234 results about "Seismic facies" patented technology

The invention relates to the technical field of seismic exploration, in particular to a seismic facies recognition semantic segmentation method based on a LightGBM algorithm. The method comprises the following steps: performing multi-dimensional attribute preprocessing on a three-dimensional post-stack seismic body to obtain a geologic body standardized feature matrix; performing region labeling based on the geologic body standardized feature matrix to obtain region labeling data; performing training-verification set division on the region labeling data to obtain seismic model training data and seismic model verification data; carrying out dynamic expansion on the seismic model training data, and carrying out geologic model constraint to obtain seismic constraint feature data; therefore, through the steps of multi-dimensional attribute processing, region labeling, dynamic expansion, noise processing, model optimization and the like on the seismic data, the problems of data isomerism, noise interference and insufficient model generalization ability in traditional seismic data analysis are solved, and the precision and reliability of seismic facies segmentation and geologic body modeling are improved.

The invention discloses a deep learning model for picking up a seismic phase from a seismic signal with a low signal-to-noise ratio. The deep learning model comprises a feature extraction trunk, a bidirectional time sequence-channel attention module (BTCA) and a multi-scale dilated convolutional layer module (DSCN). According to the feature extraction trunk, a cascaded LiteMobileBlock module is used, shallow high-resolution features are extracted from an original waveform step by step, and an initial feature map is generated; the bidirectional time sequence-channel attention module integrates the time sequence modeling capability of the bidirectional LSTM and the spectrum sensing capability (emphasizing importance of different sensors / directions) of a channel attention mechanism, and outputs fusion features; and the multi-scale cavity convolution layer module utilizes parallel cavity convolution modules with different expansion coefficients to synchronously capture a local mutation and global oscillation mode of a waveform to generate multi-scale enhancement features, further enhance the capture capability of the model on long-range time dependence in seismic signals through LSTM, and output time sequence enhancement features. According to the model, a multi-scale cavity convolution module, a bidirectional time-frequency attention mechanism module and an LSTM enhanced sequence modeling module are fused, so that the feature extraction and time sequence modeling capability of a seismic signal with a low signal-to-noise ratio is improved, and robust pickup of a seismic phase is realized.

The invention belongs to the field of ionosphere satellite magnetic field seismic anomaly extraction, and particularly relates to a satellite magnetic field seismic anomaly extraction method based on complex non-negative matrix factorization, which comprises the steps of reading Swarm A satellite magnetic field data, and removing a main magnetic field of a ground core and a lithosphere magnetic field of an earth crust to obtain a residual magnetic field; performing short-time Fourier transform on the preprocessed data to obtain a complex time-frequency matrix in a rectangular coordinate representation form of a complex domain; decomposing the complex time-frequency matrix by using complex non-negative matrix decomposition to obtain R feature components; performing inverse short-time Fourier transform on each feature component to obtain a time domain feature component; a mean square root of a time domain seismic component of each orbit is calculated, a threshold value is set by using an over-limit rate method, and abnormal points are extracted from the seismic components of the orbits. The amplitude and phase information contained in the time-frequency domain of the satellite observation data is fully utilized, so that the component related to the earthquake is identified, and the earthquake anomaly extraction is more accurate and reliable.

The invention relates to a seismic phase pickup method based on time-frequency analysis and deep learning, and belongs to the field of seismology, a deep neural network comprising an empirical mode decomposition module, a residual shrinkage module, an LSTM layer, a transformer layer, an up-sampling layer and a convolutional layer is established, a linear rectification function is adopted as an activation function, an Adam adaptive optimization function is adopted for back propagation, and a seismic phase is obtained. And taking a binary cross entropy loss function as a loss function, and taking a mean absolute error function, an error standard deviation, an accuracy rate, a recall rate and an F1 score as evaluation indexes to obtain a deep neural network model. When the frequencies of the noise, the P wave and the S wave are different, empirical mode decomposition can effectively separate the noise, the P wave and the S wave, so that the network is helped to better identify the P wave and the S wave. Even for broadband noise, the noise can be decomposed into different intrinsic mode functions, so that the noise content in the intrinsic mode functions where P waves and S waves are located is reduced. The time-frequency analysis method can effectively improve the anti-noise capability of the network.

The invention discloses a dynamic polarization window seismic waveform separation method based on a time-frequency ridge line, which utilizes a time-frequency ridge line analysis technology and combines a dynamic polarization window calculation method to accurately calculate the polarization characteristic of a seismic signal. According to the method, the polarization information in the seismic wave propagation process is captured by analyzing the dynamic change of the seismic wave in the time-frequency domain, so that the seismic phase is distinguished, and the purpose of waveform separation is achieved. And more accurate data support and analysis means are provided for the application fields of seismic exploration, seismic early warning, seismic disaster assessment and the like.

The invention discloses a method for predicting a river channel in a high sand-to-ground ratio stratum in a down-down period, and belongs to the technical field of oil-gas exploration. In a method for predicting a river channel sand body in a deep stratum in the down-down period, a lithologic model is established by utilizing an earthquake forward modeling technology to identify a river channel sand earthquake response characteristic; the method comprises the following steps: determining a river channel sand body distribution range and a sand body thickness distribution trend by using seismic attributes and a stratigraphic slicing technology, then determining a river channel sand body space distribution form by using a waveform indication inversion technology, and finally carrying out hydrocarbon detection on the basis of a fluid replacement forward modeling technology. And completing depiction of a favorable area with sand body distribution by combining attribute fusion clustering analysis with a seismic facies method. According to the method, seismic sedimentology and geomorphology are used as theoretical guidance, seismic forward modeling, attribute analysis and inversion technologies are used as technical supports, qualitative description of river channel sand body distribution is converted into specific quantitative analysis, the multiplicity of solutions of conventional river channel sand reservoir prediction is theoretically reduced, the error between the predicted thickness and the actual thickness is reduced, and the prediction accuracy of the river channel sand reservoir is improved. The problem of low prediction precision of the deep riverway type reservoir is solved.

The invention provides a seismic facies classification method and device for seismic data, and the method comprises the steps: dividing a seismic data set in a target work area into K classes, wherein K is equal to the number of seismic facies classes in the target work area; determining initial classification sample data according to the interpretation horizon in the target work area; wherein the initial classification sample data is used as initial data for performing seismic facies classification on the seismic data set; and performing seismic facies classification on the seismic data set according to the initial classification sample data and a pre-generated Markov random field model. According to the method, the actual seismic data in a certain layer are compared trace by trace, and the transverse change of the seismic signal is depicted in detail, so that the plane distribution rule of the seismic anomalous body is obtained.

Systems and methods are provided for subsurface characterization from seismic data. The system can receive a plurality of candidate velocity and anisotropic parameter models and seismic gather data. A subset of the plurality of candidate velocity and anisotropic parameter models can be selected to form a training data set. The system can generate a joint probability functions of depth differences and seismic semblances based on the training data set and generate a likelihood function based on the joint probability function. A Bayesian model can be defined using the likelihood function and the prior probability functions. The system can draw a plurality of samples from a posterior distribution of the Bayesian model using Markov Chain Monte Carlo sampling methods and calculate depth uncertainty values using the plurality of samples.

The invention belongs to but not limited to the technical field of seismic facies intelligent identification, and particularly relates to a three-dimensional seismic facies intelligent identification method and system based on V-Net, and the method comprises the steps: S1, obtaining three-dimensional seismic data and a corresponding seismic facies interpretation result; s2, establishing a learning sample for three-dimensional seismic phase identification; s3, building a seismic facies intelligent identification model based on V-Net; and S4, identifying the three-dimensional seismic facies of unknown data. According to the method, seismic facies recognition is taken as a three-dimensional image segmentation problem, and meanwhile, residual errors are added in modeling, so that the rationality of a model stretching hypothesis space is optimized, and the precision of seismic facies intelligent recognition can be effectively improved. According to the method, the residual layer is added on the basis of three-dimensional U-Net modeling, the structure of the intelligent model is optimized, the reasonability of the model expanding hypothesis space is guaranteed, and the overall feature learning ability of the network is improved.

Methods for seismic diagenetic modeling are disclosed. The methods include obtaining a seismic facies map, well data from one or more wells, and a burial map, and determining input parameters, a burial-history curve, and a thermal-history curve based on the well data. The methods may further include determining a sedimentary facies map based on the seismic facies map and the well data, for each of the sedimentary facies, determining a facies-dependent seismic diagenesis model based on the input parameters, the burial-history curve, the thermal-history curve, the sedimentary facies map, and models, determining a seismic diagenetic model based on the facies-dependent seismic diagenesis model for each of the sedimentary facies and the burial map, and determining an output map based on the seismic diagenetic model. The methods may further include identifying a drilling target based on the output map and designing a wellbore drilling plan based on the drilling target. Systems for seismic diagenetic modeling are also disclosed.

The present invention discloses a method for intelligent identification and detection of seismic phases based on the TransUNet deep learning network. The original three-dimensional seismic data is sliced ​​to obtain the original image of the two-dimensional seismic data. The original image of the two-dimensional seismic data is encoded using a convolutional neural network and a Transformer encoder structure to obtain a seismic data encoding feature map, and then the feature map is upsampled to the size of the original seismic data, thereby realizing intelligent classification of seismic phases. The present invention extracts and encodes features of seismic data using a full convolutional neural network and a Transformer encoder, and uses a U‑Net upsampling structure to decode the encoded feature map to full pixel density to realize intelligent identification and detection of seismic phases, thereby improving the problems of low efficiency and over-reliance on experience in artificial seismic phase classification; using the innate self-attention mechanism of the Transformer and the perception of local details by the U‑Net convolutional neural network, TransUNet can take into account both local detail features and global features, and can be effectively applied to intelligent classification of seismic phases, effectively improving the efficiency and accuracy of seismic phase classification.

The invention relates to the technical field of seismic source fracture analysis, in particular to a seismic source time function extraction method and device, equipment and a storage medium. The method comprises the following steps: firstly, obtaining a station waveform record of a target earthquake, positioning each earthquake event associated with the target earthquake, obtaining a historical Green function of each earthquake event, and then applying a neural network model to the station waveform of the target event and each historical Green function to obtain each visual source time function of the target earthquake; and then carrying out convolution on the historical Green function and the apparent source time function, and screening and determining an optimal apparent source time function of the target earthquake according to a convolution result corresponding to each apparent source time function. According to the method, the target apparent source time function of the target earthquake is calculated through two deterministic parameters including the known station waveform and the historical Green function, and the whole process does not involve parameters needing manual adjustment or marking, so that the stability of the accuracy of the target apparent source time function can be ensured.

The present invention discloses a method and apparatus for establishing an earthquake phase prediction model. The method comprises obtaining a plurality of first-category plane earthquake attribute data and a plurality of second-category plane earthquake attribute data that have been screened and gridded according to the same rule; screening unlabeled grids from each plane earthquake attribute data according to a set rule, extracting each first-category earthquake attribute value and each second-category earthquake attribute value of the labeled grids and the screened unlabeled grids; obtaining a first labeled sample set, a first unlabeled sample set, a second labeled sample set, and a second unlabeled sample set; and using each obtained sample set to train a selected classification model through a collaborative training method to obtain a classification model for earthquake phase prediction. The method can improve the prediction accuracy of earthquake phases and reduce the dependence on labeled samples.

The invention relates to a road cavity detection and identification method and system based on multichannel optical fiber sensing. The method comprises the following steps: acquiring and storing Rayleigh scattering light time domain signals which are acquired by a DAS underground sensing system at different positions and different time periods of a detection road section and contain active knocking signals; determining a frequency range of the active knocking signal according to a frequency spectrum obtained after the Rayleigh scattering light time domain signal is subjected to fast Fourier transform processing; carrying out filtering processing on the Rayleigh scattering light time domain signal by adopting a finite impulse response band-pass filter to obtain a filtered signal; the filtered signal is input into a PhaseNetDAS model for seismic phase pickup, and an active source seismic phase is obtained; associating the active source seismic facies with the active source event to obtain a seismic facies report including event latitude and longitude, time and seismic facies arrival time; and inverting the velocity structure of the underground medium by using the seismic phase report by using a tomography method, and determining a road cavity detection and recognition result, thereby improving the accuracy of road cavity detection.

The present application relates to the technical field of oil reservoir exploration, and particularly relates to a fracture-cave seismic facies identification method and device, which comprises the following steps: obtaining each reservoir section range of a fracture-cave reservoir; obtaining the seismic reflection energy intensity of each reservoir section of the fracture-cave reservoir; obtaining the seismic waveform characteristics of each reservoir section, wherein the seismic waveform characteristics are any one of the following: a first type of seismic waveform with both wave peaks and wave troughs, and a second type of seismic waveform with only wave peaks or only wave troughs; determining the seismic facies category of each reservoir section of the fracture-cave reservoir based on the seismic waveform characteristics and the seismic reflection energy intensity, thereby providing an important basis for the exploration and development of the carbonate rock fracture-cave oil reservoir through effective identification of the seismic facies.

The invention relates to the technical field of seismic facies intelligent identification models, in particular to a seismic facies intelligent identification model construction method and system and a storage medium. The method comprises the following steps: acquiring three-dimensional seismic data and seismic facies label data corresponding to the three-dimensional seismic data, and performing data preprocessing to generate a standard training data set; determining network structure parameters and generating a network model initial architecture based on the structural features of the standard training data set; according to the initial architecture of the network model and the standard training data set, carrying out network parameter training and generating multiple groups of candidate model parameters; through a preset model verification evaluation process, screening out an optimal model parameter from the multiple groups of candidate model parameters, and constructing a seismic facies intelligent identification model; according to the method, the seismic facies intelligent identification model is constructed, so that standardization and automation of the whole process from data preparation to model optimization are realized, and the accuracy and reliability of seismic facies identification are remarkably improved.

The invention belongs to the technical field of oil exploration. The invention discloses a method for determining the age of extrusion resurgence geology of a backlash structure of an ancient and buried hill in a broken basin, and the method comprises the steps: determining the top surface of the ancient and buried hill and the trend of a regional normal fault on a three-dimensional seismic data body of the broken basin subjected to three-dimensional inclination median filtering processing. And determining a thrust fault on a seismic section basically parallel to the trend of the regional normal fault. And recognizing the reverse thrusting anticline structure according to the morphological characteristics of the top surface of the ancient diving mountain. And identifying a slope edge seismic facies, an upper super-irregular seismic reflection interface and a seismic reflection interface corresponding to a fault displacement vanishing point of the thrust fault on an overlying stratum of a footwall of the thrust fault. And determining the starting time and the ending time of the extrusion resurgence of the thrust fault, and determining the geological time stage between the starting time and the ending time as the geological age of the extrusion resurgence of the ancient buried hill thrust structure of the fault basin. By means of the method, the ancient buried hill recoil structure of the broken basin and the geological age of extrusion resurgence of the ancient buried hill recoil structure can be accurately recognized.

The invention discloses an ultra-deep strike-slip fracture reservoir control boundary determination method and device and electronic equipment. The method comprises the following steps: determining a fracture plane distribution condition of a fault zone of a target stratum in a mining area according to a seismic coherence body and a reservoir inversion interpretation result; according to the fracture plane distribution condition and the single well drilling and completion data of at least two drilling wells in the mining area, the drilling well fracture zone vertical distance corresponding to different drilling wells is determined; determining oil and gas exploitation data of at least two drilling wells in the exploitation area according to the single well actual drilling dynamic and static data; according to the vertical distance of the fracture zone of the drilling well and the oil and gas exploitation data, determining a distance oil and gas relation graph group of different drilling wells in the exploitation area; wherein the distance oil-gas relational graph group comprises the relation between the vertical distance of the fault zone of the drilling well and the oil-gas exploitation data of the drilling well; and determining an ultra-deep strike-slip fracture reservoir control boundary according to the distance oil-gas relational graph group. By adopting the technical scheme of the invention, the accuracy of the ultra-deep strike-slip fracture reservoir control boundary determination result is improved.

The invention provides a seismic facies analysis method based on a random projection tree. The analysis method comprises the steps of 1, extracting seismic data needing to be processed according to a seismic data body and a horizon file; 2, selecting seismic layer attributes according to the seismic data; 3, carrying out the dimension reduction of the attributes of the seismic layer through employing a big data nonlinear dimension reduction method based on a random projection tree, and obtaining the attributes of the seismic layer after the dimension reduction; step 4, performing clustering according to the seismic layer attributes after dimension reduction by adopting SOM optimized by PSO to obtain a clustering result; and step 5, outputting a visualization result according to the clustering result. Nonlinear dimensionality reduction is carried out on seismic layer attributes by combining domain space search and a probability model, the method is more suitable for the nonlinear characteristic of seismic layer attribute data, and a scientific basis is provided for the oil reservoir exploration process.

The invention discloses a seismic facies-guided high-precision geological anomalous body identification method, belongs to the technical field of seismic exploration data processing, and aims to consider the change of seismic facies in different regions, perform cross-correlation time shifting on layer-leveled seismic data channel by channel, improve the layer-leveled precision and obtain high-precision geological anomalous body identification factors of a target horizon. And a good foundation is laid for post-fault development law and reservoir identification and prediction. The method is simple in algorithm, high in flexibility, wide in applicability, good in result stability and high in precision.

Seismic facies modeling of an area of study at an oil and gas exploration site includes obtaining a seismic dataset. A set of unsupervised machine learning (USML) models processes a test dataset of the seismic dataset. Respective USML models of the set are configured with different cluster numbers. A USML model and corresponding elbow point cluster number is selected from the set of USML models. The selected USML model, configured with the elbow point cluster number, processes the seismic dataset to obtain clusters of the data points. Cluster profiles based on seismic cell attributes of the data points of each cluster are generated. Seismic facies labels are assigned to the clusters based on corresponding cluster profiles. The clusters are sampled to a three-dimensional (3D) grid representation of the area of study to obtain a seismic facies model displayed in a visualization tool of a seismic modeling platform.

The invention provides a fault control fracture-cavity modeling method utilizing geological trend factor constraints, and relates to the technical field of oil-gas exploration, and the method comprises the steps: calculating seismic coherent body attributes based on post-stack seismic data, obtaining the overall space distribution characteristics of geological structure anomalies and the like, and depicting a strike-slip fracture zone boundary model U1 (x, y, z); calculating the spatial development intensity I1 (x, y, z) of the crack and the hole body based on the distance influence factors from the fracture zone; the longitudinal development strength I2 (x, y, z) of the crack and the hole body is restrained by using the buried depth trend of the stratum, and the comprehensive spatial distribution trend strength Itotal of the crack and the hole body is constructed. The space trend constraint and the stochastic simulation technology are combined, and a space distribution model of the cracks and the hole bodies is established. The model established by the method is more in line with actual space distribution characteristics of cracks and holes in a work area. The method is applied to carrying out numerical simulation research, and is helpful for finding more effective means for processing the fault control fracture-cavity body.

The present invention discloses an automatic seismic facies recognition method based on self-attention mechanism and U-shaped structure. First, an encoder structure based on Segformer is proposed. Then, a block dilation module is introduced to construct a decoder structure, and Hypercolumns technology is introduced to fuse features for seismic facies recognition. Image seismic facies segmentation is performed through a reasonable training and inference process. Finally, a seismic facies body is formed based on the seismic facies recognition result to interpret different underground sedimentary environments. Through experiments on public data and actual data, compared with the seismic facies recognition model of traditional convolutional neural networks, the automatic seismic facies recognition method proposed by the present invention has fewer noise points, more obvious boundary details, stronger lateral continuity, and can more accurately interpret different underground sedimentary environments.

The invention discloses a method and equipment for automatically picking up a seismic phase based on deep learning. The method comprises the following steps: collecting three-component seismic waveform data including P-wave arrival time and S-wave arrival time; preprocessing the three-component seismic waveform data; generating a P-wave arrival time Gaussian distribution curve and an S-wave arrival time Gaussian distribution curve, and constructing a noise distribution curve; the PSNet neural network is trained; and preprocessing three-component seismic waveform data to be processed, inputting the preprocessed three-component seismic waveform data into the trained PSNet neural network, and generating a prediction probability of each category of each time point. According to the invention, the problems of time and labor waste, low efficiency and large error of manual picking of the seismic phase are solved. The differential three-component seismic waveform data of different instruments can be used for training the PSNet neural network, and the seismic phase pickup robustness of the three-component seismic waveform data of different signal-to-noise ratios on different instruments can be effectively improved.

The invention relates to the technical field of geological exploration, in particular to an on-site exploration stratum logging method and system. The method comprises the following steps: identifying a layered boundary in a rock core image by using a deep learning image segmentation model; extracting a stratum interface in the logging curve through wavelet transform and multi-scale analysis; based on a multi-source feature fusion strategy of an attention mechanism and an auto-encoder, the stratigraphic boundary, the stratigraphic interface and the seismic facies feature vectors are combined to analyze comprehensive features of the stratigraphic boundary; according to a stratum attribute graph construction strategy of the graph neural network, associating a stratum reflection layer and the lithology classification feature vector in the radar image to obtain stratum attribute graph features; and constructing a comprehensive decision-making model of a random forest classifier and a rule engine, and generating a stratum catalog report by combining the landform boundary, the stratum boundary comprehensive features and the stratum attribute graph features. According to the method, the accuracy, objectivity and efficiency of the catalog result are improved by cooperatively processing the multi-source data.

The application discloses an unsupervised seismic facies analysis method based on a lognormal mixture variational autoencoder, and comprises the following steps: obtaining prestack seismic data, performing data preprocessing, making a data set, constructing a deep clustering model of the LMVAE, iteratively training the data set by using the deep clustering model of the LMVAE, completing unsupervised prestack seismic data reflection mode analysis, predicting a seismic facies category, and generating a prestack seismic facies map. The method models the lognormal mixture probability of the seismic data in the feature, solves the limitation of the asymmetric data in the deep feature space distribution in the seismic reflection mode analysis, and simultaneously, for the purpose of simplifying the model solution, an inference model using a reparameterization skill for direct optimization is constructed, the inference difficulty problem of the deep generation model under a complex latent structure is overcome, the accuracy of the seismic facies map is improved, and thus strong technical support is provided for the prestack seismic data reflection mode analysis.

The invention discloses a reservoir parameter prediction method and device, electronic equipment and a medium. The method comprises the following steps: determining seismic attribute parameters capable of reflecting reservoir parameter essential characteristics, and predicting a seismic facies through a self-organizing competitive neural network algorithm; adding the seismic facies as transverse constraints into the low-frequency model, and constructing a phase-controlled low-frequency model; and carrying out reservoir parameter inversion through a self-adaptive pre-stack mixed domain inversion method. According to the method, low-frequency information containing seismic facies constraints is added in low-frequency model construction, in addition, in the inversion process, through self-adaption of prior information of different distributions, and meanwhile, mixed domain constraints are added in a likelihood function, the inversion result resolution and transverse continuity are improved, the reservoir description precision is greatly improved, and the method is suitable for large-scale popularization and application. And good support is provided for high-quality reservoir prediction.

The present invention discloses a method for fine fault identification based on ant tracking technology, which is characterized in that the slice data of typical tracks are taken from the original seismic coherence volume data as the seismic coherence volume image, and the high-pass filter is applied to intercept the high-frequency area to reduce the influence of noise; the data blocks are divided according to the number of ants to deploy the ants, the initial search window is set according to the limited range of the fault strike angle, the ant moving point is selected by roulette, the fitness value of the moving path is retained and the search window is adjusted, the pheromone concentration is superimposed according to the ant moving trajectory, and the fine fault identification result is output after the tracking is completed. The present invention uses an adaptive window strategy to plan the action trajectory of the ant tracking technology, so that it has higher search efficiency in simple fault zones and better fault identification accuracy in complex fault zones; and introduces a random ant mutation mechanism to improve the description ability of hidden faults.

The application discloses a kind of seismic facies drive deep carbonate high-precision sedimentary microfacies recovery methods, comprising: basic data collection;Well-seismic calibration;Establish regional isochronous sequence stratigraphic framework;Logging interpretation;Sequence framework within sedimentary facies division: seismic facies division;Establish seismic facies-sedimentary facies conversion relationship;Isochronous sequence stratigraphic framework within division strata slice;Optimization seismic attribute;Establish single factor sedimentary index-seismic attribute relationship;Attribute volume isochronous strata slice;Seismic facies-sedimentary facies under the constraint of high-precision isochronous sedimentary microfacies drawing.The application can accurately predict sequence within sedimentary facies belt boundary by establishing sedimentation-seismic correlation under the condition of less drilling, outcrop;Further optimization can reflect the seismic attribute of sedimentary characteristics, calculate isochronous strata slice;Isochronous sedimentary facies map is obtained by attribute-sedimentary correlation conversion.In practical application, good effect is obtained, and the coincidence rate with actual drilling is high.

The invention discloses a shale-oil-rock facies combined seismic facies inversion method based on deep learning, and the method comprises the steps: carrying out the preprocessing of a pre-stack angle gather and logging data, and carrying out the seismic facies division of seismic data through employing a K-Means clustering algorithm; establishing a three-dimensional convolution deep learning model based on a space-time attention mechanism, introducing a position coding theory in natural language processing to convert a seismic facies classification result into a time sequence code, and introducing the time sequence code into the three-dimensional convolution deep learning model to form a seismic facies control deep learning model; and training the seismic phased deep learning model by adopting a semi-supervised learning method to obtain inversion longitudinal and transverse wave velocity and density result transverse resolution. Transverse resolution of longitudinal and transverse wave velocity and density results of phase-controlled deep learning inversion is superior to that of a non-phase-controlled inversion result, and the thickness of an inverted thin layer is far smaller than one fourth of the wavelength of seismic waves.