Method and system for measuring water absorption profile of well entering based on deep learning model

Abstract

The invention provides a deep learning model-based method and system for measuring a going-into-well water absorption profile, and the method comprises the steps: obtaining to-be-measured going-into-well water absorption profile data, and judging the data type of the to-be-measured going-into-well water absorption profile, the data type being clear data or noise data; if the data type of the to-be-measured well-entering water absorption profile is clear data, marking the to-be-measured well-entering water absorption profile data as clear water absorption profile data, and directly performing water absorption profile measurement on the clear water absorption profile data to obtain a measurement result; if the data type of the to-be-measured well-entering water absorption profile is noise data, marking the to-be-measured well-entering water absorption profile data as noise water absorption profile data; and obtaining de-noising data corresponding to the noise water absorption profile data, and performing water absorption profile measurement on the de-noising data corresponding to the noise water absorption profile data to obtain a measurement result. According to the invention, the accuracy and efficiency of water absorption profile measurement can be improved.

Description

Technical Field

[0001] The present invention relates to the field of oilfield engineering, and more specifically, to a method and system for measuring a water absorption profile of an inflowing well based on a deep learning model. Background Art

[0002] In the process of oilfield development, accurate measurement of downhole water absorption profile is of great significance for optimizing water injection strategy, increasing oil well production and reducing mining costs. Traditional water absorption profile measurement methods mainly rely on physical models and empirical formulas, which often require a large amount of field experimental data and have poor adaptability to complex reservoir conditions. With the development of computer technology, numerical simulation methods have been introduced into water absorption profile measurement, but these methods usually require an in-depth understanding of the physical properties of the reservoir, and the calculation process is complex, making it difficult to process a large amount of downhole data in real time. In addition, the data collected on site often contain noise, which can seriously affect the accuracy of the measurement results. Although some denoising techniques have been proposed, they often require manual intervention and are not adaptable to different types of noise.

[0003] In the process of implementing the embodiments of the present invention, the inventors found that there are at least the following problems or defects in the prior art: the existing water absorption profile measurement method is inefficient in processing noise data, and is not adaptable enough to data under complex reservoir conditions, resulting in the accuracy and real-time performance of the measurement results failing to meet the needs of oilfield development. In addition, the existing denoising technology often requires a lot of manual intervention, and the processing effects of different types of noise vary, making it difficult to achieve automated and intelligent data processing. Summary of the invention

[0004] The present invention provides a method and system for measuring a water absorption profile of a well based on a deep learning model.

[0005] In a first aspect of the present invention, a method for measuring a water absorption profile of a well based on a deep learning model is provided, comprising: Acquire the water absorption profile data of the well to be measured, and determine the data type of the water absorption profile of the well to be measured, wherein the data type is clear data or noise data; If the data type of the water absorption profile to be measured is clear data, the water absorption profile data to be measured is marked as clear water absorption profile data, and the water absorption profile is directly measured on the clear water absorption profile data to obtain a measurement result; If the data type of the water absorption profile into the well to be measured is noise data, the water absorption profile data into the well to be measured is marked as noise water absorption profile data; the denoised data corresponding to the noise water absorption profile data is obtained, and the water absorption profile is measured on the denoised data corresponding to the noise water absorption profile data to obtain a measurement result.

[0006] Furthermore, the determining of the data type of the water absorption profile to be measured includes: Converting the water absorption profile data to be measured into the well into a feature space to obtain a feature value of the water absorption profile data to be measured; If the characteristic value exceeds a preset characteristic value threshold, the data type of the water absorption profile to be measured is clear data; If the characteristic value does not exceed a preset characteristic value threshold, the data type of the water absorption profile to be measured is noise data.

[0007] Further, the obtaining of denoised data corresponding to the noise water absorption profile data includes: Decomposing the noise water absorption profile data in multiple feature dimensions respectively to obtain feature images of corresponding dimensions; De-noising is performed on the feature images of each dimension respectively to obtain the de-noised data corresponding to the feature images of each dimension; The denoised data of the characteristic images of all dimensions are fused to obtain the denoised data corresponding to the noise water absorption profile data.

[0008] Furthermore, obtaining the denoised data corresponding to the feature image of each dimension includes: Acquire deep learning filters of multiple scales, and respectively use the deep learning filters of multiple scales to convolve the feature image to obtain convolution results of corresponding scales; Performing weighted averaging on the convolution results of multiple scales to obtain a weighted average result of the feature image; Based on the feature image and its weighted average result, denoising data of the feature image of the corresponding dimension is obtained.

[0009] Furthermore, the convolution result of the th scale is:

[0010] in, Indicates that the horizontal coordinate in the feature image is , the vertical axis is The pixel value at the location of Indicates Deep learning filters at different scales; Take 1, 2, 3; in,

[0011] in, Indicates The scale parameters of deep learning filters of different scales.

[0012] Furthermore, the weighted average result of the feature image is expressed as:

[0013] in, Indicates The weight coefficient of each scale, .

[0014] Furthermore, the denoised data of the feature image of the corresponding dimension is expressed as:

[0015] Furthermore, the trained deep learning model is used to measure the water absorption profile; The deep learning model is implemented based on a neural network model or a deep learning network model.

[0016] Further, the deep learning model is trained by performing the following operations: Acquire a training set; the training set includes training samples and their measurement results; the training samples include clear water absorption profile data samples and noisy water absorption profile data samples; For a noise water absorption profile data sample, obtaining denoised data corresponding to the noise water absorption profile data sample; The denoised data corresponding to each noise water absorption profile data sample and the measurement results corresponding to the corresponding noise water absorption profile data sample are sequentially input into the deep learning model for training; Each clear water absorption profile data sample and its measurement results are sequentially input into the deep learning model for training; After the training end conditions are met, the trained deep learning model is obtained.

[0017] In a second aspect of the present invention, a system for measuring a water absorption profile of a well based on a deep learning model is provided, comprising: A data acquisition module, used to acquire the water absorption profile data of the well to be measured; A data type determination module, used to determine the data type of the water absorption profile to be measured, wherein the data type is clear data or noise data; A clear data measurement module, for marking the water absorption profile data to be measured as clear water absorption profile data when the data type of the water absorption profile to be measured is clear data, and directly performing water absorption profile measurement on the clear water absorption profile data to obtain a measurement result; The noise data measurement module is used to mark the water absorption profile data to be measured as noise water absorption profile data when the data type of the water absorption profile to be measured is noise data; obtain the denoised data corresponding to the noise water absorption profile data, and perform water absorption profile measurement on the denoised data corresponding to the noise water absorption profile data to obtain a measurement result.

[0018] According to the above-mentioned embodiments of the present invention, at least the following beneficial effects are achieved: the use of a deep learning model to intelligently classify and process downhole water absorption profile data can significantly improve the degree of automation of data processing and the accuracy of measurement results. By converting the data to be measured into a feature space and obtaining eigenvalues, clear data and noise data can be effectively distinguished, thereby adopting corresponding processing strategies for different types of data. For noise data, more accurate denoised data can be obtained by decomposing and denoising in multiple feature dimensions, thereby improving the accuracy of water absorption profile measurement. In addition, the use of deep learning filters of multiple scales to convolve and weighted average the feature images can further improve the denoising effect and ensure the reliability of the measurement results.

[0019] By training the deep learning model, it can be adapted to different reservoir conditions and data characteristics, thereby improving the generalization ability of the model. In actual operation, this method can reduce the reliance on manual intervention, realize fast and automated water absorption profile measurement, and provide more efficient and accurate data support for oilfield development. At the same time, this method can also reduce the complexity of data processing, save time and cost, and is of great significance to improving the efficiency and economic benefits of oilfield development. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood by reading the detailed description below with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, in which: Figure 1 A schematic diagram of a flow chart of a method for measuring a water absorption profile of a well based on a deep learning model provided in one embodiment of the present invention; Figure 2 A schematic diagram of the structure of a well water absorption profile measurement system based on a deep learning model provided by an embodiment of the present invention; Figure 3 The schematic diagram schematically shows the structure of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0021] The principles and spirit of the present invention will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are provided only to enable those skilled in the art to better understand and implement the present invention, and are not intended to limit the scope of the present invention in any way. On the contrary, these embodiments are provided to make the present invention more thorough and complete, and to fully convey the scope of the present invention to those skilled in the art.

[0022] Those skilled in the art know that the embodiments of the present invention can be implemented as a system, device, apparatus, method or computer program product. Therefore, the present invention can be specifically implemented in the following forms, namely: complete hardware, complete software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0023] It should be noted that any number of elements in the drawings is for illustration rather than limitation, and any naming is only for distinction and does not have any limiting meaning.

[0024] Reference below Figure 1 , Figure 1 A schematic diagram of a flow chart of a method for measuring a water absorption profile of a well based on a deep learning model provided by an embodiment of the present invention. Figure 1 As shown, a method 100 for measuring a water absorption profile of a well based on a deep learning model includes: Step 101, obtaining the water absorption profile data of the well to be measured, and determining the data type of the water absorption profile of the well to be measured, wherein the data type is clear data or noise data; Step 102, if the data type of the water absorption profile to be measured is clear data, mark the water absorption profile data to be measured as clear water absorption profile data, and directly measure the water absorption profile of the clear water absorption profile data to obtain a measurement result; Step 103: If the data type of the water absorption profile into the well to be measured is noise data, mark the water absorption profile data into the well to be measured as noise water absorption profile data; obtain denoised data corresponding to the noise water absorption profile data, and perform water absorption profile measurement on the denoised data corresponding to the noise water absorption profile data to obtain a measurement result.

[0025] It should be noted that the present invention relates to a method for measuring the water absorption profile of an inflow well based on a deep learning model, wherein the water absorption profile of an inflow well refers to the distribution of water absorption capacity at different depths underground, which is a key oilfield development parameter used to evaluate the water absorption efficiency of an oil well and optimize water injection strategies. The water absorption profile data of an inflow well refers to specific numerical information used to describe the distribution of such water absorption capacity, which may come from underground sensors or other measuring equipment.

[0026] Specifically, the method first needs to obtain the water absorption profile data to be measured, which usually involves deploying sensors on site in the oil field to collect fluid flow data at different depths. The data type may include parameters such as pressure, flow, and temperature, which can reflect the dynamic behavior of the downhole fluid.

[0027] More specifically, when determining the data type, a classifier based on statistics or machine learning algorithms can be set, which can determine whether the data is clear data or noisy data based on the characteristic values ​​of the data. The characteristic values ​​can be statistical parameters of the data, such as mean, variance or peak, or high-level features extracted by a deep learning model.

[0028] Preferably, for data marked as noise, a deep learning model can be used for denoising. In this process, a convolutional neural network (CNN) can be used to extract feature images, and an autoencoder or a generative adversarial network (GAN) can be used to generate denoised data. Specific parameter settings may include the number of network layers, the number of neurons, the type of activation function, etc.

[0029] Furthermore, multi-scale analysis methods, such as wavelet transform, can be used to process feature images of different scales to improve the denoising effect. When fusing denoised data, different fusion strategies, such as averaging, weighted averaging, or model-based fusion, can be used to obtain more accurate denoising results. These steps can be further refined to adapt to specific oilfield conditions and data characteristics.

[0030] In some embodiments, determining the data type of the water absorption profile to be measured includes: Converting the water absorption profile data to be measured into the well into a feature space to obtain a feature value of the water absorption profile data to be measured; If the characteristic value exceeds a preset characteristic value threshold, the data type of the water absorption profile to be measured is clear data; If the characteristic value does not exceed a preset characteristic value threshold, the data type of the water absorption profile to be measured is noise data.

[0031] It should be noted that this method determines the data type by converting the water absorption profile data of the well to be measured into the feature space and obtaining the eigenvalues ​​of these data. Here, the feature space refers to a space composed of multiple features, which can represent the key information of the original data, and the eigenvalues ​​are the specific values ​​of these features, which can be used to distinguish different attributes or states of the data.

[0032] Specifically, when implementing this step, we first need to determine which features can effectively represent the water absorption profile data and distinguish between clear data and noise data. These features may include statistical features (such as mean, variance), frequency domain features (such as power spectral density), time domain features (such as autocorrelation), etc.

[0033] Furthermore, the original data is mapped to the feature space through feature extraction algorithms, such as principal component analysis (PCA) or autoencoders, and the eigenvalues ​​are calculated. Next, a preset eigenvalue threshold needs to be set. When the eigenvalue exceeds this threshold, the data is judged as clear data; otherwise, it is noise data. The threshold setting can be based on statistical analysis of historical data, or the optimal value can be determined by cross-validation.

[0034] Preferably, the feature extraction process can be further refined, for example, a deep convolutional neural network (CNN) can be used to automatically learn feature representations instead of manually selecting features. In addition, the calculation of feature values ​​can be combined with multiple feature extraction techniques to improve the accuracy of classification.

[0035] Furthermore, when setting the threshold, a dynamic adjustment strategy can be used to automatically adjust the threshold according to the characteristics of real-time data to adapt to different measurement environments and conditions. In addition, it is also possible to consider introducing anomaly detection algorithms in machine learning, such as Isolation Forest or One-Class SVM, to more intelligently identify and process noisy data.

[0036] In some embodiments, the obtaining of denoised data corresponding to the noise water absorption profile data includes: Decomposing the noise water absorption profile data in multiple feature dimensions respectively to obtain feature images of corresponding dimensions; De-noising is performed on the feature images of each dimension respectively to obtain the de-noised data corresponding to the feature images of each dimension; The denoised data of the characteristic images of all dimensions are fused to obtain the denoised data corresponding to the noise water absorption profile data.

[0037] It should be noted that the method involves obtaining denoised data corresponding to the noisy water absorption profile data, wherein the denoised data refers to useful information extracted from the noise data after denoising. This step is for those water absorption profile data that are determined to be noise data, and the purpose is to reduce or eliminate the impact of noise on the measurement results.

[0038] Specifically, when implementing this step, the noise water absorption profile data needs to be decomposed in multiple feature dimensions. This can be achieved through wavelet transform, Fourier transform or other multi-scale analysis methods. After decomposition, a feature image will be obtained in each dimension, and these feature images contain the information of the original data in this dimension.

[0039] Furthermore, the feature image of each dimension is denoised, which can be done through a deep learning model, such as using a convolutional neural network (CNN) to process the feature image to identify and remove noise. After denoising, the denoised data of the feature images of all dimensions need to be fused to reconstruct the complete denoised data. Fusion can be achieved through simple averaging, weighted averaging, or more complex fusion strategies, such as a fusion network based on deep learning.

[0040] Preferably, the denoising process can adopt a variety of deep learning architectures, such as using CNNs of different depths and widths, or combining attention mechanisms to improve the denoising effect. When decomposing the feature image, different decomposition layers and scales can be selected to adapt to different types of noise features.

[0041] Furthermore, the fusion strategy of denoised data can also be optimized according to the characteristics of the data. For example, a fusion method based on a physical model can be introduced to utilize the physical characteristics of the reservoir to guide the data fusion process.

[0042] In some embodiments, obtaining denoised data corresponding to the feature image of each dimension includes: Acquire deep learning filters of multiple scales, and respectively use the deep learning filters of multiple scales to convolve the feature image to obtain convolution results of corresponding scales; Performing weighted averaging on the convolution results of multiple scales to obtain a weighted average result of the feature image; Based on the feature image and its weighted average result, denoising data of the feature image of the corresponding dimension is obtained.

[0043] It should be noted that this method describes in detail how to obtain the denoised data corresponding to the feature image of each dimension. This involves convolving the feature image with deep learning filters of multiple scales, and then taking a weighted average of these convolution results to obtain the denoised data. Here, deep learning filters refer to filters designed under the deep learning framework, which can capture complex patterns in the data.

[0044] Specifically, when implementing this step, we first need to determine the number of scales of the deep learning filter and the parameter settings for each scale. For example, we can choose three scales of filters, each scale corresponding to a different receptive field size to capture features at different levels.

[0045] Furthermore, these filters are used to perform convolution operations on the feature images of each dimension. The convolution operation can be regarded as the process of sliding the filter on the image and calculating the weighted sum. After the convolution is completed, the convolution results of each scale will be obtained. Next, these convolution results need to be weighted averaged. The choice of weights can be determined based on experimental results or optimization algorithms to ensure that the final denoised data can retain useful information and remove noise to the greatest extent.

[0046] Preferably, the convolution operation can use different kernel functions, such as Gaussian kernel, Laplacian kernel or other custom kernel functions, to adapt to different types of feature images. In the process of weighted averaging, the weight coefficient can be optimized by cross-validation to ensure the best denoising effect.

[0047] Furthermore, we can also consider introducing regularization terms to prevent overfitting, or use batch normalization to speed up the training process. In practice, we can also explore different network architectures, such as residual networks or densely connected networks, to improve denoising performance.

[0048] In some embodiments, the convolution result of the th scale is:

[0049] in, Indicates that the horizontal coordinate in the feature image is , the vertical axis is The pixel value at the location of Indicates Deep learning filters at different scales; Take 1, 2, 3; in,

[0050] in, Indicates The scale parameters of deep learning filters of different scales.

[0051] It should be noted that this sentence describes how to use deep learning filters of different scales to perform convolution operations on feature images, and gives a specific mathematical expression for the convolution operation. Here, convolution is a mathematical operation used to extract features from an image, and the scale parameter refers to the size or range of the filter, which determines the receptive field of the filter on the image.

[0052] Specifically, when implementing this step, we first need to determine the scale parameters of the deep learning filter. For example, we can set three filters of different scales, each of which corresponds to different scale parameters σ1, σ2, and σ3. These parameters control the smoothness of the filter. The larger the scale parameter, the stronger the smoothing effect of the filter. Then, these filters are used to perform convolution operations on the feature image. The convolution operation can be implemented by the following mathematical expression:

[0053] in, Represents the pixel value at the position with horizontal coordinate u and vertical coordinate v in the feature image, Represents the deep learning filter of the i-th scale, i is 1, 2, 3.

[0054] Preferably, the specific form of the deep learning filter can be a Gaussian function, whose mathematical expression is:

[0055] in, Indicates the scale parameter. In practical applications, different filter forms can be selected according to the characteristics of the feature image, such as Laplace filter, Sobel filter, etc.

[0056] Furthermore, the convolution operation can be implemented through the convolution layer in the deep learning framework, or the parameters of the filter can be automatically learned through the optimization algorithm.

[0057] In some embodiments, the weighted average result of the feature image is expressed as:

[0058] in, Indicates The weight coefficient of each scale, .

[0059] It should be noted that this sentence describes how to combine the convolution results of different scales by weighted averaging to obtain the weighted average result of the feature image. Here, the weighted average result refers to the comprehensive result calculated after giving different weights to each scale according to its importance, and the weight coefficient is used to indicate the proportion of each scale in the final result.

[0060] Specifically, when implementing this step, it is necessary to determine the weight coefficient of each scale. For example, weight coefficients α, β, and γ can be set for the convolution results of the three scales respectively. These weight coefficients need to satisfy the condition of α+β+γ=1. The setting of the weight coefficient can be determined based on experimental results or through an optimization algorithm to ensure that the final weighted average result can reflect the essential characteristics of the feature image to the greatest extent. The calculation of weighted average can be achieved by the following mathematical expression:

[0061] in, , and Represent the convolution results of three different scales respectively.

[0062] Preferably, the weight coefficients may be determined by a variety of methods, for example, the best weight coefficients may be selected by cross-validation, or an optimization algorithm such as a genetic algorithm may be used to automatically find the best weight combination.

[0063] Furthermore, in the process of weighted averaging, a regularization term can be introduced to prevent overfitting, or an adaptive weight adjustment strategy can be used to dynamically adjust the weight coefficient according to the content of the feature image.

[0064] In some embodiments, the denoised data of the feature image of the corresponding dimension is represented as:

[0065] It should be noted that this sentence describes how to combine the weighted average result of the feature image with the original feature image to obtain the denoised data of the feature image of the corresponding dimension. Here, denoised data refers to the data after the noise interference is removed from the original data through processing, and the feature image refers to the image represented in the feature space, which contains the important information of the original water absorption profile data.

[0066] Specifically, when implementing this step, it is necessary to multiply the weighted average result with the original feature image to obtain denoised data. This process can be achieved through the following mathematical expression:

[0067] in, represents the weighted average result of the feature image, represents the original feature image, and Represents the final denoised data. In this process, the weighted average result As a weight factor, with the original feature image Multiply to enhance the useful signal and suppress the noise.

[0068] Preferably, the multiplication operation can be adjusted according to the specific characteristics of the feature image. For example, a threshold can be introduced, and only when the weighted average result exceeds the threshold, it is multiplied with the original feature image to further suppress noise.

[0069] Furthermore, nonlinear activation functions, such as ReLU (Rectified Linear Unit), can also be used to enhance the denoising effect. In practical applications, different fusion strategies can also be considered, such as a fusion network based on deep learning, to automatically learn the best denoising strategy.

[0070] In some embodiments, the water absorption profile measurement is performed using a trained deep learning model; The deep learning model is implemented based on a neural network model or a deep learning network model.

[0071] It should be noted that this sentence describes the process of using a trained deep learning model to measure water absorption profiles. Here, a deep learning model refers to a complex mathematical model trained with a large amount of data, which can learn and simulate the relationship between water absorption profile data and measurement results, while a neural network model or deep learning network model refers to the specific architecture that constitutes a deep learning model. They are composed of multiple layers of neurons and can process and learn complex nonlinear relationships.

[0072] Specifically, when implementing this step, you first need to build a deep learning model, which can be based on a convolutional neural network (CNN), a recurrent neural network (RNN), or other types of neural network models. The construction of the model includes determining the number of layers in the network, the number of neurons in each layer, the type of activation function, etc.

[0073] Furthermore, a large amount of water absorption profile data and corresponding measurement results need to be collected, which will be used to train the model. During the training process, the parameters of the model will be continuously adjusted to minimize the difference between the predicted results and the actual measured results. After the training is completed, the model will be able to measure new water absorption profile data and give predicted measurement results.

[0074] Preferably, the training of deep learning models can use a variety of optimization algorithms, such as stochastic gradient descent (SGD), Adam optimizer, etc., to accelerate the training process and improve the performance of the model. In addition, regularization techniques, such as dropout or weight decay, can be introduced to prevent the model from overfitting. In the model evaluation stage, methods such as cross-validation can be used to evaluate the generalization ability of the model.

[0075] More specifically, you can also consider using transfer learning techniques to use models pre-trained on other related tasks as a starting point to improve training efficiency and model performance. These refined steps and alternatives can further improve the accuracy and robustness of the model.

[0076] In some embodiments, the deep learning model is trained by performing the following operations: Acquire a training set; the training set includes training samples and their measurement results; the training samples include clear water absorption profile data samples and noisy water absorption profile data samples; For a noise water absorption profile data sample, obtaining denoised data corresponding to the noise water absorption profile data sample; The denoised data corresponding to each noise water absorption profile data sample and the measurement results corresponding to the corresponding noise water absorption profile data sample are sequentially input into the deep learning model for training; Each clear water absorption profile data sample and its measurement results are sequentially input into the deep learning model for training; After the training end conditions are met, the trained deep learning model is obtained.

[0077] It should be noted that this sentence describes the specific steps of training a deep learning model. In this process, the training set refers to the data set used to train the model, which contains multiple training samples and their corresponding measurement results. The training sample refers to a single data instance in the training set, which can be a clear water absorption profile data sample or a noisy water absorption profile data sample. For the noisy water absorption profile data sample, its corresponding denoised data needs to be obtained before it can be used for training.

[0078] Specifically, when implementing this step, you first need to collect enough water absorption profile data, which should include clear data and noisy data to ensure that the model can handle different types of situations. For the noisy data, you need to denoise it first to extract useful information. The denoised data and the original clear data will together constitute the training set.

[0079] Further, this data and its corresponding measurement results are fed into a deep learning model for training. During the training process, the model's weights and biases are continuously updated to minimize the prediction error. This process usually involves forward propagation and backpropagation algorithms, where forward propagation is used to generate prediction results and backpropagation is used to adjust the model parameters based on the error.

[0080] Preferably, different strategies can be used to optimize model performance during training. For example, different learning rates can be set to control the speed of model parameter update. Early stopping can also be used to prevent overfitting, that is, training is stopped when the performance on the validation set no longer improves. In addition, data augmentation techniques can be used to expand the training set and improve the generalization ability of the model.

[0081] More specifically, you can also consider using different network architectures, such as residual networks (ResNet) or densely connected networks (DenseNet), to improve the learning ability of the model.

[0082] The above-mentioned embodiments of the present invention have the following beneficial effects: The method for measuring the water absorption profile of the well based on the deep learning model of the present invention can realize the efficient processing and accurate measurement of the downhole water absorption profile data. By automatically judging the data type and performing corresponding clear data measurement or noise data processing, the method can significantly improve the efficiency and accuracy of the measurement process. In particular, for noisy data, the method can effectively extract useful water absorption profile information from the noise through feature decomposition, multi-dimensional denoising and data fusion technology, thereby improving the reliability of the measurement results.

[0083] In addition, this method can further improve the quality of denoised data by using deep learning filters at multiple scales and weighted averaging techniques. This method can not only process feature images of different scales, but also optimize the denoising process through weighted averaging results, making the final water absorption profile measurement results more accurate. In this way, this method can be widely used in various complex reservoir conditions and provide strong technical support for oilfield development.

[0084] like Figure 2 As shown, in some embodiments, a system 200 for measuring a water absorption profile of a well based on a deep learning model comprises: The data acquisition module 201 is used to acquire the water absorption profile data to be measured. A data type determination module 202 is used to determine the data type of the water absorption profile to be measured, and the data type is clear data or noise data; The clear data measurement module 203 is used to mark the water absorption profile data to be measured as clear water absorption profile data when the data type of the water absorption profile to be measured is clear data, and directly perform water absorption profile measurement on the clear water absorption profile data to obtain a measurement result; The noise data measurement module 204 is used to mark the water absorption profile data to be measured as noise water absorption profile data when the data type of the water absorption profile to be measured is noise data; obtain the denoised data corresponding to the noise water absorption profile data, and perform water absorption profile measurement on the denoised data corresponding to the noise water absorption profile data to obtain a measurement result.

[0085] It can be understood that the modules recorded in the deep learning model-based well water absorption profile measurement system 200 are similar to the reference Figure 1 Therefore, the operations, features and beneficial effects described above for the method for measuring the water absorption profile of the well based on the deep learning model are also applicable to the system 200 for measuring the water absorption profile of the well based on the deep learning model and the modules contained therein, and will not be repeated here.

[0086] Reference below Figure 3 , which shows a schematic diagram of a structure 300 of an electronic device suitable for implementing some embodiments of the present invention. The electronic devices in some embodiments of the present invention may include, but are not limited to, mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 3 The terminal device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0087] like Figure 3 As shown, the electronic device 300 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage device 308 to a random access memory (RAM) 303. In the RAM 303, various programs and data required for the operation of the electronic device 300 are also stored. The processing device 301, the ROM 302, and the RAM 303 are connected to each other via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0088] Typically, the following devices may be connected to the I / O interface 305: an input device 306 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 307 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 308 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 309. The communication device 309 may allow the electronic device 300 to communicate with other devices wirelessly or by wire to exchange data. Figure 3 The electronic device 300 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead. Figure 3 Each block shown in the figure may represent one device, or may represent multiple devices as required.

[0089] Furthermore, the storage medium of the embodiment of the present application stores program instructions that can implement all the above methods, wherein the program instructions can be stored in the above storage medium in the form of a software product, including several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) or a processor to execute all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, or terminal devices such as a computer, a server, a mobile phone, and a tablet.

[0090] The above descriptions are only some preferred embodiments of the present invention and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present invention is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above inventive concept. For example, the above features are replaced with (but not limited to) technical features with similar functions disclosed in the embodiments of the present invention.

Claims

1. A method for measuring water absorption profile of a well based on a deep learning model, characterized in that: include: Acquire the water absorption profile data of the well to be measured, and determine the data type of the water absorption profile of the well to be measured, wherein the data type is clear data or noise data; If the data type of the water absorption profile to be measured is clear data, the water absorption profile data to be measured is marked as clear water absorption profile data, and the water absorption profile is directly measured on the clear water absorption profile data to obtain a measurement result; If the data type of the water absorption profile into the well to be measured is noise data, the water absorption profile data into the well to be measured is marked as noise water absorption profile data; the denoised data corresponding to the noise water absorption profile data is obtained, and the water absorption profile is measured on the denoised data corresponding to the noise water absorption profile data to obtain a measurement result.

2. The method for measuring water absorption profile of a well based on a deep learning model according to claim 1, characterized in that: The determining of the data type of the water absorption profile to be measured includes: Converting the water absorption profile data to be measured into the well into a feature space to obtain a feature value of the water absorption profile data to be measured; If the characteristic value exceeds a preset characteristic value threshold, the data type of the water absorption profile to be measured is clear data; If the characteristic value does not exceed a preset characteristic value threshold, the data type of the water absorption profile to be measured is noise data.

3. The method for measuring water absorption profile of a well based on a deep learning model according to claim 1, characterized in that: The step of obtaining denoised data corresponding to the noise water absorption profile data comprises: Decomposing the noise water absorption profile data in multiple feature dimensions respectively to obtain feature images of corresponding dimensions; De-noising is performed on the feature images of each dimension respectively to obtain the de-noised data corresponding to the feature images of each dimension; The denoised data of the characteristic images of all dimensions are fused to obtain the denoised data corresponding to the noise water absorption profile data.

4. The method for measuring water absorption profile of a well based on a deep learning model according to claim 3, characterized in that: The step of obtaining denoised data corresponding to the feature image of each dimension includes: Acquire deep learning filters of multiple scales, and respectively use the deep learning filters of multiple scales to convolve the feature image to obtain convolution results of corresponding scales; Performing weighted averaging on the convolution results of multiple scales to obtain a weighted average result of the feature image; Based on the feature image and its weighted average result, denoising data of the feature image of the corresponding dimension is obtained.

5. The method for measuring water absorption profile of a well based on a deep learning model according to claim 4, characterized in that: The convolution result of the scale is: in, Indicates that the horizontal coordinate in the feature image is , the vertical axis is The pixel value at the location of Indicates Deep learning filters at different scales; Take 1, 2, 3; in, in, Indicates The scale parameters of deep learning filters of different scales.

6. The method for measuring water absorption profile of a well based on a deep learning model according to claim 5, characterized in that: The weighted average result of the feature image is expressed as: in, Indicates The weight coefficient of each scale, .

7. The method for measuring water absorption profile of a well based on a deep learning model according to claim 6, characterized in that: The denoised data of the feature image of the corresponding dimension is expressed as: 。 8. The method for measuring water absorption profile of a well based on a deep learning model according to any one of claims 1 to 7, characterized in that: Water absorption profile measurement using the trained deep learning model; The deep learning model is implemented based on a neural network model or a deep learning network model.

9. The method for measuring water absorption profile of a well based on a deep learning model according to claim 8, characterized in that: Train the deep learning model by doing the following: Acquire a training set; the training set includes training samples and their measurement results; the training samples include clear water absorption profile data samples and noisy water absorption profile data samples; For a noise water absorption profile data sample, obtaining denoised data corresponding to the noise water absorption profile data sample; The denoised data corresponding to each noise water absorption profile data sample and the measurement results corresponding to the corresponding noise water absorption profile data sample are sequentially input into the deep learning model for training; Each clear water absorption profile data sample and its measurement results are sequentially input into the deep learning model for training; After the training end conditions are met, the trained deep learning model is obtained.

10. A system for measuring water absorption profile of a well based on a deep learning model, characterized in that: include: A data acquisition module, used to acquire the water absorption profile data of the well to be measured; A data type determination module, used to determine the data type of the water absorption profile to be measured, wherein the data type is clear data or noise data; A clear data measurement module, for marking the water absorption profile data to be measured as clear water absorption profile data when the data type of the water absorption profile to be measured is clear data, and directly performing water absorption profile measurement on the clear water absorption profile data to obtain a measurement result; The noise data measurement module is used to mark the water absorption profile data to be measured as noise water absorption profile data when the data type of the water absorption profile to be measured is noise data; obtain the denoised data corresponding to the noise water absorption profile data, and perform water absorption profile measurement on the denoised data corresponding to the noise water absorption profile data to obtain a measurement result.

Images