Multi-task learning speed model prediction method

Abstract

The invention relates to a multi-task learning speed model prediction method, and belongs to a speed model prediction method based on in-well seismic exploration DAS-VSP data. And constructing a real geologic structure-like velocity model set and a vertical seismic profile data set, constructing and training a multi-task learning network, and realizing velocity model prediction by using DAS-VSP data. The method has the advantages that the problem that samples are deficient in the network training process is effectively solved, the multi-task learning network of the encoder-double decoder architecture is constructed, the sub-tasks are used for carrying out structural constraint on the main task, the accuracy of the prediction result is enhanced, and an attention mechanism is introduced to improve the DAS-VSP data key feature recognition capability. According to the method, the dependence of a traditional method on prior information is effectively overcome, the calculation efficiency and the prediction precision are remarkably improved, more accurate data support is provided for subsequent imaging and interpreters, and the influence of the DAS-VSP technology in the field of oil and gas resource exploration is improved.

Description

Technical Field

[0001] The present invention relates to a velocity model prediction method based on DAS-VSP data of borehole seismic exploration, and in particular to a prediction method for accurately predicting formation structure and velocity values ​​under multi-task learning. Background Art

[0002] Seismic exploration analyzes the response of artificial seismic waves, including wave velocity, travel time, reflection and refraction angles, to infer the lithology and structure of the strata. As an important means of modern seismic exploration, vertical seismic profile (VSP) has the advantages of high resolution and accurate depth detection. In recent years, distributed acoustic sensors (DAS) have gradually been applied to the field of VSP seismic exploration due to their advantages of low cost, high sensitivity, high precision, large array and repeatable acquisition.

[0003] An accurate velocity model is a key prerequisite for high-resolution seismic imaging technologies such as reverse time migration and prestack depth migration. At present, seismic exploration is gradually developing towards more complex underground structures and oil and gas reservoirs. Although traditional methods such as full waveform inversion (FWI) have high accuracy and strong interpretability, this step-by-step iterative reconstruction of the velocity model requires setting a large number of parameters, which is time-consuming and computationally expensive. The setting of the initial velocity model will also affect the results. In addition, it is extremely difficult to use FWI to predict the velocity model of DAS_VSP data. Therefore, exploring accurate and effective velocity model prediction methods suitable for DAS-VSP data is an urgent problem to be solved in the current field of seismic exploration.

[0004] In recent years, with the rapid development of computing power, many deep learning (DL) models trained with large-scale data sets have emerged to directly predict velocity models from prestack surface seismic data. However, no researchers have yet used deep learning methods to accurately predict velocity models from prestack DAS-VSP data. The main problems are: the lack of velocity model sets required for deep learning network training; the realization of complex mappings such as DAS-VSP data to velocity models is often interfered by various irrelevant information; without appropriate stratigraphic constraints, the network cannot accurately reconstruct fine underground structures. Summary of the invention

[0005] The present invention provides a multi-task learning velocity model prediction method to solve the problems faced by the current traditional velocity model prediction method, such as complex engineering background and difficulty in predicting velocity models from DAS-VSP data. At the same time, the velocity model set required for training deep learning methods is scarce, and the realization of complex mappings such as DAS-VSP data to velocity models is often interfered by various irrelevant information. Without appropriate stratum constraints, the network cannot accurately reconstruct fine underground structures, which are key problems.

[0006] The technical solution adopted by the present invention comprises the following steps:

[0007] Step 1: Construct a velocity model set of quasi-real geological structures and a DAS-VSP dataset;

[0008] (1) Construct a set of velocity models V that resemble real geological structures;

[0009] The velocity model construction process includes drawing formation curves and filling velocity values. Drawing formation structure curves requires the use of linear and trigonometric functions to simulate continuous and fluctuating formation structures.

[0010] The curve equation of the mth layer can be expressed as follows:

[0011]

[0012] where a j ,T j ,θ j , j = 1, 2, 3 are the trigonometric function amplitude, period, and phase parameters, which are used to simulate formation fluctuations in different situations. m is the formation inclination parameter, c m is the layer depth parameter:

[0013] Construct horizontal velocity model, inclined velocity model, wave velocity model and fault velocity model;

[0014] According to the above velocity model construction method, p random independent velocity models v are generated in batches. i , i∈[1,p], forming a velocity model set V={v i |i=1,2,…p};

[0015] (2) Construct the DAS-VSP dataset D and obtain DAS-VSP data by forward modeling of the velocity model;

[0016] Step 2: Build and train a multi-task learning network;

[0017] (1) Constructing a multi-task learning network

[0018] The multi-task learning network adopts an encoder-dual decoder architecture. The main decoder performs the main task of velocity model prediction, and the sub-decoder performs the sub-task of velocity model structure reconstruction.

[0019] (2) Train the multi-task learning network;

[0020] Establish a training set. After training, the multi-task learning network can i The velocity model v is predicted under the condition of i ;

[0021] Step 3: DAS-VSP data is used to make velocity model predictions;

[0022] The output of the main task decoder is the result of velocity model prediction based on DAS-VSP data.

[0023] In step 1 (1) of the present invention, in constructing a set of velocity models V of quasi-real geological structures, a horizontal velocity model is constructed according to the following steps:

[0024] 1) According to formula (1), set a j =0, j=1,2,3, r1=0, c1>0 to generate a straight line as the top interface of the velocity model;

[0025] 2) According to formula (1), set a j =0,j=1,2,3,r m =0, c m >c m-1 Iteratively generate the lower layer interface until all the stratum interfaces are generated;

[0026] 3) Following the principle that the deeper the medium, the greater the propagation velocity, the top layer velocity value is set to q1, ensuring that the mth layer velocity value is q m -q m-1 >q min Under these conditions, iteratively add velocity values ​​at each layer.

[0027] In step 1 (1) of the present invention, in constructing a velocity model set V of a similar real geological structure, an inclined velocity model is constructed according to the following steps:

[0028] 1) According to formula (1), set a j =0, j=1,2,3, r1≠0, c1>0 generates a slant line as the top interface of the velocity model;

[0029] 2) According to formula (1), set a j =0,j=1,2,3,r m ≠0, (y m ) min>(y m-1 ) max Iteratively generate the lower layer interface until all the stratum interfaces are generated;

[0030] 3) Following the principle that the deeper the medium, the greater the propagation velocity, the top layer velocity value is set to q1, ensuring that the mth layer velocity value is q m -q m-1 >q min Under these conditions, iteratively add velocity values ​​at each layer.

[0031] In step 1 (1) of the present invention, in constructing a velocity model set V of a similar real geological structure, a wave velocity model is constructed according to the following steps:

[0032] 1) According to formula (1), set a min j max ,j=1,2,3,T min <T j <T max ,j=1,2,3,θ min <θ j <θ max ,j=1,2,3,r1≠0,c1>0 to generate a fluctuation curve as the top interface of the velocity model;

[0033] 2) According to formula (1), set a min j max ,j=1,2,3,T min <T j <T max ,j=1,2,3,θ min <θ j <θ max ,j=1,2,3,r m ≠0, (y m ) min >(y m-1 ) max Iteratively generate the lower layer interface until all the stratum interfaces are generated;

[0034] 3) Following the principle that the deeper the medium, the greater the propagation velocity, the top layer velocity value is set to q1, ensuring that the mth layer velocity value is q m -q m-1 >q min Under these conditions, velocity values ​​are iteratively added to each layer.

[0035] In step 1 (1) of the present invention, in constructing a velocity model set V of a similar real geological structure, a fault velocity model is constructed based on the wave velocity model according to the following steps:

[0036] ​​​​1) Select the constructed wave velocity model, randomly set the fault starting point at the top interface, set the fault line slope z<0, and generate the fault line;

[0037] 2) Select the velocity model on the right side of the fault line and move down along the fault line h, h <c1;

[0038] 3) To avoid the loss of speed value caused by the movement of the speed model, the speed value q of the mth layer is guaranteed. m -q m-1 >q min Under these conditions, iteratively add speed values ​​to each speed value missing layer.

[0039] The step (2) of step 1 of the present invention, constructing the DAS-VSP data set D, is specifically as follows:

[0040] From the velocity model set V = {v i |i=1,2,…p} select velocity model v i ,i∈[1,p], set appropriate observation parameters and layout to build a two-dimensional observation system, ensure that DAS-VSP data covers all parts of the velocity model and can capture subtle changes in the underground medium, use elastic wave forward modeling to simulate the actual observation process, and use a single Ricker wavelet to simulate the wavelet form received by the detector.

[0041]

[0042] Where A is the amplitude, f0 is the main frequency of the Ricker wavelet, and t0 is the wavelet delay time. The propagation process of seismic waves in the stratum is simulated by the wave equation, and the corresponding DAS-VSP forward modeling data d is generated by the finite difference method. i ,i∈[1,p],

[0043] Traverse all velocity models in the forward velocity model set V to form a DAS-VSP data set D = {d i |i=1,2,…p}.

[0044] The step 2 (1) of the present invention, constructing a multi-task learning network comprises:

[0045] The encoder consists of m1 Conv+BN+ReLu and m2 feature extraction modules. Each feature extraction module consists of m3 Max pooling and m4 Conv+BN+ReLu, which can realize the high-level feature extraction of DAS-VSP data. Multiple convolutional layers can fully extract the features in DAS-VSP data. The activation function can ensure the nonlinear relationship between layers and improve the mapping ability of the network from DAS-VSP data to high-level features.

[0046] The decoder is composed of a main task decoder and a subtask decoder. The main task decoder is composed of n1 DeConv, n2 feature reconstruction modules and n3 Conv+BN+ReLu. Each feature reconstruction module is composed of n4 Conv+BN+ReLu and n5 DeConv. The main task decoder is used to convert high-level features into velocity model prediction results, wherein multiple deconvolution layer decoding can reconstruct high-level features into a fine velocity model, and the activation function can improve the mapping ability between the network's high-level features and the velocity model; the subtask decoder is also composed of n1 DeConv, n2 feature reconstruction modules and n3 Conv+BN+ReLu. Each feature reconstruction module is composed of n4 Conv+BN+ReLu and n5 DeConv. The subtask decoder is only used to convert high-level features into velocity model structure results, wherein multiple deconvolution layer decoding can reconstruct high-level features into a fine velocity model stratigraphic structure, and the activation function can improve the mapping ability between the network's high-level features and the velocity model structure;

[0047] A jump connection is made between each Max pooling result of the encoder and the DeConv results of the same-sized feature maps in the main decoder and sub-decoder to weaken the loss of DAS-VSP feature information in the downsampling process and fully retain the feature information related to speed reconstruction; an attention module is introduced at the end of the network jump connection to filter out redundant information irrelevant to speed model reconstruction extracted by the encoder and highlight key speed information and high-level structural features; a multi-task learning network architecture is adopted to use the structural information predicted by the sub-task decoder as the structural loss to constrain the main task decoder to achieve high-precision speed model reconstruction.

[0048] In step 2 (2) of the present invention, training the multi-task learning network is as follows:

[0049] Randomly selected from the constructed quasi-real geological structure velocity model set V and the DAS-VSP dataset D Group Paired Data Constructing the training set and

[0050] DAS-VSP data i and velocity model v i The forward relationship between them is expressed as d i =F(v i ), for velocity model prediction, it is necessary to obtain the DAS-VSP data d i The corresponding velocity model is derived from Right now:

[0051]

[0052] Where G1 represents the mapping function represented by the main task decoder and θ represents the model parameters of the multi-task network.

[0053] The loss function is used to update parameters in the velocity model prediction process.

[0054] The loss function and parameter updating process used in the present invention are as follows:

[0055] 1) Mean square error loss L MSE

[0056] The mean square error (MSE) loss function is defined as follows:

[0057]

[0058] Where m represents the number of pixels of the velocity model, and k represents the amount of training data;

[0059] 2) Cross entropy loss L contour

[0060] The cross entropy loss function is defined as follows:

[0061]

[0062] Where C(v i ) is the velocity model v i The outline structure, is C(v i ), s i Represents the result of the softmax operation on the two channels of the subtask decoder:

[0063]

[0064] in The result of the nth, n=1,2th channel of the subtask decoder;

[0065] 3) Overall loss and model parameter θ update

[0066] The overall loss function is defined as follows:

[0067] L total =a1L MSE +a2L contour (7)

[0069] Where α1 and α2 represent L MSE and L contourThe weight occupied in the overall loss. The loss function is used to calculate the deviation between the true value and the predicted value of the velocity model in the training set. Different model parameters θ correspond to different losses. The goal of model training is to find a set of model parameters that minimize the loss. At this point, the network model is considered to have achieved optimal performance. The multi-task learning network is trained and the Adam optimizer is used to accelerate the convergence of the training process. After training, the multi-task learning network can i The velocity model v is predicted under the condition of i .

[0070] The prediction method of step 3 of the present invention is as follows:

[0071] Select the remaining ones from the constructed quasi-realistic geological structure velocity model set V and the DAS-VSP dataset D Group paired data {v i ,d i}, forming a test set and

[0072] The multi-task learning network is used to test the velocity model prediction on DAS-VSP data, that is, from the test set D test Select DAS-VSP data As the network test input, the main task decoder output is the result of the velocity model prediction of DAS-VSP data.

[0073] The present invention proposes a solution for intelligent DAS-VSP velocity model prediction. By automatically constructing a set of velocity models similar to real geological structures, massive data support is provided for network training. At the same time, a multi-task network model is constructed and trained, and an attention gating mechanism is introduced to prompt the network to focus on features that are highly relevant to velocity model prediction and weaken the interference of irrelevant information. In addition, the stratigraphic structure extraction subtask is used to constrain the velocity model prediction main task to perform high-precision velocity model reconstruction. The velocity model prediction solution provided by the present invention is intended to provide a solid foundation for subsequent high-precision imaging of DAS-VSP data.

[0074] The advantages of the present invention are: the multi-task learning network proposed in the present invention realizes an intelligent DAS-VSP data prediction velocity model solution. First, a sufficient set of real geological structure velocity models is automatically constructed, which can effectively solve the problem of the lack of velocity model sets. Secondly, a multi-task learning network is constructed and trained and an attention mechanism is introduced to prompt the network to focus on features that are highly relevant to velocity model prediction. In addition, the sub-task of extracting stratigraphic structure is used to constrain the main task of velocity model prediction, thereby reducing the instability of the prediction results and improving the prediction accuracy of the velocity model. The current traditional method of velocity model prediction requires a large amount of computing resources and is heavily dependent on the initial model. It is even more difficult to use traditional methods to predict the velocity model for DAS-VSP data. The present invention provides a low-cost, high-efficiency and high-precision velocity model prediction method, which is crucial to the DAS-VSP imaging quality, geological stratigraphic interpretability and oil and gas production efficiency, and provides strong technical support for DAS technology in the field of seismic exploration. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Figure 1 It is a schematic diagram of the velocity model and 2D observation system constructed including horizontal, inclined, wavy and fault shapes, where the black inverted triangle is the detector deployed by the DAS system and the white circle is the source deployed by the DAS system;

[0076] Figure 2 yes Figure 1 Four DAS-VSP records obtained by elastic wave forward modeling for the four velocity models;

[0077] Figure 3 It is a multi-task learning network structure diagram, which consists of an encoder-dual decoder structure and an overall Y-shaped structure. The main task decoder is mainly used for high-precision fitting of speed values, and the sub-task decoder assists the main decoder in smooth fitting of layer boundaries. It also includes a feature extraction module and a feature reconstruction module. The feature extraction module consists of 1 Max pooling and 2 Conv+BN+ReLu, and the feature reconstruction module consists of 2 Conv+BN+ReLu and 1 DeConv.

[0078] Figure 4 This is a comparison chart of the prediction results of the multi-task learning network processing the fluctuating speed model and the ground truth, where a) is the ground truth and b) is the prediction result of the trained multi-task learning network.

[0079] Figure 5 This is a comparison chart of the prediction results of the multi-task learning network processing the fault velocity model and the ground truth, where a) is the ground truth and b) is the prediction result of the trained multi-task learning network;

[0080] Figure 6 It corresponds to Figure 4 The velocity and depth curve of the undulating velocity model at an offset of 240m, which intuitively shows the difference between the predicted result and the ground truth;

[0081] Figure 7 It corresponds to Figure 5 The velocity and depth curve of the fault velocity model at an offset of 240m, which intuitively shows the difference between the predicted result and the ground truth;

[0082] Figure 8 This is a graph comparing the experimental results of the multi-task learning network and the network without subtask constraints, where a) is the ground truth, b) is the prediction result of the multi-task learning network, and c) is the prediction result of the network without subtask constraints. DETAILED DESCRIPTION

[0083] The following steps are involved:

[0084] Step 1: Construct a velocity model set of quasi-real geological structures and a DAS-VSP dataset;

[0085] (1) Construct a set of velocity models V that resemble real geological structures;

[0086] The velocity model construction process includes drawing formation curves and filling velocity values. Drawing formation structure curves requires the use of linear and trigonometric functions to simulate continuous and fluctuating formation structures.

[0087] The curve equation of the mth layer can be expressed as follows:

[0088]

[0089] where a j ,T j ,θ j , j = 1, 2, 3 are the trigonometric function amplitude, period, and phase parameters, which are used to simulate formation fluctuations in different situations; r m is the formation inclination parameter, c m is the layer depth parameter;

[0090] Follow these steps to build a horizontal velocity model:

[0091] 1) According to formula (1), set a j =0, j=1,2,3, r1=0, c1>0 to generate a straight line as the top interface of the velocity model;

[0092] 2) According to formula (1), set a j=0,j=1,2,3,r m =0, c m >c m-1 Iteratively generate the lower layer interface until all the stratum interfaces are generated;

[0093] 3) Following the principle that the deeper the medium, the greater the propagation velocity, the top layer velocity value is set to q1, ensuring that the mth layer velocity value is q m -q m-1 >q min Under these conditions, iteratively add speed values ​​to each layer;

[0094] Follow these steps to build a tilted velocity model:

[0095] 1) According to formula (1), set a j =0, j=1,2,3, r1≠0, c1>0 generates a slant line as the top interface of the velocity model;

[0096] 2) According to formula (1), set a j =0,j=1,2,3,r m ≠0, (y m ) min >(y m-1 ) max Iteratively generate the lower layer interface until all the stratum interfaces are generated;

[0097] 3) Following the principle that the deeper the medium, the greater the propagation velocity, set the top layer velocity value to q1, ensuring that the mth layer velocity value is q m -q m-1 >q min Under these conditions, iteratively add speed values ​​to each layer;

[0098] Follow these steps to build a wave-like velocity model:

[0099] 1) According to formula (1), set a min j max ,j=1,2,3,T min <T j <T max ,j=1,2,3,θ min <θ j <θ max ,j=1,2,3,r1≠0,c1>0 to generate a fluctuation curve as the top interface of the velocity model;

[0100] 2) According to formula (1), set a min j max ,j=1,2,3,T min <T​​​​j <T max ,j=1,2,3,θ min <θ j <θ max ,j=1,2,3,r m ≠0, (y m ) min >(y m-1 ) max Iteratively generate the lower layer interface until all the stratum interfaces are generated;

[0101] 3) Following the principle that the deeper the medium, the greater the propagation velocity, set the top layer velocity value to q1, ensuring that the mth layer velocity value is q m -q m-1 >q min Under these conditions, iteratively add velocity values ​​at each layer.

[0102] Based on the wave velocity model, the fault velocity model is constructed by following the steps below:

[0103] 1) Select the constructed wave velocity model, randomly set the fault starting point at the top interface, set the fault line slope z<0, and generate the fault line;

[0104] 2) Select the velocity model on the right side of the fault line and move down along the fault line h, h <c1;

[0105] 3) To avoid the loss of speed value caused by the movement of the speed model, the speed value q of the mth layer is guaranteed. m -q m-1 >q min Under these conditions, iteratively add speed values ​​to each speed value missing layer;

[0106] According to the above velocity model construction method, p random independent velocity models v are generated in batches. i , i∈[1,p], forming a velocity model set V={v i |i=1,2,…p};

[0107] (2) Construct the DAS-VSP dataset D and obtain DAS-VSP data by forward modeling of the velocity model;

[0108] From the velocity model set V = {v i |i=1,2,…p} select velocity model v i , i∈[1,p], set appropriate observation parameters and layout to build a two-dimensional observation system, ensure that DAS-VSP data covers all parts of the velocity model and can capture subtle changes in the underground medium, and use the elastic wave forward modeling method to simulate the actual observation process.

[0109] Equation 2) The wavelet form received by the analog detector is:

[0110]

[0111] Where A is the amplitude, f0 is the main frequency of the Ricker wavelet, and t0 is the wavelet delay time. The wave equation is used to simulate the propagation process of seismic waves in the stratum, and the corresponding DAS-VSP forward modeling data d is generated by the finite difference method. i ,i∈[1,p];

[0112] Traverse all velocity models in the forward velocity model set V to form a DAS-VSP data set D = {d i |i=1,2,…p};

[0113] Step 2: Build and train a multi-task learning network;

[0114] (1) Constructing a multi-task learning network

[0115] The multi-task learning network adopts an encoder-dual decoder architecture. The main decoder performs the main task of velocity model prediction, and the sub-decoder performs the sub-task of velocity model structure reconstruction.

[0116] The encoder consists of m1 Conv+BN+ReLu and m2 feature extraction modules. Each feature extraction module consists of m3 Max pooling and m4 Conv+BN+ReLu, which can realize the high-level feature extraction of DAS-VSP data. Multiple convolutional layers can fully extract the features in DAS-VSP data. The activation function can ensure the nonlinear relationship between layers and improve the mapping ability of the network from DAS-VSP data to high-level features.

[0117] The decoder is composed of a main task decoder and a subtask decoder. The main task decoder is composed of n1 DeConv, n2 feature reconstruction modules and n3 Conv+BN+ReLu. Each feature reconstruction module is composed of n4 Conv+BN+ReLu and n5 DeConv. The main task decoder is used to convert high-level features into velocity model prediction results, wherein multiple deconvolution layer decoding can reconstruct high-level features into a fine velocity model, and the activation function can improve the mapping ability between the network's high-level features and the velocity model; the subtask decoder is also composed of n1 DeConv, n2 feature reconstruction modules and n3 Conv+BN+ReLu. Each feature reconstruction module is composed of n4 Conv+BN+ReLu and n5 DeConv. The subtask decoder is only used to convert high-level features into velocity model structure results, wherein multiple deconvolution layer decoding can reconstruct high-level features into a fine velocity model stratigraphic structure, and the activation function can improve the mapping ability between the network's high-level features and the velocity model structure;

[0118] Make jump connections between each Max pooling result of the encoder and the DeConv results of the same-sized feature maps in the main decoder and sub-decoder to weaken the loss of DAS-VSP feature information during downsampling and fully retain the feature information related to speed reconstruction; introduce an attention module at the end of the network jump connection to filter out redundant information irrelevant to speed model reconstruction extracted by the encoder and highlight key speed information and high-level structural features; adopt a multi-task learning network architecture, use the structural information predicted by the sub-task decoder as the structural loss, and constrain the main task decoder to achieve high-precision speed model reconstruction;

[0119] (2) Training multi-task learning network

[0120] Randomly selected from the constructed quasi-real geological structure velocity model set V and the DAS-VSP dataset D Group Paired Data Constructing the training set and

[0121] DAS-VSP data i and velocity model v i The forward relationship between them is expressed as d i =F(v i ), for velocity model prediction, it is necessary to obtain the DAS-VSP data d i The corresponding velocity model is derived from Right now:

[0122]

[0123] Where G1 represents the mapping function represented by the main task decoder and θ represents the model parameters of the multi-task network.

[0124] The loss function and parameter update process used in the velocity model prediction process of the present invention are as follows:

[0125] 1) Mean square error loss L MSE

[0126] The mean square error (MSE) loss function is defined as follows:

[0127]

[0128] Where m represents the number of pixels of the velocity model, and k represents the amount of training data;

[0129] 2) Cross entropy loss L contour

[0130] The cross entropy loss function is defined as follows:

[0131]

[0132] Where C(v i ) is the velocity model v i The outline structure, is C(v i ), s i Represents the result of the softmax operation on the two channels of the subtask decoder:

[0133]

[0134] in The result of the nth, n=1,2th channel of the subtask decoder;

[0135] 3) Update the overall loss and model parameters θ;

[0136] The overall loss function is defined as follows:

[0137] L total =a1L MSE +a2L contour (7)

[0139] Where α1 and α2 represent L MSE and L contour The weight occupied in the overall loss. The loss function is used to calculate the deviation between the true value and the predicted value of the velocity model in the training set. Different model parameters θ correspond to different losses. The goal of model training is to find a set of model parameters that minimize the loss. At this point, the network model is considered to have achieved optimal performance. The multi-task learning network is trained and the Adam optimizer is used to accelerate the convergence of the training process. After training, the multi-task learning network can i The velocity model v is predicted under the condition of i ;

[0140] Step 3: DAS-VSP data for velocity model prediction

[0141] Select the remaining ones from the constructed quasi-realistic geological structure velocity model set V and the DAS-VSP dataset D Group paired data {v i ,d i}, forming a test set and The multi-task learning network is used to predict the velocity model of DAS-VSP data, that is, from the test set D test Select DAS-VSP data As the network test input, the main task decoder output is the result of the velocity model prediction of DAS-VSP data.

[0142] The present invention is further described below through experimental examples.

[0143] According to geological conditions and actual needs, this experimental example automatically draws a large number of velocity models, covering different stratigraphic structures and velocity distributions. According to Table 1, the DAS system layout and forward simulation stratigraphic parameters are constructed to ensure that the DAS-VSP data covers all parts of the velocity model and can capture subtle changes in the underground medium. In addition, the elastic wave forward modeling method is used to simulate the DAS-VSP data generated by the actual observation system, and the multi-task learning network is trained and tested.

[0144] Step 1: Construct a velocity model set of quasi-real geological structures and a DAS-VSP dataset;

[0145] (1) Construct a set of velocity models V that resemble real geological structures

[0146] Referring to the real geological structure, the size of each velocity model is set to 500×100, the grid size is 5m×5m, the simulated stratigraphic structure size is 2500m×500m, and the velocity parameter q is set 1＝ 2000,q min =200, formation tilt parameter -1 <r m <1, layer depth 10 <c m <20, for the fluctuating velocity model, set a min =-1, a max =1,T min =-2, T max =2,θ min =-π,θ max =π, and a total of 8000 velocity models v, including horizontal, inclined, wavy and fault velocity models, were drawn in batches at a ratio of 2:3:10:5, forming a velocity model set V = {v i |i=1,2,…8000};

[0147] Figure 1 Schematic diagrams showing four structural velocity models are shown. These diagrams include Figure 1 a) Horizontal, Figure 1 b) Inclined, Figure 1 c) Fluctuation and Figure 1 d) Different geological structures such as faults provide rich training data and test samples for subsequent velocity model prediction tasks;

[0148] (2) Construct DAS-VSP dataset D

[0149] Select velocity model v from velocity model dataset V and set it as follows Figure 1 The two-dimensional DAS observation system in the forward modeling is shown in Table 1.

[0150] Table 1. DAS system layout and forward simulation formation parameters

[0151]

[0152] A single shot source is set at an offset of 500m, and 500 geophones are set at 5m intervals in a well with a depth of 2500m. The sixth-order finite difference method is used to perform DAS-VSP data forward simulation, and the PML (Perfectly Matched Layer) boundary absorption condition is used to absorb non-physical reflections from each boundary. The simulation uses a Ricker wavelet with a main frequency of 35Hz close to the real DAS system acquisition and recording, and simulates DAS-VSP data with a sampling frequency of 2000Hz. The observation time of each record is 2 seconds. The velocity model v in the velocity model set V is traversed, and DAS-VSP data is generated forward and constitutes a seismic record data set D = {d i |i=1,2,…8000};

[0153] Figure 2 Shown with Figure 1 DAS-VSP records corresponding to the four velocity models, Figure 2 a) is the DAS-VSP data corresponding to the horizontal velocity model, Figure 2 b) is the DAS-VSP data corresponding to the inclined velocity model. Figure 2 c) is the DAS-VSP data corresponding to the fluctuating velocity model. Figure 2 d) DAS-VSP data corresponding to the fault velocity model;

[0154] Step 2: Build and train a multi-task learning network

[0155] (1) Constructing a multi-task learning network

[0156] The multi-task learning model in this example is Figure 3 As shown in Figure 1, the multi-task learning network adopts an encoder-dual decoder architecture, where the main decoder performs the main task of velocity model prediction and the sub-decoder performs the sub-task of velocity model structure reconstruction.

[0157] The encoder consists of 2 Conv+BN+ReLu and 4 feature extraction modules, each of which consists of 1 Maxpooling and 2 Conv+BN+ReLu. It realizes the advanced feature extraction of DAS-VSP data, where multiple convolutional layers can fully extract the features in DAS-VSP data, and the activation function can ensure the nonlinear relationship between layers, improving the mapping ability of the network from DAS-VSP data to advanced features;

[0158] The decoder consists of a main task decoder and a subtask decoder. The main task decoder consists of 1 DeConv, 3 feature reconstruction modules and 2 Conv+BN+ReLu. Each feature reconstruction module consists of 2 Conv+BN+ReLu and 1 DeConv. The main task decoder is used to convert high-level features into velocity model prediction results. Among them, multiple deconvolution layer decoding can reconstruct high-level features into a detailed velocity model, and the activation function can improve the mapping ability between the network's high-level features and the velocity model; the subtask decoder is also composed of 1 DeConv, 3 feature reconstruction modules and 2 Conv+BN+ReLu. Each feature reconstruction module consists of 2 Conv+BN+ReLu and 1 DeConv. The subtask decoder is only used to convert high-level features into velocity model structure results. Among them, multiple deconvolution layer decoding can reconstruct high-level features into a detailed velocity model stratigraphic structure, and the activation function can improve the mapping ability between the network's high-level features and the velocity model structure;

[0159] A jump connection is made between each Max pooling result of the encoder and the DeConv results of the same-sized feature maps in the main decoder and sub-decoder to weaken the loss of DAS-VSP feature information in the downsampling process and fully retain the feature information related to speed reconstruction; an attention module is introduced at the end of the network jump connection to filter out redundant information irrelevant to speed model reconstruction extracted by the encoder and highlight key speed information and high-level structural features; a multi-task learning network architecture is adopted to use the structural information predicted by the sub-task decoder as the structural loss to constrain the main task decoder to achieve high-precision speed model reconstruction.

[0160] Table 2 Multi-task learning network parameter settings

[0161] Encoder Main decoder Auxiliary decoder Conv(1,64,3),BN,ReLu deconv(1024,512,2),concat deconv(1024,512,2),concat Conv(64,64,3),BN,ReLu Conv(1024,1024,3),BN,ReLu Conv(1024,1024,3),BN,ReLu max pooling(2,2) Conv(512,512,3),BN,ReLu Conv(512,512,3),BN,ReLu Conv(64,128,3),BN,ReLu deconv(512,256,2),concat deconv(512,256,2),concat Conv(128,128,3),BN,ReLu Conv(512,256,3),BN,ReLu Conv(512,256,3),BN,ReLu max pooling(2,2) Conv(256,256,3),BN,ReLu Conv(256,256,3),BN,ReLu Conv(128,256,3),BN,ReLu deconv(256,128,2),concat deconv(256,128,2),concat Conv(256,256,3),BN,ReLu Conv(256,128,3),BN,ReLu Conv(256,128,3),BN,ReLu max pooling(2,2) Conv(128,128,3),BN,ReLu Conv(128,128,3),BN,ReLu Conv(256,512,3),BN,ReLu deconv(128,64,2),concat deconv(128,64,2),concat Conv(512,512,3),BN,ReLu Conv(128,64,3),BN,ReLu Conv(128,64,3),BN,ReLu max pooling(2,2) Conv(64,64,3),BN,ReLu Conv(64,64,3),BN,ReLu Conv(512,1024,3),BN,ReLu Conv(64,1,1) Conv(64,2,1) Conv(1024,1024,3),BN,ReLu

[0162] (2) Training multi-task learning network

[0163] Artificially set the initial model weight parameter α2 / α1=10 6 In the constructed velocity model set and paired DAS-VSP dataset, 6400 pairs of paired data were randomly selected and the training set V was constructed. train = {vi |i=1,2,…6400} and D train ={d i |i=1,2,…6400}, every 16 pairs are regarded as a batch, and are sent to the model for training batch by batch. The loss function L between the label and the predicted value is calculated using formula (4) and formula (5) respectively. MSE and L contour The gradient of the loss function is back-propagated, and the Adam optimizer is used to continuously modify the parameters of each layer of the network using the forward propagation and back-propagation algorithms to alternately update the network parameters. After training for 200 epochs, the overall loss L in formula (7) is total Minimize and converge. At this point, the multi-task learning network achieves the expected performance and will be used to predict the speed model.

[0164] Step 3: DAS-VSP data for velocity model prediction

[0165] After the multi-task learning network model training is completed, the speed model prediction performance test is performed on the model using DAS-VSP data. In this example, data other than the training data is selected to construct the test set V test = {v j |j=1,2,…1600} and D test ={d j |j=1,2,…1600}, select paired data for speed model prediction. Input the DAS-VSP records in the paired data into the trained network model, and compare the speed model prediction results output by the main task decoder with the label speed model in the paired data to analyze the speed model prediction performance of the multi-task learning network. The prediction results are shown in Figure 4 and Figure 5 As shown, two groups of representative wave-shaped and fault velocity models are selected for visual comparison. Figure 4 a) and Figure 4 b) are the prediction results and ground truth of the fluctuating velocity model, Figure 5 a) and Figure 5 b) The prediction results and ground truth of the fault velocity model. It can be seen that the network can accurately predict the velocity value and stratigraphic structure of each layer, and can restore the microstructure in the stratigraphic layer. In order to further refine the difference between the prediction results and the ground truth, Figure 6 Neutralization Figure 7 They are drawn in Figure 4 and Figure 5The velocity and depth curves of the wave velocity model and fault velocity model at an offset of 240 meters are shown in Fig. 1. It can be seen that the method of the present invention does not rely on the initial velocity model, but can accurately fit the velocity value of each layer, indicating the accuracy of the model. To further verify the performance of the multi-task learning network, the network without subtask constraints is trained and compared. Figure 8 a) is the velocity model groundtruth, Figure 8 b) is the prediction result of the multi-task network. Figure 8 c) is the prediction result of the network without subtask constraints. It can be seen that the multi-task network has obvious advantages in both structural characterization and deep velocity value prediction, indicating the stability of the model. In order to quantify the prediction effect of the method of the present invention, the following four indicators are selected: MAE (Mean Absolute Error), MSE (Mean Squared Error), UQI (Universal Quality Index) and LPIPS (Learned Perceptual Image Patch Similarity, the results are shown in Table 3.

[0166] Table 3. Statistical indicators of the prediction results of the present invention

[0167]

[0168] It can be observed that MAE and LPIPS reach 10 -2 , MSE reaches 10 -3 , while UQI is close to 1. The results show that the model has high reliability and demonstrates the superior performance of the method of the present invention. Through the above qualitative and quantitative analysis, it is verified that the multi-task learning network model can not only accurately predict the formation velocity distribution, but also effectively capture the subtle structural changes in the formation, providing high-precision technical support for the DAS-VSP data in the velocity model prediction process.

[0169] In summary, the current traditional methods for velocity model prediction require a lot of computing resources and are heavily dependent on the initial model. It is even more difficult to use traditional methods to predict the velocity model for DAS-VSP data. The present invention proposes an intelligent multi-task learning scheme for predicting velocity models using DAS-VSP data. First, a sufficient set of real geological structure velocity models is automatically constructed to solve the problem of the lack of velocity model sets. Secondly, a multi-task learning network model is constructed and trained and an attention mechanism is introduced to prompt the network to focus on features that are highly relevant to velocity model prediction. In addition, the sub-task of extracting stratigraphic structure is used to constrain the main task of velocity model prediction, thereby reducing the instability of the prediction results and improving the reconstruction accuracy of the velocity model.

[0170] The present invention automatically extracts high-level features from DAS-VSP data through a multi-task learning network to achieve high-precision velocity model prediction, providing a low-cost, high-efficiency and high-precision velocity model prediction method, which is crucial to DAS-VSP imaging quality, geological stratum interpretability and oil and gas production efficiency, and provides strong technical support for DAS technology in the field of seismic exploration.

Claims

1. A multi-task learning speed model prediction method, characterized in that: The following steps are involved: Step 1: Construct a velocity model set of quasi-real geological structures and a DAS-VSP dataset; (1) Construct a set of velocity models V that resemble real geological structures; The velocity model construction process includes drawing formation curves and filling velocity values. Drawing formation structure curves requires the use of linear and trigonometric functions to simulate continuous and fluctuating formation structures. The curve equation of the mth layer can be expressed as follows: where a j ,T j ,θ j , j = 1, 2, 3 are the trigonometric function amplitude, period, and phase parameters, which are used to simulate formation fluctuations in different situations. m is the formation inclination parameter, c m is the layer depth parameter: Construct horizontal velocity model, inclined velocity model, wave velocity model and fault velocity model; According to the above velocity model construction method, p random independent velocity models v are generated in batches. i , i∈[1,p], forming a velocity model set V={v i |i=1,2,…p}; (2) Construct the DAS-VSP dataset D and obtain DAS-VSP data by forward modeling of the velocity model; Step 2: Build and train a multi-task learning network; (1) Constructing a multi-task learning network The multi-task learning network adopts an encoder-dual decoder architecture. The main decoder performs the main task of velocity model prediction, and the sub-decoder performs the sub-task of velocity model structure reconstruction. (2) Train the multi-task learning network; Establish a training set. After training, the multi-task learning network can i The velocity model v is predicted under the condition of i ; Step 3: DAS-VSP data is used to make velocity model predictions; The output of the main task decoder is the result of velocity model prediction based on DAS-VSP data.

2. A multi-task learning speed model prediction method according to claim 1, characterized in that: In the step 1 (1), in constructing a velocity model set V of a similar real geological structure, a horizontal velocity model is constructed according to the following steps: 1) According to formula (1), set a j =0, j=1,2,3, r1=0, c1>0 to generate a straight line as the top interface of the velocity model; 2) According to formula (1), set a j =0,j=1,2,3,r m =0, c m >c m-1 Iteratively generate the lower layer interface until all the stratum interfaces are generated; 3) Following the principle that the deeper the medium, the greater the propagation velocity, the top layer velocity value is set to q1, ensuring that the mth layer velocity value is q m -q m-1 >q min Under these conditions, iteratively add velocity values ​​at each layer.

3. A multi-task learning speed model prediction method according to claim 1, characterized in that: In the step 1 (1), in constructing a velocity model set V of a similar real geological structure, an inclined velocity model is constructed according to the following steps: 1) According to formula (1), set a j =0, j=1,2,3, r1≠0, c1>0 generates a slant line as the top interface of the velocity model; 2) According to formula (1), set a j =0,j=1,2,3,r m ≠0, (y m ) min >(y m-1 ) max Iteratively generate the lower layer interface until all the stratum interfaces are generated; 3) Following the principle that the deeper the medium, the greater the propagation velocity, the top layer velocity value is set to q1, ensuring that the mth layer velocity value is q m -q m-1 >q min Under these conditions, iteratively add velocity values ​​at each layer.

4. A multi-task learning speed model prediction method according to claim 1, characterized in that: In the step 1 (1), in constructing a velocity model set V of a similar real geological structure, a wave velocity model is constructed according to the following steps: 1) According to formula (1), set a min j max ,j=1,2,3,T min <T j <T max ,j=1,2,3,​​ θ min <θ j <θ max ,j=1,2,3,r1≠0,c1>0 to generate a fluctuation curve as the top interface of the velocity model; 2) According to formula (1), set a min j max ,j=1,2,3,T min <T j <T max ,j=1,2,3,θ min <θ j <θ max ,j=1,2,3,r m ≠0, (y m ) min >(y m-1 ) max Iteratively generate the lower layer interface until all the stratum interfaces are generated;​​ 3) Following the principle that the deeper the medium, the greater the propagation velocity, the top layer velocity value is set to q1, ensuring that the mth layer velocity value is q m -q m-1 >q min Under these conditions, velocity values ​​are iteratively added to each layer.

5. A multi-task learning speed model prediction method according to claim 4, characterized in that: In the step 1 (1), in constructing a velocity model set V of a similar real geological structure, a fault velocity model is constructed based on the wave velocity model according to the following steps: 1) Select the constructed wave velocity model, randomly set the fault starting point at the top interface, set the fault line slope z<0, and generate the fault line; 2) Select the velocity model on the right side of the fault line and move down along the fault line h, h <c1; 3) To avoid the loss of speed value caused by the movement of the speed model, the speed value q of the mth layer is guaranteed. m -q m-1 >q min Under these conditions, iteratively add speed values ​​to each speed value missing layer.

6. A multi-task learning speed model prediction method according to claim 1, characterized in that: The step 1 (2) of constructing the DAS-VSP dataset D is as follows: From the velocity model set V = {v i |i=1,2,…p} select velocity model v i ,i∈[1,p], set appropriate observation parameters and layout to build a two-dimensional observation system, ensure that DAS-VSP data covers all parts of the velocity model and can capture subtle changes in the underground medium, use elastic wave forward modeling to simulate the actual observation process, and use a single Ricker wavelet to simulate the wavelet form received by the detector. Where A is the amplitude, f0 is the main frequency of the Ricker wavelet, and t0 is the wavelet delay time. The wave equation is used to simulate the propagation process of seismic waves in the stratum, and the corresponding DAS-VSP forward modeling data d is generated by the finite difference method. i ,i∈[1,p], Traverse all velocity models in the forward velocity model set V to form a DAS-VSP data set D = {d i |i=1,2,…p}.

7. A multi-task learning speed model prediction method according to claim 1, characterized in that: The step 2 (1) of constructing a multi-task learning network includes: The encoder is composed of m1 Conv+BN+ReLu and m2 feature extraction modules. Each feature extraction module is composed of m3 Maxpooling and m4 Conv+BN+ReLu, which can realize the high-level feature extraction of DAS-VSP data. Multiple convolutional layers can fully extract the features in DAS-VSP data. The activation function can ensure the nonlinear relationship between layers and improve the mapping ability of the network from DAS-VSP data to high-level features. The decoder is composed of a main task decoder and a subtask decoder. The main task decoder is composed of n1 DeConv, n2 feature reconstruction modules and n3 Conv+BN+ReLu. Each feature reconstruction module is composed of n4 Conv+BN+ReLu and n5 DeConv. The main task decoder is used to convert high-level features into velocity model prediction results, wherein multiple deconvolution layer decoding can reconstruct high-level features into a fine velocity model, and the activation function can improve the mapping ability between the network's high-level features and the velocity model; the subtask decoder is also composed of n1 DeConv, n2 feature reconstruction modules and n3 Conv+BN+ReLu. Each feature reconstruction module is composed of n4 Conv+BN+ReLu and n5 DeConv. The subtask decoder is only used to convert high-level features into velocity model structure results, wherein multiple deconvolution layer decoding can reconstruct high-level features into a fine velocity model stratigraphic structure, and the activation function can improve the mapping ability between the network's high-level features and the velocity model structure; A jump connection is made between each Max pooling result of the encoder and the DeConv results of the same-sized feature maps in the main decoder and sub-decoder to weaken the loss of DAS-VSP feature information in the downsampling process and fully retain the feature information related to speed reconstruction; an attention module is introduced at the end of the network jump connection to filter out redundant information irrelevant to speed model reconstruction extracted by the encoder and highlight key speed information and high-level structural features; a multi-task learning network architecture is adopted to use the structural information predicted by the sub-task decoder as the structural loss to constrain the main task decoder to achieve high-precision speed model reconstruction.

8. A multi-task learning speed model prediction method according to claim 1, characterized in that: In step 2 (2), training the multi-task learning network is as follows: Randomly selected from the constructed quasi-real geological structure velocity model set V and the DAS-VSP dataset D Group Paired Data Constructing the training set and DAS-VSP data i and velocity model v i The forward relationship between them is expressed as d i =F(v i ), for velocity model prediction, it is necessary to obtain the DAS-VSP data d i The corresponding velocity model is derived from Right now: Where G1 represents the mapping function represented by the main task decoder, θ represents the model parameters of the multi-task network, The loss function is used to update parameters in the velocity model prediction process.

9. A multi-task learning speed model prediction method according to claim 8, characterized in that: The loss function and parameter update process used are as follows: 1) Mean square error loss L MSE The mean square error (MSE) loss function is defined as follows: Where m represents the number of pixels of the velocity model, and k represents the amount of training data; 2) Cross entropy loss L contour The cross entropy loss function is defined as follows: Where C(v i ) is the velocity model v i The outline structure, is C(v i ), s i Represents the result of the softmax operation on the two channels of the subtask decoder: in The result of the nth, n=1,2th channel of the subtask decoder; 3) Overall loss and model parameter θ update The overall loss function is defined as follows: <h2 style=";text-align:left;direction:ltr">L<h2 style=";text-align:left;direction:ltr"> total <h2 style=";text-align:left;direction:ltr"> =a1L<h2 style=";text-align:left;direction:ltr"> MSE <h2 style=";text-align:left;direction:ltr"> +a2L<h2 style=";text-align:left;direction:ltr"> contour (7) where α1 and α2 represent L MSE and L contour The weight occupied in the overall loss. The loss function is used to calculate the deviation between the true value and the predicted value of the velocity model in the training set. Different model parameters θ correspond to different losses. The goal of model training is to find a set of model parameters that minimize the loss. At this point, the network model is considered to have achieved optimal performance, using the paired data in the training set The multi-task learning network is trained and the Adam optimizer is used to accelerate the convergence of the training process. After training, the multi-task learning network can i The velocity model v is predicted under the condition of i .

10. A multi-task learning speed model prediction method according to claim 1, characterized in that: The prediction method of step 3 is as follows: Select the remaining ones from the constructed quasi-realistic geological structure velocity model set V and the DAS-VSP dataset D Group paired data {v i ,d i }, forming a test set and The multi-task learning network is used to test the velocity model prediction on DAS-VSP data, that is, from the test set D test Select DAS-VSP data As the network test input, the main task decoder output is the result of the velocity model prediction of DAS-VSP data.

Images