional signal detection algorithm to perform a water pool signal test to obtain a comparison result. Based on the comparison result, the network structure of the neural network model is iteratively optimized and the parameters of the neural network model are updated.

3. The method for real-time monitoring of marine strata activity according to claim 2, characterized in that: Pre-built signal location algorithms, including: The Bayesian probability inversion theory is used to construct the objective function, and then the DAS monitoring vibration event location algorithm is constructed; The most likely position of the predicted vibration event is taken as the initial position, and the DAS monitoring vibration event location algorithm is inverted and solved through the Markov chain Monte Carlo algorithm to obtain the precise location of the vibration event.

4. The method for real-time monitoring of marine strata activity according to claim 3, characterized in that: The objective function is constructed using Bayesian probability inversion theory, including: The Bayesian formula is expressed as: Wherein, d is the observed vibration signal arrival data, m is the vibration signal position parameter, p(m) represents the prior probability distribution of the vibration signal spatial position, p(d|m) represents the conditional probability of the observed vibration signal arrival data d obtained when the vibration event spatial position m is given, p(d) is the probability of obtaining the observed data d, which is set as a normalization constant, and p(m|d) is the posterior probability, which is the possible position of the vibration event spatial position m under the condition of the observed vibration signal arrival data d; Since the spatial position of the vibration event and the probability of various situations in the observed data are the same, the prior probability p(m) and the observed data probability p(d) are set as constants; Then the maximum a posteriori probability p max (m|d) is only related to the value of the likelihood function p(d|m): p max (m|d)∝p(d|m),(2) Assuming that the time-pickup error conforms to the Gaussian distribution, the likelihood function is expressed in the form of Gaussian probability: In the formula, is the real arrival time data of the vibration signal, d is the arrival time data calculated according to the vibration event position, σ is the standard deviation, represents the travel time mismatch function.

5. The method for real-time monitoring of marine strata activity according to claim 4, characterized in that: Construct a mismatch function, including: Considering the spatial continuity of the seismic records of the distributed optical fiber acoustic sensing marine stratum activity monitoring system, its mismatch function Ψ1 is expressed as: nr represents the number of detection channels, ns1 represents the number of vibration events; t = O + T is the recorded arrival time, O is the start time of the event, and T is the travel time from the source point to the detection channel; and denote the observed travel times of events i and j to detection channel r respectively; and indicates the calculated P-wave and S-wave travel times from events i and j to detection channel r; obs and cal represent the observed and calculated values; The mismatch function Ψ1 uses the arc shape of the arrival time connection curve of the distributed optical fiber acoustic wave sensing marine stratum activity monitoring system to constrain the absolute spatial position of the microseismic event. The spatial likelihood function of the vibration event is expressed as The spatial position corresponding to the maximum likelihood function is the most likely position of the shock event.

6. The method for real-time monitoring of marine strata activity according to claim 4, characterized in that: The most likely position of the predicted vibration event is used as the initial position, and the vibration event is accurately located using the Markov Chain Monte Carlo algorithm: ① Initialization: Input the initial model m containing the spatial position of noise and vibration signals 0，0 =[σ0,x0,y0,z0], calculate the posterior probability size p(d|m 0，0 ); ② Generate a Markov Chain, the outer loop is performed N times, each time only the arrival error σ changes; the inner loop is performed M times, each time only the earthquake location parameters are updated; Assume that the i-th arrival error of the outer loop is σ i At this time, update the j-th position parameter [x j ,y j ,z j ]The steps are: [1] In the previous step, the model parameter m i,j-1 =[σ i ,x j-1 ,y j-1 ,z j-1 ], random perturbations generate new models m i,j =[σ i ,x j ,y j ,z j ]; [2] Calculation model m i,j The posterior probability p(d|m i,j ); [3] Calculate the reception probability [4] Compare the received probability α with a random number u between [0,1]; if α≥u, receive new model parameters; If α < u, then m i,j = m i,j-1 ; [5] Return to [1] for the next iteration; ③ After the internal and external cycles are completed, the final posterior probability distribution is generated by the subsequent stable part statistics, and its maximum posterior solution is the most likely location of the vibration event.

7. The method for real-time monitoring of marine strata activity according to claim 1, characterized in that: Based on the observation data and formation activity data, time-frequency analysis is performed to obtain the evolution law of the vibration signal, which specifically includes: The observation data and formation activity data are subjected to time-frequency analysis by wavelet transform, Hilbert-Huang transform or generalized S transform. The wavelet transform formula is: x(t) is the original signal, ψ(t) is the mother wavelet function, a and b are the scaling and translation amounts respectively; The Hilbert-Huang transform formula is: x(t) is the original signal; The generalized S transform formula is: x(t) is the original signal, λ and p are the control parameters of the Gaussian window function; The vibration signal evolution law includes the evolution law of the occurrence time, spatial position, signal time-frequency parameters and waveform intensity of the formation activity.

8. The method for real-time monitoring of marine strata activity according to claim 1, characterized in that: The communication optical cable at the bottom of the ocean and the demodulator on land are connected to form a distributed optical fiber acoustic wave sensing ocean stratum activity monitoring system, which specifically includes: Select the communication optical cables suitable for laying on the bottom of the ocean, accurately measure their lengths, and determine their spatial locations; Select a suitable location on land to install the fiber optic vibration signal demodulator, ensure that it is placed in a waterproof and moisture-proof equipment room, and provide continuous 24-hour power supply; Connect the demodulator on land to the submarine communication optical cable to monitor the vibration signals of different parts of the optical cable in real time.

9. The method for real-time monitoring of marine strata activity according to claim 1, characterized in that: Continuous observation of vibration signals generated by seabed strata activity in time and space, including: The spatial sampling interval is set to 100 meters and the temporal sampling rate exceeds 500 Hz, ensuring that the vibration signals around the communication optical cable at the bottom of the ocean are measured continuously every day.

Images