Geophysical detection system based on seabed crawler

Abstract

The invention provides a geophysical detection system based on a seabed crawl device, and belongs to the technical field of geophysics, the geophysical detection system comprises a survey ship, a seabed crawl device, a gravimeter, a shallow profiler, a seismograph source, a seismograph receiving array and a control chip, the survey ship is connected with the seabed crawl device through a traction rope, and the control chip is connected with the seabed crawl device through a traction rope. A dragging cable is arranged at the tail part of the seabed crawler, and related instruments are sequentially and fixedly arranged on the dragging cable; the control chip is electrically connected with a driving unit of the seabed crawler and other instruments and performs data interaction with the driving unit and the other instruments; and a geophysical detection control module is arranged in the control chip and is used for setting parameters of a seismograph source and preprocessing data acquired by the receiver, finally obtaining geophysical detection data and sending the geophysical detection data to an investigation ship by using data transmission equipment. According to the invention, the technical problem of low precision of the existing seabed geophysical exploration method is solved.

Description

Technical Field

[0001] The present invention belongs to the field of geophysical technology, and in particular, relates to a geophysical detection system based on a seafloor crawler. Background Art

[0002] In recent years, with the continuous increase in marine resource exploration and development activities, the application demand based on seabed geophysical detection technology has also increased. Comprehensive exploration of seabed geological structures and mineral resources can provide important geological basis for marine oil and gas exploitation, deep-sea mining, etc. Traditional seabed geophysical exploration mainly relies on ship platforms and uses towed detection equipment such as gravimeters, magnetometers and multi-channel seismic measurement systems to conduct comprehensive detection of the seabed. This method has the advantages of wide detection range and rich information, but it also has some problems.

[0003] First, due to the water depth and navigation conditions, ship platforms are difficult to enter some sea areas with complex terrain, such as ridges, canyons and other areas, resulting in detection blind spots. Secondly, towed detection equipment is easily affected by ocean currents, surges, etc., and it is difficult to ensure the quality and accuracy of detection data. Furthermore, traditional detection methods often require large ships and professional operators, and the operating costs are high. In addition, the complex and changeable geological structure of the seabed makes it difficult for a single detection method to fully depict the detailed information of the seabed strata.

[0004] In order to overcome the above problems, seabed geophysical detection systems based on underwater robots have emerged in recent years. Such systems usually use unmanned underwater vehicles (AUVs) or underwater crawling robots (ROVs) to carry detection equipment and directly contact the seabed for detection. They can penetrate into complex terrain areas and improve the quality and accuracy of data. At the same time, underwater robots are flexible to operate and can be programmed to operate autonomously, which greatly reduces labor costs. In addition, by integrating multiple detection methods, such as gravimeters, magnetometers, sonar depth sounding, etc., more abundant and comprehensive geological information can be obtained.

[0005] However, the existing underwater robot-based detection systems also have some defects. Ship-borne geophysical detection and offshore towed geophysical detection have high noise and low detection accuracy due to the influence of wind, waves and currents on the sea surface. In addition, the thick seawater layer will weaken or even completely cover the high-frequency information of the geophysical field, which will also reduce the detection accuracy. In offshore towed geophysical detection, geophysical equipment such as magnetometers will be disturbed by the towed body, resulting in low accuracy. Summary of the invention

[0006] In view of this, the present invention provides a geophysical exploration system based on a seabed crawler, which can solve the technical problem of low accuracy in existing seabed geophysical exploration methods.

[0007] The present invention is achieved in that:

[0008] The present invention provides a geophysical exploration system based on a seabed crawler, which comprises: a survey ship, a seabed crawler, a gravimeter, a shallow layer profiler, a seismograph source, a seismograph receiving array and a control chip, wherein the survey ship is connected to the seabed crawler through a traction rope, a towing cable is arranged at the tail of the seabed crawler, and the gravimeter, the shallow layer profiler, the seismograph source and the seismograph receiving array are fixedly arranged on the towing cable in sequence; an electronic cabin is arranged in the seabed crawler, and the data control chip is arranged in the electronic cabin, and the control chip is electrically connected with the driving unit of the seabed crawler, the gravimeter, the shallow layer profiler, the seismograph source and the seismograph receiving array to exchange data; a geophysical exploration control module is arranged in the control chip, which is used to set the parameters of the seismograph source and pre-process the data collected by the receiver to finally obtain geophysical exploration data; a data transmission device is also arranged in the electronic cabin, and the control chip is electrically connected with the data transmission device, and the data transmission device is used to send the geophysical exploration data to the survey ship. In addition, the gravimeter may also be directly fixed on the seafloor crawler.

[0009] On the basis of the above technical solution, the geophysical detection system based on the seafloor crawler of the present invention can also be improved as follows:

[0010] Among them, it also includes a self-contained geomagnetic diurnal variation station, which is carried by a crawler and deployed on the seabed, and is recovered by the crawler after completing the operation.

[0011] Furthermore, the survey vessel is equipped with a satellite navigation system, an underwater positioning system, and a power supply unit.

[0012] Furthermore, the seabed crawler is a tracked seabed crawling robot or a seabed ore collecting machine or a seabed mining vehicle.

[0013] Furthermore, the transmitting array vibration surface of the seismic source of the seismograph is in contact with the seabed sediments.

[0014] Furthermore, it also includes a positioning beacon, which is fixed on the towing cable.

[0015] Furthermore, the data transmission device sends the geophysical detection data to the survey ship via a data cable or a wireless channel.

[0016] Furthermore, the seabed crawler is also provided with a sensor module, which at least includes a depth gauge, an altimeter, a gyroscope and an accelerometer.

[0017] Furthermore, it also includes a magnetometer, an electromagnetic instrument excitation source, and an electromagnetic instrument receiving array which are fixedly arranged on the towing cable in sequence.

[0018] Among them, the gravimeter, magnetometer, electromagnetic excitation source, electromagnetic receiving array, shallow layer profiler, seismograph source, and seismograph receiving array may include zero, one, or more. Among them, the seismograph source may be a frequency-adjustable source such as a percussion drill, an air gun, or a heavy hammer striking device.

[0019] The geophysical detection control module is used to perform the following steps:

[0020] S10, controlling the seafloor crawler to crawl in a predetermined detection area, obtaining the position information of the seafloor crawler, and controlling the seismic source of the seismograph to continuously emit seismic waves and receive reflected waves through the seismograph receiving array;

[0021] S20, preprocessing the received reflected wave data, including denoising, filtering and gain control;

[0022] S30, using a sliding time window method to perform segmented analysis on the preprocessed reflected wave data, and extracting the reflected wave characteristics in each time window, including frequency, amplitude and phase characteristics;

[0023] S40, identifying abnormal points in the reflected wave characteristics according to a preset threshold, including amplitude abnormality, frequency abnormality and phase abnormality;

[0024] S50, time-aligning the identified abnormal points with the transmitted seismic wave sequence, and establishing a corresponding matrix between the abnormal points and the specific seismic waves;

[0025] S60, performing singular value decomposition on the corresponding matrix to obtain a basic matrix and a change matrix;

[0026] S70, inputting the variation matrix into a pre-trained seismograph source parameter model to obtain corresponding optimization parameters, including source intensity and emission frequency; and controlling the seismograph source to re-emit seismic waves according to the optimization parameters, and receiving new reflected wave data through the seismograph receiving array;

[0027] S80, processing the newly received reflected wave data, including dynamic correction, spectrum analysis and amplitude recovery, to improve data quality;

[0028] S90, fusing and analyzing the processed reflected wave data with the data collected by the gravimeter and the shallow layer profiler to generate comprehensive geophysical exploration data; further, the geophysical exploration data also fuses the gravity field, magnetic field, electromagnetic, and shallow profile data.

[0029] The seismic source parameter model of the seismograph adopts an adversarial neural network model. The generator is specifically a multi-layer perceptron, which contains 3 hidden layers, with 128, 64, and 32 neurons in each layer respectively. The ReLU activation function is used, and the tanh activation function is used in the output layer. The discriminator is specifically a convolutional neural network, which contains 3 convolutional layers and 2 fully connected layers. The LeakyReLU activation function is used in the convolutional layers, the ReLU activation function is used in the fully connected layers, and the sigmoid activation function is used in the output layer. The initial input of the generator is specifically a variation matrix and a 100-dimensional random noise vector. The initial input of the discriminator is specifically the initially set seismic source parameters and the corresponding reflected wave characteristics.

[0030] The loss function of the seismic source parameter model of the seismograph adopts the Wasserstein loss function.

[0031] Specifically, the generator is as follows:

[0032] 1. Input layer: Receive two inputs. One is a vector containing 100 random noise values, which are sampled from the standard normal distribution. The other is a variation matrix (assuming the dimension is m×n).

[0033] 2. Matrix processing layer: Use 1×1 convolution to reduce the dimension of the variation matrix and compress it into a k-dimensional vector (k < m×n). Concatenate the compressed k-dimensional vector with the 100-dimensional random noise vector to obtain a (100 + k)-dimensional vector.

[0034] 3. Fully connected layer: Include the first hidden layer: 256 neurons, using the LeakyReLU activation function; the second hidden layer: 128 neurons, using the LeakyReLU activation function; the third hidden layer: 64 neurons, using the LeakyReLU activation function.

[0035] 4. Output layer: Use the tanh activation function, and the output dimension is the same as the number of required seismic source parameters.

[0036] The steps for establishing the training data set of the seismic source parameter model of the seismograph are as follows:

[0037] 1. Data collection: Collect a large amount of historical seismic detection data, including seismic source parameters and corresponding reflected wave data under different seabed environments, ensuring that the data covers various seabed geological conditions and detection equipment parameters.

[0038] 2. Data preprocessing: Denoise the original data to eliminate environmental noise and equipment noise. Standardize the reflected wave data to make data of different scales comparable. Normalize the seismic source parameters to ensure that all parameters are within the range of [-1, 1].

[0039] 3. Feature extraction: Perform time-frequency analysis on the reflected wave data, extract the frequency, amplitude and phase characteristics, calculate the statistical characteristics of the reflected wave data, such as mean, variance, skewness and kurtosis, and extract the correlation indicators between the source parameters and the reflected wave characteristics.

[0040] 4. Data segmentation: The processed dataset is randomly divided into training set (70%), validation set (15%) and test set (15%).

[0041] 5. Data augmentation: Perform data augmentation on the training set, including adding random noise, small scaling and translation, etc., to generate synthetic data, and expand the data set by interpolation and extrapolation of existing data.

[0042] 6. Change matrix generation: For each set of training data, calculate the change trend of the reflection wave characteristics, generate the corresponding change matrix, and pair the change matrix with the original source parameters and reflection wave characteristics to form a complete training sample.

[0043] The training steps of the seismograph source parameter model are:

[0044] 1. Model initialization:

[0045] a. Initialize the generator and discriminator according to the provided structure.

[0046] b. Initialize the Adam optimizer for training the generator and discriminator respectively.

[0047] 2. The training cycle is described in computer pseudo code as follows:

[0048]

[0049] 3. Model fine-tuning:

[0050] Analyze loss curves and performance metrics during training.

[0051] Adjust the learning rate, batch size, or model structure as needed.

[0052] If overfitting occurs, consider using regularization techniques or early stopping.

[0053] 4. Final evaluation: Evaluate the performance of the final model on the test set and calculate key indicators such as mean square error, correlation coefficient, etc.

[0054] This training process makes full use of the information of the change matrix and stabilizes the training process through the framework of Wasserstein GAN. In this way, the model should be able to learn a more accurate strategy for generating source parameters.

[0055] Wherein, the step S10 specifically includes: controlling the seafloor crawler to crawl along the predetermined detection area, and obtaining its position and attitude information through the positioning sensor on the crawler; at the same time, controlling the seismic source of the seismograph to continuously transmit seismic waves to the seafloor with preset parameters (such as source intensity, transmission frequency, etc.), and the seismic wave signals reflected back are received by the seismograph receiving array. The position and attitude information of the crawler, plus the transmission time of the seismic wave and the reception time of the receiving array, can calculate the seafloor position corresponding to the reflected wave signal, laying the foundation for subsequent stratigraphic analysis.

[0056] Among them, the step S20 specifically includes: firstly, denoising the received reflected wave signal to eliminate the environmental noise and the noise interference of the measuring equipment itself. Common denoising methods include time domain filtering, frequency domain filtering, etc.; then filtering is performed to focus on removing some useless high-frequency or low-frequency components and improve the signal-to-noise ratio of the signal. Butterworth filter, Chebyshev filter, etc. can be used; finally, adaptive gain control is performed to amplify weak signals, suppress strong signals, and keep the overall dynamic range at an appropriate level. An adaptive algorithm based on LMSE can be used. Through the above preprocessing, the quality of the reflected wave signal can be significantly improved, laying the foundation for the subsequent time-frequency feature extraction.

[0057] Wherein, the step S30 specifically includes: using the sliding time window method to perform segmented analysis on the pre-processed reflected wave data, dividing the entire reflected wave signal into several time windows, each window containing a certain number of sampling points; for each time window, extracting its frequency, amplitude and phase characteristics, the frequency characteristics can be obtained by short-time Fourier transform or wavelet transform, the amplitude characteristics can calculate the root mean square value of the signal, and the phase characteristics can directly extract the phase angle of the signal. In this way, the characteristic description of the reflected wave in the two dimensions of time and frequency is obtained, providing a basis for the subsequent abnormal point identification.

[0058] Wherein, the step S40 specifically includes: identifying abnormal points in the reflection wave characteristics according to a preset threshold. For frequency characteristics, if the difference between the main frequency in a certain time window and the adjacent window exceeds 10%, it is determined to be a frequency anomaly; for amplitude characteristics, if the root mean square value of a certain window exceeds 2 times the average value, it is determined to be an amplitude anomaly; for phase characteristics, if the phase change rate of a certain window exceeds 30 degrees / ms, it is determined to be a phase anomaly. Through the above threshold judgment, the abnormal points in the reflection wave characteristics can be preliminarily screened out, providing important clues for subsequent geological analysis.

[0059] The step S50 specifically includes: according to the time position of the abnormal point, aligning it with the transmitted seismic wave sequence, and establishing the corresponding relationship between the abnormal point and the specific seismic wave. First, according to the seismic wave transmission time and the receiving time of the receiving array, the seabed position corresponding to each reflected wave signal is calculated, and then the abnormal point identified in step S40 is mapped to these positions to determine the seismic wave transmission sequence corresponding to them. Through this time alignment, a corresponding matrix between the abnormal point and the specific seismic wave can be established, which provides important information for the subsequent optimization of the source parameters.

[0060] The step S60 specifically includes: using the singular value decomposition (SVD) algorithm to decompose the abnormal point-seismic wave correspondence matrix to obtain three matrices U, Σ and V. Among them, U and V are regarded as basic matrices, which contain the basic characteristics of the seabed stratum structure; Σ is regarded as a variable matrix, which reflects the dynamic changes of the stratum structure. In this way, the original abnormal point-seismic wave correspondence relationship is decomposed into a basic part and a variable part, providing a more detailed input for the subsequent optimization of the source parameters.

[0061] Wherein, the step S70 specifically includes: inputting the change matrix obtained in step S60 into the pre-trained seismograph source parameter model to obtain optimized source parameters. The parameter model adopts the architecture of adversarial generative network (GAN), including two main parts: generator and discriminator. The generator is responsible for outputting optimized source parameters according to the input change matrix and random noise; the discriminator is responsible for judging whether these parameters are consistent with the real data. During the training process, the generator and the discriminator are trained adversarially with the Wasserstein loss function as the target until the convergence condition is reached.

[0062] The step S80 specifically includes: first, according to the position information of the crawler, the reflected wave data is subjected to dynamic correction to eliminate the time drift and spatial dislocation caused by the movement of the crawler; then, the corrected reflected wave data is subjected to spectrum analysis to extract richer frequency information, which is helpful to identify the characteristics of different lithological formations; at the same time, the amplitude value is also normalized to eliminate the attenuation caused by the propagation distance. Through the above processing, the quality of the reflected wave data can be further improved, laying the foundation for the subsequent geophysical data fusion analysis.

[0063] Compared with the prior art, the beneficial effects of the geophysical exploration system based on the seafloor crawler provided by the present invention are:

[0064] 1. The crawler directly contacts the seabed and can obtain more accurate position and attitude data, providing a basis for the spatial registration of detection data. Seabed geophysical detection based on seabed crawlers can avoid the attenuation of the geophysical field by the seawater layer and improve the high-frequency information of the geophysical field obtained by marine geophysical detection; the use of seabed geophysical sensor arrays towed behind the crawler can avoid the interference of the electromagnetic field of the operating platform (survey ship, towed body or crawler) on the geophysical sensor and obtain high-precision geophysical field information;

[0065] 2. The crawler has strong working continuity and can operate for a long time on the seabed, which greatly improves the continuity and integrity of data collection. It can break through terrain restrictions and go deep into complex seabed landforms such as submarine canyons and ridges, expanding the detection range.

[0066] 3. By adopting advanced signal processing and parameter optimization algorithms, the quality and reliability of detection data are further improved.

[0067] In summary, the present invention solves the technical problem of low precision in existing seabed geophysical detection methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 A schematic diagram of the composition of a geophysical detection system based on a seafloor crawler provided by the present invention;

[0069] Figure 2 A flow chart of steps executed by a geophysical exploration control module;

[0070] Figure 3 is a schematic diagram of a wired system according to Embodiment 2 of the present invention;

[0071] Figure 4 This is a system diagram of a wireless method according to the second embodiment of the present invention. DETAILED DESCRIPTION

[0072] In order to make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0073] like Figure 1As shown, it is a schematic diagram of the composition of a geophysical exploration system based on a seabed crawler provided by the present invention, comprising: a survey ship, a seabed crawler, a gravimeter, a shallow layer profiler, a seismograph source, a seismograph receiving array and a control chip, wherein the survey ship is connected to the seabed crawler by a traction rope, a towing cable is provided at the tail of the seabed crawler, and the gravimeter, the shallow layer profiler, the seismograph source and the seismograph receiving array are fixedly arranged on the towing cable in sequence; an electronic cabin is provided in the seabed crawler, and a data control chip is provided in the electronic cabin, and the control chip is electrically connected to the driving unit, gravimeter, shallow layer profiler, seismograph source and seismograph receiving array of the seabed crawler and performs data exchange; a geophysical exploration control module is provided in the control chip, which is used to set the parameters of the seismograph source and pre-process the data collected by the receiver to finally obtain geophysical exploration data; a data transmission device is also provided in the electronic cabin, and the control chip is electrically connected to the data transmission device, and the data transmission device is used to send the geophysical exploration data to the survey ship. Furthermore, it also includes a magnetometer, an electromagnetic excitation source, and an electromagnetic receiving array fixedly arranged on the towing cable in sequence. Among them, the gravimeter, magnetometer, electromagnetic excitation source, electromagnetic receiving array, shallow layer profiler, seismic source, and seismic receiving array can all include zero, one, or more. In addition, the gravimeter can also be directly fixedly arranged on the seafloor crawler.

[0074] The seismograph source may be a frequency-adjustable source such as a percussion drill, an air gun, or a heavy hammer striking device.

[0075] The geophysical detection control module is used to perform the following steps:

[0076] S10, controlling the seafloor crawler to crawl in a predetermined detection area, obtaining the position information of the seafloor crawler, and controlling the seismic source of the seismograph to continuously emit seismic waves and receive reflected waves through the seismograph receiving array;

[0077] S20, preprocessing the received reflected wave data, including denoising, filtering and gain control;

[0078] S30, using a sliding time window method to perform segmented analysis on the preprocessed reflected wave data, and extracting the reflected wave characteristics in each time window, including frequency, amplitude and phase characteristics;

[0079] S40, identifying abnormal points in the reflected wave characteristics according to a preset threshold, including amplitude abnormality, frequency abnormality and phase abnormality;

[0080] S50, time-aligning the identified abnormal points with the transmitted seismic wave sequence, and establishing a corresponding matrix between the abnormal points and the specific seismic waves;

[0081] S60, performing singular value decomposition on the corresponding matrix to obtain a basic matrix and a change matrix;

[0082] S70, inputting the variation matrix into a pre-trained seismograph source parameter model to obtain corresponding optimization parameters, including source intensity and emission frequency; and controlling the seismograph source to re-emit seismic waves according to the optimization parameters, and receiving new reflected wave data through the seismograph receiving array;

[0083] S80, processing the newly received reflected wave data, including dynamic correction, spectrum analysis and amplitude recovery, to improve data quality;

[0084] S90. The processed reflected wave data is integrated and analyzed with the data collected by the gravimeter and the shallow layer profiler to generate comprehensive geophysical detection data.

[0085] The specific implementation methods of the above steps are described in detail below:

[0086] The specific implementation of step S10 is to first control the seabed crawler to crawl along the predetermined detection area, and at the same time obtain its position and posture information through the positioning sensor on the crawler. These position and posture information provide a basis for the subsequent spatial registration of the detection data.

[0087] At the same time, the control chip is also responsible for coordinating the work of the seismograph source, allowing it to continuously emit seismic waves to the seabed with preset parameters (such as source strength, transmission frequency, etc.). These seismic waves will propagate in the seabed strata and be reflected back and received by the seismograph receiving array.

[0088] The position and attitude information of the crawler, together with the emission time of the seismic wave and the reception time of the receiving array, can calculate the seabed position corresponding to the reflected wave signal, laying the foundation for subsequent stratigraphic analysis. The purpose of this step is to obtain basic detection data of the seabed environment, including the movement trajectory of the crawler and the reflection information of the seismic wave.

[0089] The specific implementation method of step S20 is to pre-process the received reflected wave signal. First, denoising is performed to eliminate environmental noise and noise interference of the measuring device itself. Common denoising methods include time domain filtering, frequency domain filtering, etc.

[0090] Then filter processing is performed to remove some useless high-frequency or low-frequency components and improve the signal-to-noise ratio. The filtering method can use Butterworth filter, Chebyshev filter, etc.

[0091] Finally, adaptive gain control is performed to amplify weak signals and suppress strong signals, so that the overall dynamic range is maintained at an appropriate level, which is conducive to subsequent signal analysis. Gain control can use an adaptive algorithm based on LMSE (minimum mean square error).

[0092] Through the above preprocessing, the quality of the reflected wave signal can be significantly improved, laying the foundation for the subsequent time-frequency feature extraction. The purpose of this step is to perform preliminary noise elimination and energy adjustment on the original reflected wave data, creating favorable conditions for subsequent signal analysis.

[0093] The specific implementation of step S30 is to use the sliding time window method to perform segmented analysis on the pre-processed reflected wave data. First, the entire reflected wave signal is divided into several time windows, each window containing a certain number of sampling points.

[0094] For each time window, extract its frequency, amplitude and phase features. The frequency feature can be obtained by short-time Fourier transform or wavelet transform; the amplitude feature can calculate the root mean square value of the signal; the phase feature can directly extract the phase angle of the signal.

[0095] In this way, the characteristic description of the reflected wave in two dimensions of time and frequency is obtained. These characteristics can reflect the geological information of strata at different depths and provide a basis for the subsequent identification of abnormal points.

[0096] The purpose of this step is to extract the time-frequency characteristics of the reflected wave signal in detail, laying the foundation for the judgment of abnormal points and the optimization of source parameters. Through segmented analysis, the changing laws of the reflected wave in the time and frequency domains can be better captured.

[0097] The specific implementation of step S40 is to identify abnormal points in the reflected wave characteristics according to a preset threshold value. First, reasonable threshold values ​​are set for the three types of characteristics: frequency, amplitude and phase.

[0098] For frequency features, if the difference between the main frequency in a certain time window and the adjacent window exceeds 10%, it is judged as a frequency anomaly. For amplitude features, if the root mean square value of a certain window exceeds 2 times the average value, it is judged as an amplitude anomaly. For phase features, if the phase change rate of a certain window exceeds 30 degrees / ms, it is judged as a phase anomaly.

[0099] Through the above threshold judgment, the abnormal points in the reflection wave characteristics can be preliminarily screened out. These abnormal points often correspond to the discontinuity surface or lithology change of the seabed strata, providing important clues for subsequent geological analysis.

[0100] The purpose of this step is to mine geological anomaly information from the time-frequency characteristics of the reflection wave, and provide a basis for the next step of source parameter optimization. Reasonable setting of thresholds is crucial for accurate identification of anomalies.

[0101] The specific implementation method of step S50 is to align the time position of the abnormal point with the transmitted seismic wave sequence according to the time position of the abnormal point, and establish a corresponding relationship between the abnormal point and the specific seismic wave.

[0102] First, the seabed positions corresponding to each reflected wave signal are calculated based on the seismic wave emission time and the receiving time of the receiving array. Then, the abnormal points identified in step S40 are mapped to these positions to determine the seismic wave emission sequences corresponding to them.

[0103] Through this time alignment, a correspondence matrix between abnormal points and specific seismic waves can be established. This matrix describes the distribution characteristics of abnormal points in time and space, providing important information for subsequent optimization of earthquake source parameters.

[0104] The purpose of this step is to establish a corresponding relationship between the identified abnormal points and the seismic wave emission process, laying the foundation for the next step of array decomposition. This time alignment work requires full use of the crawler's positioning information and the propagation characteristics of seismic waves.

[0105] The specific implementation method of step S60 is to perform array decomposition on the abnormal point-seismic wave correspondence matrix established in step S50 to obtain a basic matrix and a change matrix.

[0106] First, the singular value decomposition (SVD) algorithm is used to decompose the abnormal point-seismic wave correspondence matrix to obtain three matrices:

[0107] A=UΣV T ;

[0108] Where A is the original matrix, U and V are orthogonal matrices, and Σ is a diagonal matrix containing the singular values ​​of matrix A.

[0109] Next, U and V are considered as basic matrices, which contain the basic characteristics of the seabed stratigraphic structure; Σ is considered as a variable matrix, which reflects the dynamic changes of the stratigraphic structure.

[0110] In this way, the original abnormal point-seismic wave correspondence is decomposed into the basic part and the variable part, providing more detailed input for the subsequent optimization of earthquake source parameters.

[0111] The purpose of this step is to extract the basic characteristics and dynamic changes of the stratigraphic structure from the abnormal point-seismic wave correspondence through the matrix decomposition method, so as to provide more valuable information for the next step of parameter optimization.

[0112] The specific implementation method of step S70 is to input the change matrix obtained in step S60 into a pre-trained seismograph source parameter model to obtain optimized source parameters.

[0113] This parameter model adopts the architecture of Generative Adversarial Network (GAN), which includes two main parts: generator and discriminator. The generator is responsible for outputting optimized source parameters based on the input change matrix and random noise; the discriminator is responsible for determining whether these parameters are consistent with the real data.

[0114] The specific structure of the generator is as follows:

[0115] 1. Input layer: receives the change matrix and 100-dimensional random noise vector

[0116] 2. Matrix processing layer:

[0117] a. Use 1x1 convolution to reduce the dimension of the change matrix and compress it into a k-dimensional vector

[0118] b. Concatenate the k-dimensional vector with the 100-dimensional noise vector to obtain a (100+k)-dimensional vector

[0119] 3. Fully connected layer:

[0120] a. First hidden layer: 256 neurons, using LeakyReLU activation function

[0121] b. Second hidden layer: 128 neurons, using LeakyReLU activation function

[0122] c. The third hidden layer: 64 neurons, using the LeakyReLU activation function

[0123] 4. Output layer: Use tanh activation function, the output dimension is the same as the number of source parameters

[0124] The discriminator adopts the structure of convolutional neural network, which includes 3 convolutional layers and 2 fully connected layers.

[0125] During the training process, the generator and the discriminator are trained adversarially with the Wasserstein loss function as the target until the convergence condition is reached. After the training is completed, the change matrix is ​​input into the generator to obtain the optimized source parameters.

[0126] The purpose of this step is to use deep learning methods to automatically optimize the parameters of the seismograph source according to the abnormal characteristics of the reflected waves, creating conditions for the next step of data fusion analysis.

[0127] The specific implementation method of step S80 is to further process the newly received reflected wave data, including dynamic correction, spectrum analysis and amplitude recovery.

[0128] First, according to the position information of the crawler, the reflected wave data is corrected to eliminate the time drift and spatial dislocation caused by the crawler movement. This step ensures the consistency of the reflected wave data in time and space.

[0129] Next, the corrected reflected wave data is subjected to spectrum analysis to extract richer frequency information. These frequency characteristics help to identify the characteristics of different lithology formations. At the same time, the amplitude value is also normalized to eliminate the attenuation caused by the propagation distance.

[0130] Through the above processing, the quality of reflection wave data can be further improved, laying the foundation for the subsequent geophysical data fusion analysis. The purpose of this step is to conduct in-depth analysis of the optimized reflection wave data and extract more abundant and reliable geological information.

[0131] The specific implementation of step S90 is to fuse and analyze the processed reflected wave data with the data collected by the gravimeter and the shallow layer profiler to generate comprehensive geophysical exploration data. Furthermore, the geophysical exploration data also fuses gravity field, magnetic field, electromagnetic, and shallow profile data.

[0132] Specifically, the principle of the present invention is:

[0133] First, the seafloor crawler is used as a mobile platform. Compared with the traditional ship towing method, it can penetrate into complex seafloor terrain areas and obtain more accurate position and attitude data. The crawler integrates a variety of detection equipment such as gravimeters, shallow layer profilers and seismic measurement systems, which can comprehensively survey seafloor geological information. At the same time, the crawler has strong working continuity and can operate for a long time on the seafloor, making up for the short operating time of underwater vehicles.

[0134] Secondly, the present invention adopts advanced algorithms in signal processing and parameter optimization. Through time-frequency domain feature extraction and abnormal point identification, the geological anomaly information contained in the reflected wave signal can be discovered, providing a basis for the subsequent optimization of the source parameters. Using the adversarial generative network (GAN) framework, through the adversarial training of the generator and the discriminator, the working parameters of the seismic source can be adaptively adjusted to make it more in line with the characteristics of the current seabed environment, thereby obtaining more accurate earthquake detection data.

[0135] Finally, the present invention uses Bayesian data fusion technology to integrate the information obtained by different detection methods, such as gravity data, acoustic profile data and seismic data, into a unified probabilistic framework for comprehensive analysis. This multi-source information fusion method can give full play to the advantages of various detection methods and obtain more comprehensive and reliable geological information.

[0136] In general, the technical solution of the present invention fully considers the complexity of the seabed environment and the requirements of the detection task. By innovatively integrating multiple detection methods, optimizing algorithm control and data fusion, it effectively overcomes the shortcomings of the existing technology and greatly improves the efficiency and data quality of seabed geophysical exploration.

[0137] In order to better understand and implement the present invention, a specific embodiment 1 of the geophysical detection control module in the present invention is provided below. The steps of this embodiment 1 are specifically described as follows:

[0138] The specific implementation of step S10 is as follows:

[0139] First, the seafloor crawler is controlled to crawl along the predetermined detection area. The crawler is equipped with a positioning sensor, which can obtain its real-time position coordinates P t =[x t ,y t ,z t ] T and attitude angle θ t =[θ x ,θ y ,θ z ] T , where t represents the timestamp. These pose information will provide the basis for the subsequent spatial registration of the detection data.

[0140] At the same time, the control chip is also responsible for coordinating the work of the seismograph source. The seismograph source is set to continuously transmit seismic waves to the seabed with predetermined parameters, such as source strength S, transmission frequency f, etc. These seismic waves propagate in the seabed strata and are received by the seismograph receiving array as reflected wave signals r(t). According to the transmission time t of the seismic wave s and the receiving time t of the receiving array r , the seabed position P corresponding to the reflected wave signal can be calculated r , the specific calculation formula is:

[0141]

[0142] Here, c represents the speed of sound waves in water.

[0143] Through the position information of the crawler and the propagation characteristics of seismic waves, the corresponding relationship between the reflected wave signal and the position of the seabed strata can be established, which lays the foundation for subsequent stratum analysis.

[0144] The specific implementation of step S20 is as follows:

[0145] First, the received reflected wave signal r(t) is denoised to eliminate the environmental noise and the noise interference of the measurement equipment itself. Common denoising methods include time domain filtering and frequency domain filtering.

[0146] Time domain filtering can use Wiener filter, whose transfer function is:

[0147]

[0148] Among them, S s (f) and S n (f) represents the power spectral density of signal and noise respectively. By adaptively estimating these two spectral densities, the optimal Wiener filter can be constructed.

[0149] Frequency domain filtering can use Butterworth filter, whose amplitude-frequency characteristic is:

[0150]

[0151] Among them, f c is the cutoff frequency, and n is the order of the filter. Reasonable selection of f c and n can effectively remove useless high-frequency or low-frequency components.

[0152] Next, the denoised signal is adaptively gain controlled to amplify weak signals and suppress strong signals, so that the overall dynamic range remains at an appropriate level. Gain control can be performed using an adaptive algorithm based on LMSE (minimum mean square error), and its recursive formula is:

[0153] g(n)=g(n-1)+μe(n)r(n-1);

[0154] Where g(n) is the gain value of the nth sampling point, μ is the step size parameter, and e(n) is the error of the nth sampling point, which is defined as the difference between the expected output and the actual output.

[0155] Through the above preprocessing, the quality of the reflected wave signal can be significantly improved, laying the foundation for the subsequent time-frequency feature extraction.

[0156] The specific implementation of step S30 is as follows:

[0157] First, the sliding time window method is used to segment the preprocessed reflection wave data r(t). Specifically, the entire reflection wave signal is divided into N time windows, each with a length of T. w , there is a certain overlap between adjacent windows. Let the reflected wave signal in the i-th time window be r i (t), where i = 1, 2,…, N.

[0158] For each time window r i (t), extract its frequency, amplitude and phase characteristics. The frequency characteristics can be obtained by short-time Fourier transform (STFT):

[0159]

[0160] Among them, R i (f) is the spectrum of the i-th time window, and f is the frequency variable. i The main frequency f can be extracted from (f) i max And other features.

[0161] The Amplitude feature calculates the root mean square (RMS) value of a signal:

[0162]

[0163] The phase feature can directly extract the phase angle φ of the signal i .

[0164] Through the above processing, the characteristic description of the reflected wave in two dimensions of time and frequency is obtained. Provide a basis for subsequent outlier identification.

[0165] The specific implementation of step S40 is as follows:

[0166] According to the pre-set threshold, identify the abnormal points in the reflection wave characteristics. Specifically, for the three types of characteristics, frequency, amplitude and phase, set reasonable thresholds respectively:

[0167] Frequency anomaly threshold:

[0168] Amplitude anomaly threshold:

[0169] Phase abnormality threshold: |φ i -φ i-1 |>30° / ms;

[0170] in, Represents the global average amplitude value.

[0171] If the characteristic value in a certain time window exceeds any of the above thresholds, it is determined that there are abnormal points in the window. These abnormal points often correspond to discontinuities or lithology changes in the seabed strata, providing important clues for subsequent geological analysis.

[0172] Through the above threshold judgment, geological anomaly information can be preliminarily screened out from the time-frequency characteristics of the reflected wave, providing a basis for the next step of source parameter optimization.

[0173] The specific implementation of step S50 is as follows:

[0174] According to the time position of the abnormal point, it is time-aligned with the emitted seismic wave sequence to establish the corresponding relationship between the abnormal point and the specific seismic wave.

[0175] First, according to the seismic wave emission time t s and the receiving array receiving time t r , calculate the seabed position P corresponding to each reflected wave signal r , as shown in step S10. Then, the abnormal points identified in step S40 are mapped to these positions to determine the seismic wave emission sequences corresponding to them.

[0176] Specifically, for the abnormal point in the i-th time window, the corresponding seismic wave emission time can be calculated as:

[0177]

[0178] Through this time alignment, an M×N outlier-seismic wave correspondence matrix A can be established, where M is the number of outliers and N is the number of seismic wave emissions. Matrix A describes the distribution characteristics of outliers in time and space, laying the foundation for subsequent array decomposition.

[0179] The specific implementation of step S60 is as follows:

[0180] Perform singular value decomposition (SVD) on the abnormal point-seismic wave correspondence matrix A established in step S50 to obtain:

[0181] A=UΣV T ;

[0182] Among them, U and V are orthogonal matrices, and Σ is a diagonal matrix containing the singular values ​​of matrix A.

[0183] Next, U and V are considered as basic matrices, which contain the basic characteristics of the seabed stratigraphic structure; Σ is considered as a variable matrix, which reflects the dynamic changes of the stratigraphic structure.

[0184] Specifically, the basic matrices U and V describe the inherent connection between the anomaly point and the seismic wave, while the variation matrix Σ reflects the dynamic change law of this connection.

[0185] In this way, the original abnormal point-seismic wave correspondence is decomposed into the basic part and the variable part, providing more detailed input for the subsequent optimization of earthquake source parameters.

[0186] The specific implementation of step S70 is as follows:

[0187] The change matrix Σ obtained in step S60 is input into the pre-trained seismograph source parameter model to obtain the optimized source parameters.

[0188] This parameter model adopts the architecture of Generative Adversarial Network (GAN), which includes two main parts: generator G and discriminator D. The generator is responsible for outputting the optimized source parameters S according to the input change matrix Σ and random noise z; the discriminator is responsible for determining whether these parameters are consistent with the real data.

[0189] The specific structure of the generator is as follows:

[0190] 1. Input layer: receives two inputs

[0191] a. A vector z containing 100 random noise values, which are sampled from a standard normal distribution;

[0192] b. Change matrix Σ;

[0193] 2. Matrix processing layer:

[0194] a. Use 1×1 convolution to reduce the dimension of the change matrix Σ and compress it into a k-dimensional vector;

[0195] b. Concatenate the compressed k-dimensional vector with the 100-dimensional random noise vector z to obtain a (100+k)-dimensional vector;

[0196] 3. Fully connected layer:

[0197] a. The first hidden layer: 256 neurons, using LeakyReLU activation function;

[0198] b. Second hidden layer: 128 neurons, using LeakyReLU activation function;

[0199] c. The third hidden layer: 64 neurons, using LeakyReLU activation function;

[0200] 4. Output layer: Using tanh activation function, the output dimension is the same as the number of source parameters.

[0201] The discriminator adopts the structure of convolutional neural network, which includes 3 convolutional layers and 2 fully connected layers.

[0202] During the training process, the generator and the discriminator are trained adversarially with the Wasserstein loss function as the target until the convergence condition is reached. After the training is completed, the change matrix Σ is input into the generator to obtain the optimized source parameter S. This method makes full use of the information of the change matrix and can learn a more accurate source parameter generation strategy.

[0203] The specific implementation of step S80 is as follows:

[0204] First, according to the crawler's posture information P t and θ t, dynamic correction is performed on the reflected wave data r(t) to eliminate the time drift and spatial dislocation caused by the movement of the crawler.

[0205] Specifically, the coordinate transformation of the reflected wave signal can be performed by the following formula:

[0206]

[0207] Among them, r′(t) is the corrected reflected wave signal. This ensures the consistency of the reflected wave data in time and space.

[0208] Then, the corrected reflection wave data r′(t) is subjected to spectrum analysis to extract richer frequency information R′(f)=STFT(r′(t)). These frequency features are helpful in identifying the characteristics of different lithology formations.

[0209] At the same time, the amplitude of the reflected wave is also normalized to eliminate the attenuation caused by the propagation distance. The specific formula is:

[0210]

[0211] Through the above processing, the quality of reflection wave data can be further improved, laying the foundation for subsequent geophysical data fusion analysis.

[0212] The specific implementation of step S90 is as follows:

[0213] The reflected wave data processed in step S80 is and R′(f), are fused and analyzed with the data of the gravimeter and shallow-sediment profiler obtained in step S10 to generate comprehensive geophysical exploration data.

[0214] Specifically, the Bayesian data fusion method can be used to integrate the data obtained by different detection devices into a unified probabilistic framework. and R′(f) obey Gaussian distribution, N(0,C r ), where the covariance matrix C r It can be estimated through the signal preprocessing step. At the same time, the gravity data g and the shallow layer profile data v also obey the Gaussian distribution, N(0,C g ) and N(0,C v ).

[0215] According to Bayesian theory, the optimal estimate of data fusion can be obtained as:

[0216]

[0217] in, is the covariance matrix of the fused data.

[0218] In this way, we get the comprehensive geophysical detection data x fused , including multi-source information such as reflection wave, gravity and shallow stratum profile. This data fusion method can effectively improve the reliability and accuracy of detection results and provide a richer and more reliable basis for subsequent geological interpretation.

[0219] In order to better understand and implement the present invention, the following provides an embodiment 2 of a specific application scenario of the present invention: Figure 3-4 As shown, this Example 2 describes a seafloor geophysical survey operation conducted in the northern continental slope area of ​​the South China Sea. The water depth in this area ranges from 500 to 1500 meters, and the seafloor topography is complex, with many canyons and submarine landslides. The goal of this survey is to conduct a detailed geophysical survey of the area to understand the seafloor geological structure, gravity anomaly distribution, and shallow sediment characteristics.

[0220] 1. Detection system configuration

[0221] The geophysical exploration system based on seafloor crawlers used in this survey includes the following components:

[0222] 1.1 The survey vessel used is 85 meters in length, 16 meters in width, and has a displacement of 3,500 tons. The following equipment is equipped on board:

[0223] Satellite navigation system: adopts Beidou / GPS dual-mode positioning system, with positioning accuracy better than 0.5 meters;

[0224] Underwater positioning system: Using the ultra-short baseline (USBL) positioning system, the positioning accuracy is better than 1.5 meters at a water depth of 1000 meters;

[0225] Power supply unit: provides 380V AC power, with a maximum output power of 200kW;

[0226] 1.2 The submarine crawler adopts a crawler-type submarine crawling robot, model "Hailong I", and its main parameters are as follows:

[0227] Size: 3.5 meters long, 2.2 meters wide, 1.8 meters high;

[0228] Weight: 3 tons in water;

[0229] Maximum operating depth: 3000 meters;

[0230] Maximum crawling speed: 2 km / h;

[0231] Battery life: 48 hours (full battery power mode);

[0232] 1.3 Towing cable

[0233] Length: 200m;

[0234] Diameter: 32 mm;

[0235] Material: Steel wire rope core with polyurethane sheath;

[0236] Load-bearing capacity: 5 tons;

[0237] 1.4 The gravimeter uses a seabed gravimeter, model "Gravity Detection 100", and its main parameters are as follows:

[0238] Measuring range: 0-2000mGal;

[0239] Accuracy: 0.01mGal;

[0240] Drift: <0.1mGal / month;

[0241] Sampling frequency: 1Hz;

[0242] 1.5 The shallow layer profiler adopts a parametric shallow layer profiler, model "seabed SBP2000", the main parameters are as follows:

[0243] Working frequency: 27kHz;

[0244] Resolution: 10cm;

[0245] Penetration depth: up to 100 meters (depending on sediment type);

[0246] Transmitting power: 2kW;

[0247] 1.6 The seismograph source uses an air gun source, model "Ocean Source 500", and the main parameters are as follows:

[0248] Capacity: 500 cubic inches;

[0249] Working pressure: 2000psi;

[0250] Main frequency: 10150Hz;

[0251] Peak sound pressure level: 220dB re 1μPa@1m;

[0252] 1.7 The seismograph receiving array adopts a towed cable receiving array, model "Ocean Receiver 96", the main parameters are as follows:

[0253] Number of channels: 96;

[0254] Track spacing: 12.5 meters;

[0255] Sensitivity: 170dB re 1V / μPa;

[0256] Frequency response: 51000Hz;

[0257] 1.8 The control chip uses a high-performance embedded processor, model "Ocean Control 3000", the main parameters are as follows:

[0258] CPU: Quad-core ARM CortexA72, main frequency 2.0GHz;

[0259] RAM: 16GB;

[0260] Storage: 512GB SSD;

[0261] Interface: Gigabit Ethernet, optical fiber, RS232 / 485;

[0262] 1.9 The data transmission equipment adopts the optical-electric composite cable transmission system, and the main parameters are as follows:

[0263] Transmission bandwidth: 10Gbps;

[0264] Maximum transmission distance: 50 km;

[0265] Number of optical fiber cores: 4 cores;

[0266] Power transmission: 1000V DC, maximum power 100kW;

[0267] 1.10 The self-contained geomagnetic diurnal variation station uses a three-component magnetometer, model "Ocean Magnetic Field 3C", and its main parameters are as follows:

[0268] Measuring range: ±100000nT;

[0269] Resolution: 0.01nT;

[0270] Sampling frequency: 1Hz;

[0271] Battery life: 30 days;

[0272] 1.11 The positioning beacon adopts acoustic release positioning beacon, model is "Ocean Positioning AR", the main parameters are as follows:

[0273] Working frequency: 816kHz;

[0274] Sound source level: 190dB re 1μPa@1m;

[0275] Battery life: 18 months

[0276] Maximum operating depth: 6000 meters;

[0277] 2. Detection operation process

[0278] 2.1 Preparation

[0279] (1) Equipment inspection and installation

[0280] Before the detection operation, the technicians conducted a comprehensive inspection of all equipment. The gravimeter, shallow layer profiler, seismometer source and seismometer receiving array were fixed on the towing cable in sequence, with the interval set to 15 meters. The self-contained geomagnetic diurnal variation station was installed in the dedicated card slot of the seabed crawler.

[0281] (2) System Testing

[0282] The entire system is fully tested on the survey ship deck to ensure that all components are working properly and data transmission is smooth. The test content includes: seabed crawler drive system test; various detection equipment function test; data transmission system test; control system test;

[0283] (3) Parameter settings

[0284] According to the characteristics of the detection area and the detection target, set the following parameters:

[0285] Seabed crawler crawling speed: 1.5 km / h;

[0286] Gravimeter sampling interval: 1 second;

[0287] Subsurface profiler emission interval: 0.5 seconds;

[0288] Seismograph source emission interval: 10 seconds;

[0289] Seismograph receiving array sampling rate: 1000Hz;

[0290] 2.2 Equipment Deployment

[0291] (1) The survey vessel sails to the starting point of the planned survey area (116°30' east longitude, 20°15' north latitude) and slows down to 2 knots.

[0292] (2) The technicians began to lay out the towed cable and the detection equipment fixed on it. First, the seismograph receiving array was put into the water, followed by the seismograph source, the shallow layer profiler and the gravimeter.

[0293] (3) When the towing cable is laid to 180 meters, use the deck crane to slowly lower the seabed crawler into the water.

[0294] (4) Depending on the water depth (the water depth here is about 1000 meters), continue to release the cable until the submarine crawler is about 50 meters away from the seabed.

[0295] (5) Control the seabed crawler to slowly descend and land. Use the depth gauge and altimeter to monitor the crawler's position in real time to ensure a safe landing.

[0296] 2.3 Detection Operation

[0297] (1) After landing, the submarine crawler began to crawl along a preset path. The path was set as a series of parallel survey lines with a spacing of 500 meters and a total coverage area of ​​10 square kilometers.

[0298] (2) The geophysical detection control module performs the following steps:

[0299] S10: Control the seafloor crawler to crawl in the predetermined detection area to obtain the position information of the seafloor crawler. At the same time, control the seismograph source to emit seismic waves every 10 seconds, with a peak sound pressure level of 215dB re 1μPa@1m and a main frequency of 50Hz. The seismograph receiving array continuously receives reflected waves.

[0300] S20: Preprocessing the received reflected wave data, including:

[0301] Denoising: wavelet transform method was used, and the threshold was set to 3 times the standard deviation;

[0302] Filtering: Use a bandpass filter with a passband of 10-200Hz;

[0303] Gain control: adopt program-controlled gain method, the gain curve is g(t)=1+0.5t, where t is the two-way travel time (unit: second);

[0304] S30: Use the sliding time window method to perform segmented analysis on the preprocessed reflection wave data, with the time window length set to 100ms and the overlap rate set to 50%. Extract the reflection wave features in each time window, including:

[0305] Frequency characteristics: main frequency, bandwidth;

[0306] Amplitude characteristics: RMS amplitude, peak value;

[0307] Phase characteristics: instantaneous phase, phase consistency;

[0308] S40: Identify abnormal points in the reflection wave characteristics according to a preset threshold value:

[0309] Abnormal amplitude: exceeding the mean ± 3 times the standard deviation;

[0310] Frequency abnormality: The main frequency deviates from the expected value by more than 20%;

[0311] Phase anomaly: phase consistency is less than 0.6;

[0312] S50: Time-align the identified abnormal points with the transmitted seismic wave sequence to establish a corresponding matrix between the abnormal points and specific seismic waves. The matrix dimension is m×n, where m is the number of abnormal points and n is the number of seismic wave sequences.

[0313] S60: Perform singular value decomposition (SVD) on the corresponding matrix to obtain a basic matrix and a change matrix.

[0314] S70: Input the change matrix into the pre-trained seismograph source parameter model to obtain the optimized parameters:

[0315] Source intensity: adjustment range is 210-220dB re 1μPa@1m;

[0316] Transmitting frequency: Adjustable range is 30-70Hz;

[0317] The seismic source is controlled to re-emit seismic waves according to the optimized parameters, and new reflected wave data are received through the seismic receiving array.

[0318] S80: Process the newly received reflected wave data:

[0319] Dynamic Correction: Use parabolic scanning method, velocity analysis interval is 500 meters;

[0320] Spectral analysis: short-time Fourier transform is used, and the time window length is 50ms;

[0321] Amplitude recovery: spherical diffusion and absorption compensation are used, and the absorption coefficient is set to 0.5dB / km / Hz;

[0322] S90: The processed reflected wave data is integrated and analyzed with the data collected by the gravimeter and the shallow layer profiler to generate comprehensive geophysical detection data.

[0323] (3) At the predetermined location (116°32' east longitude, 20°17' north latitude), the seafloor crawler is controlled to stop and the self-contained geomagnetic diurnal variation station is deployed to the seafloor using a manipulator. After deployment, the seafloor crawler continues to crawl along the predetermined path.

[0324] (4) The gravimeter continuously collects gravity data with a sampling interval of 1 second. The data is preliminarily processed by the control chip, including tidal correction and instrument drift correction.

[0325] (5) The shallow layer profiler emits a sound wave every 0.5 seconds, with a frequency of 5kHz and a transmission power of 1.5kW. The received echo signal is subjected to envelope detection and time-depth conversion to generate a shallow seabed profile.

[0326] (6) The seismograph receiving array continuously receives seismic waves at a sampling rate of 1000 Hz. After the raw data undergo the aforementioned processing steps, an ocean bottom seismic profile is generated.

[0327] 2.4 Data Transmission

[0328] (1) The control chip sends the processed geophysical detection data to the survey ship in real time through the data transmission device. The data transmission rate is 5Gbps.

[0329] (2) The data receiving system on the survey vessel continuously receives and stores the detection data, and simultaneously performs real-time display and preliminary analysis.

[0330] 2.5 Job Monitoring

[0331] (1) The monitoring personnel on the survey vessel monitor the entire detection process through real-time transmitted data, including: the position, posture and motion status of the seafloor crawler, the working status and data quality of each detection equipment, and the seafloor topography and geological conditions;

[0332] (2) If any abnormal situation is found, the monitoring personnel can make real-time adjustments to the seabed crawler and various detection equipment through the control system.

[0333] 2.6 Self-contained geomagnetic daily change station recovery

[0334] (1) After the detection operation is completed, the survey ship controls the seafloor crawler to return to the deployment location of the self-contained geomagnetic diurnal variation station based on the position information of the positioning beacon.

[0335] (2) After the seafloor crawler reaches the designated location, a manipulator is used to retrieve the self-contained geomagnetic diurnal variation station and its accompanying positioning beacon and fix them on the crawler.

[0336] 2.7 Equipment Recycling

[0337] (1) The survey vessel uses a composite cable winch to reel in the cable, causing the seafloor crawler to slowly rise.

[0338] (2) When the submarine crawler approaches the sea surface, it is lifted by a deck crane and recovered to the deck.

[0339] (3) Continue to reel in the cables, and successively recover the gravimeter, shallow-sediment profiler, seismograph source, and seismograph receiving array.

[0340] (4) After all equipment is recovered, technicians will inspect, clean and maintain the equipment.

[0341] 3. Data Processing and Analysis

[0342] 3.1 Gravity data processing

[0343] (1) The original gravity data is processed as follows:

[0344] Instrument drift correction: using a linear drift model with a drift rate of 0.05 mGal / day;

[0345] Correction: Consider the effects of the earth's rotation and the motion of the measuring platform;

[0346] Free air correction: using elevation data measured by GPS;

[0347] Bouguer correction: Assume that the density of seawater is 1.03 g / cm 3 , the density of seafloor sediment is 2.0g / cm 3 ;

[0348] (2) Generate free air anomaly maps and Bouguer anomaly maps with a grid spacing of 100 meters.

[0349] 3.2 Magnetic data processing

[0350] (1) The data collected by the self-contained geomagnetic diurnal variation station are processed as follows: diurnal variation correction: using the data of the nearest land magnetic observatory, IGRF correction: removing the influence of the Earth's main magnetic field;

[0351] (2) Generate a total magnetic intensity anomaly map with a grid spacing of 100 meters.

[0352] 3.3 Shallow stratum profile data processing

[0353] (1) The shallow stratum profile data is processed as follows:

[0354] Denoising: Use median filtering with a window size of 5×5;

[0355] Gain control: Automatic gain control (AGC) is applied with a time window length of 20ms;

[0356] Deconvolution: Use predictive deconvolution with an operator length of 30ms;

[0357] Static correction: Static correction based on seabed topography data;

[0358] (2) Generate a two-dimensional shallow subsurface profile with a horizontal resolution of 1 meter and a vertical resolution of 0.1 meter.

[0359] 3.4 Seismic data processing

[0360] (1) Perform the following advanced processing on seismic data:

[0361] Multiple wave suppression: using predictive deconvolution method, the operator length is 300ms;

[0362] Speed ​​analysis: Perform speed analysis every 500 meters to generate a speed model;

[0363] Prestack time migration: using the Kirchhoff migration algorithm, the migration aperture is 3000 m;

[0364] Post-stack depth conversion: converting time profiles into depth profiles based on velocity models;

[0365] (2) Generate a two-dimensional seismic profile with a horizontal resolution of 12.5 meters and a vertical resolution of 5 meters.

[0366] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A geophysical exploration system based on a seafloor crawler, characterized in that: include: A survey ship, a seabed crawler, a gravimeter, a shallow-layer profiler, a seismograph source, a seismograph receiving array and a control chip, wherein the survey ship is connected to the seabed crawler by a towing rope, a towing cable is provided at the tail of the seabed crawler, and the gravimeter, shallow-layer profiler, seismograph source and seismograph receiving array are fixedly arranged on the towing cable in sequence; an electronic cabin is provided in the seabed crawler, and the data control chip is provided in the electronic cabin, and the control chip is electrically connected with the driving unit of the seabed crawler, the gravimeter, shallow-layer profiler, seismograph source and seismograph receiving array to exchange data; a geophysical detection control module is provided in the control chip, which is used to set the parameters of the seismograph source and pre-process the data collected by the receiver to finally obtain geophysical detection data; a data transmission device is also provided in the electronic cabin, and the control chip is electrically connected with the data transmission device, and the data transmission device is used to send the geophysical detection data to the survey ship.

2. A geophysical exploration system based on a seafloor crawler according to claim 1, characterized in that: It also includes a self-contained geomagnetic diurnal variation station, which is carried by a crawler and deployed on the seabed, and is recovered by the crawler after the operation is completed.

3. A geophysical exploration system based on a seafloor crawler according to claim 2, characterized in that: The seabed crawler is a tracked seabed crawling robot, a multi-legged seabed crawling robot or a seabed mining vehicle.

4. A geophysical exploration system based on a seafloor crawler according to claim 3, characterized in that: The data transmission device sends the geophysical detection data to the survey ship via a data cable or a wireless channel.

5. A geophysical exploration system based on a seafloor crawler according to claim 4, characterized in that: The transmitting array vibration surface of the seismic source of the seismograph is in contact with the seabed sediment.

6. A geophysical exploration system based on a seafloor crawler according to claim 5, characterized in that: Also included is a positioning beacon, which is fixed on the towing cable.

7. A geophysical exploration system based on a seafloor crawler according to claim 6, characterized in that: The survey vessel is equipped with a satellite navigation system, an underwater positioning system, and a power supply unit.

8. A geophysical exploration system based on a seafloor crawler according to claim 7, characterized in that: The seabed crawler is also provided with a sensor module, which at least includes a depth gauge, an altimeter, a gyroscope and an accelerometer.

9. A geophysical exploration system based on a seafloor crawler according to claim 8, characterized in that: It also includes a magnetometer, an electromagnetic instrument excitation source, and an electromagnetic instrument receiving array which are sequentially fixedly arranged on the towing cable.

10. A geophysical exploration system based on a seafloor crawler according to claim 1, characterized in that: The geophysical detection control module is used to perform the following steps: S10, controlling the seafloor crawler to crawl in a predetermined detection area, obtaining the position information of the seafloor crawler, and controlling the seismic source of the seismograph to continuously emit seismic waves and receive reflected waves through the seismograph receiving array; S20, preprocessing the received reflected wave data, including denoising, filtering and gain control; S30, using a sliding time window method to perform segmented analysis on the preprocessed reflected wave data, and extracting the reflected wave characteristics in each time window, including frequency, amplitude and phase characteristics; S40, identifying abnormal points in the reflected wave characteristics according to a preset threshold, including amplitude abnormality, frequency abnormality and phase abnormality; S50, time-aligning the identified abnormal points with the transmitted seismic wave sequence, and establishing a corresponding matrix between the abnormal points and the specific seismic waves; S60, performing singular value decomposition on the corresponding matrix to obtain a basic matrix and a change matrix; S70, inputting the variation matrix into a pre-trained seismograph source parameter model to obtain corresponding optimization parameters, including source intensity and emission frequency; and controlling the seismograph source to re-emit seismic waves according to the optimization parameters, and receiving new reflected wave data through the seismograph receiving array; S80, processing the newly received reflected wave data, including dynamic correction, spectrum analysis and amplitude recovery, to improve data quality; S90. The processed reflected wave data is integrated and analyzed with the data collected by the gravimeter and the shallow layer profiler to generate comprehensive geophysical detection data.

Images