Response characteristic calibration method, device and equipment of logging equipment and storage medium

Abstract

The invention relates to a response characteristic calibration method, device and equipment of logging equipment and a storage medium. The method comprises the following steps: carrying out two-parameter response modeling on key parameters of logging equipment, and constructing a pressure-fluidity coupling response model; data acquisition parameters and signal filtering parameters of logging equipment are configured in a partitioned manner, and a multi-layer response parameter set is established. And processing the logging pressure and fluidity data sequence to obtain a logging data deviation characteristic spectrum and a response anomaly type matrix. And performing multi-target parameter optimization calculation on different layers of response parameter sets to generate a layered calibration parameter sequence. The layered calibration parameter sequence is analyzed, response benchmark analysis, response gain analysis and response compensation analysis are achieved, and a multi-layer response calibration instruction is obtained. And the calibration instruction is subjected to a comparison test under three working conditions, and a pressure-fluidity two-parameter calibration curve is generated. According to the method, accurate calibration of the logging equipment under various working conditions is realized, and the reliability of a calibration result is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of well logging equipment, and in particular to a response characteristic calibration method, device, equipment and storage medium of well logging equipment. Background Art

[0002] In the process of oil and gas exploration, formation pressure and mobility parameters are of great significance to reservoir evaluation. At present, these two key parameters are mainly recorded in the open hole section of the well through logging equipment. However, due to differences in design parameters such as the sealing area and suction method of logging equipment from different manufacturers, there are large differences in the measurement data obtained in the same wellbore and the same test point. This difference is particularly prominent in dense formations. The mobility values ​​read by different types of logging equipment may differ by several to dozens of times, which seriously affects the accuracy of reservoir evaluation.

[0003] Some formation pressure simulation systems have been developed in the prior art. By simulating the formations and formation pressures of cores of different lithologies indoors, ground tests and simulation tests can be performed during the development of logging equipment. However, these systems have obvious technical limitations. They can only verify the data accuracy and reliability of the formation pressure parameters measured by the logging equipment, but cannot verify the core fluidity parameters at the same time, which makes it difficult to ensure the comprehensive reliability of the logging data. In addition, most of the existing calibration methods adopt a single parameter calibration mode, lacking in-depth consideration of the coupling relationship between pressure and fluidity parameters. Under complex formation conditions, the response characteristics of logging equipment often show nonlinear and time-varying characteristics, and traditional linear calibration methods are difficult to meet the needs of high-precision measurements. At the same time, the existing calibration system lacks the ability to partition different formation types and cannot perform targeted parameter optimization for different formation characteristics such as dense layers and conventional layers. Summary of the invention

[0004] The main purpose of the present invention is to provide a method, device, equipment and storage medium for calibrating the response characteristics of logging equipment. The present invention realizes accurate calibration of the logging equipment under various working conditions and ensures the reliability of the calibration results.

[0005] To achieve the above object, the present invention provides a method for calibrating the response characteristics of a well logging device, comprising the following steps: The double-parameter response modeling of the logging equipment's sealing area coefficient, suction rate coefficient and pressure measurement range was carried out to obtain the pressure-fluidity coupling response model. According to the pressure-fluidity coupling response model, data acquisition parameters and signal filtering parameters of the logging equipment are configured in different intervals to obtain a dense layer response parameter set, a conventional layer response parameter set and a mixed layer response parameter set; The logging pressure data sequence and the mobility data sequence are input into a dual-channel deep neural network for dynamic feature extraction and response mode classification to obtain the logging data deviation characteristic spectrum and response anomaly type matrix; According to the logging data deviation characteristic spectrum and the response anomaly type matrix, a multi-objective parameter optimization calculation is performed on the dense layer response parameter set, the conventional layer response parameter set and the mixed layer response parameter set to obtain a hierarchical calibration parameter sequence; Based on the layered calibration parameter sequence, a response benchmark analysis is performed in sequence through a formation permeability evaluation matrix, a response gain analysis is performed through a pressure gradient evaluation matrix, and a response compensation analysis is performed through a mobility attenuation evaluation matrix to obtain a multi-layer response calibration instruction; The multi-layer response calibration instructions are subjected to comparative tests under standard permeability conditions, variable pressure permeability conditions and composite permeability conditions to obtain a pressure-fluidity dual-parameter calibration curve.

[0006] The present invention also provides a response characteristic calibration device for a well logging device, comprising: A modeling module is used to perform dual-parameter response modeling on the setting area coefficient, suction rate coefficient and pressure measurement range of the logging equipment to obtain a pressure-fluidity coupling response model; A configuration module, for performing interval configuration of data acquisition parameters and signal filtering parameters of the logging equipment according to the pressure-fluidity coupling response model, to obtain a dense layer response parameter set, a conventional layer response parameter set, and a mixed layer response parameter set; A classification module is used to input the logging pressure data sequence and the mobility data sequence into a dual-channel deep neural network for dynamic feature extraction and response mode classification to obtain the logging data deviation characteristic spectrum and response anomaly type matrix; A calculation module, configured to perform multi-objective parameter optimization calculation on the dense layer response parameter set, the conventional layer response parameter set and the mixed layer response parameter set according to the logging data deviation characteristic spectrum and the response anomaly type matrix, to obtain a hierarchical calibration parameter sequence; An analysis module, for performing response benchmark analysis through a formation permeability evaluation matrix, response gain analysis through a pressure gradient evaluation matrix, and response compensation analysis through a mobility attenuation evaluation matrix based on the layered calibration parameter sequence, to obtain a multi-layer response calibration instruction; The test module is used to perform comparative tests on the multi-layer response calibration instructions under standard permeability conditions, variable pressure permeability conditions and composite permeability conditions to obtain a pressure-fluidity dual-parameter calibration curve.

[0007] The present invention also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the above methods when executing the computer program.

[0008] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of any of the above-mentioned methods are implemented.

[0009] In summary, the technical solution provided by the present invention achieves an accurate description of the pressure-fluidity coupling response by establishing a dual-parameter response modeling mechanism; adopts a multi-layer response parameter partition configuration strategy to solve the problem of differentiated calibration under different formation conditions; introduces a dual-channel deep neural network for feature extraction and anomaly recognition, which improves the intelligence level of the calibration process; designs a multi-objective parameter optimization algorithm to achieve accurate calibration of logging equipment under various working conditions; and ensures the reliability of the calibration results through working condition test verification of multi-layer response calibration instructions. The present invention establishes a complete set of logging equipment response characteristic calibration system, which effectively solves the problems of single logging equipment calibration, low precision and poor adaptability in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 is a schematic diagram of the steps of a method for calibrating the response characteristics of a well logging device in one embodiment of the present invention; Figure 2 is a structural block diagram of a response characteristic calibration device for a well logging device in one embodiment of the present invention; Figure 3 It is a schematic block diagram of the structure of a computer device according to an embodiment of the present invention.

[0011] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0012] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0013] Reference Figure 1 This embodiment provides a method for calibrating the response characteristics of a well logging device, comprising the following steps: S1, dual-parameter response modeling is performed on the setting area coefficient, suction rate coefficient and pressure measurement range of the logging equipment to obtain the pressure-fluidity coupling response model; Among them, the setting area coefficient and the suction rate coefficient of the logging equipment are linearly mapped and transformed to obtain the initial response coefficient matrix of the logging equipment, which describes the initial response characteristics of the logging equipment under different physical environments. Based on the initial response coefficient matrix of the logging equipment, the pressure measurement range is piecewise linearized to obtain the pressure response function of the logging equipment. The piecewise linearization method helps to reduce the dependence on complex nonlinear models and simplify the calculation difficulty of the response function while maintaining a certain accuracy and adaptability. In this process, the pressure response function is divided into multiple segmented linear expressions, each segment represents the response characteristics of the equipment under different pressure ranges. The initial response coefficient matrix and the pressure response function are input into the second-order differential equation solver for solution to obtain the fluidity response function of the logging equipment. The fluidity response function describes the response characteristics of the logging equipment under different flow conditions. The second-order differential equation solver can capture the dynamic response of the equipment during the change of pressure and flow by solving the dynamic equations involving pressure and flow. According to the pressure response function and fluidity response function of the logging equipment, the response characteristics of the logging equipment are nonlinearly fitted and calculated to obtain the dual-parameter response characteristic curve. Nonlinear fitting can effectively reduce the deviation caused by complex physical conditions, and make the coupling relationship between pressure and flow get the best expression, forming an accurate dual-parameter response characteristic curve, revealing the mutual dependence and mutual influence of pressure and flow. The dual-parameter response characteristic curve is input into the adaptive filter for signal filtering processing to remove high-frequency noise and unexpected random fluctuations in the curve to improve the smoothness and accuracy of the response curve. After signal filtering, the correlation matrix of pressure-fluidity response is obtained. The singular value decomposition operation is performed on the pressure-fluidity response correlation matrix to obtain the response eigenvector of the logging equipment. Based on the response eigenvector, the least squares fitting operation is performed on the response parameters of the logging equipment to obtain the coupling coefficient matrix of the logging equipment. The least squares fitting is used to minimize the fitting error so that the response model can more accurately describe the characteristics of the logging equipment. The coupling coefficient matrix is ​​one of the important results in the dual-parameter response model, which characterizes the degree of coupling between pressure and flow. The coupling coefficient matrix is ​​substituted into the state space equation for numerical solution to construct the pressure-fluidity coupling response model. The state space equation is used to describe the evolution law of the system in time. After substituting into the coupling coefficient matrix, the coupling response results of the system under different states are obtained, forming a complete pressure-fluidity coupling response model.

[0014] S2, according to the pressure-fluidity coupling response model, the data acquisition parameters and signal filtering parameters of the logging equipment are configured in different intervals to obtain a dense layer response parameter set, a conventional layer response parameter set and a mixed layer response parameter set; Specifically, the sampling frequency range of the pressure-fluidity coupling response model is segmented and quantified to obtain the dynamic sampling matrix of the logging equipment. The dynamic characteristics of the pressure and flow coupling response model are accurately quantified to capture the sampling behavior of the logging equipment under different formation conditions and ensure that it has sufficient resolution to reflect the actual logging needs. When performing response analysis based on the dynamic sampling matrix, the response sensitivity of the sealing area coefficient is calculated to obtain the sensitivity partition sequence of the logging equipment, which effectively describes the response change characteristics of the equipment under different formation conditions, especially the change trend of sensitivity. The sensitivity partition sequence of the logging equipment is input into the formation classifier for permeability characteristic analysis to obtain the formation classification matrix of the logging equipment. The formation classification matrix contains the permeability characteristics of each sampling point, which are used to divide the formation into a dense layer, a conventional layer, and a mixed layer. Based on this classification result, signal processing is performed for different types of formations, and the logging signal is processed by Gaussian filtering operation to obtain a multi-layer filtering parameter sequence. Gaussian filtering is a signal processing method that effectively smooths the logging data and removes high-frequency noise, ensuring that the logging signal has high quality in subsequent processing. The multi-layer filter parameter sequence is input into the adaptive quantizer for parameter allocation, and the partition parameter matrix of each layer is generated. The role of the partition parameter matrix is ​​to make a detailed distinction between the response parameters of different formations, so that the response parameters of the logging equipment of each formation type can be optimally configured. The multi-layer response compensation coefficient is obtained by performing fluid response compensation calculation on the partition parameter matrix of each layer. The response deviation caused by flow changes in different formations is corrected to ensure the response consistency and accuracy of the logging equipment under the formation conditions of each layer. Based on the multi-layer response compensation coefficient, the logging data is subjected to layered response correction calculation to obtain the layered response correction matrix to ensure that the response results of the logging equipment under different formation types have high accuracy and consistency. The layered response correction matrix is ​​substituted into the response characteristic allocation model for parameter configuration to obtain the dense layer response parameter set, the conventional layer response parameter set and the mixed layer response parameter set. The response characteristic allocation model is a multi-layer allocation algorithm based on formation type and measurement parameters. Its core is to establish response characteristic mapping relationships for dense layers, conventional layers and mixed layers respectively, and realize the optimal allocation of response parameters of logging equipment under different formation conditions through the layered adaptive weight configuration mechanism. This multi-layer allocation algorithm can dynamically adjust the weights of the response parameters according to the characteristics of different formations and the actual measurement requirements of the equipment to ensure the best logging accuracy under each formation condition. The dense layer requires a higher pressure response sensitivity due to its low permeability, while the conventional layer requires the equipment to maintain a balanced response under relatively stable flow conditions, and the mixed layer needs to take into account the dual characteristics of pressure and flow due to the complexity of the formation characteristics. Through the above steps, a hierarchical adaptive weight configuration mechanism is adopted to effectively optimize the response characteristics of the logging equipment under different formation conditions, and achieve more accurate and reliable logging data acquisition.

[0015] S3, inputting the logging pressure data sequence and the mobility data sequence into a dual-channel deep neural network for dynamic feature extraction and response mode classification, and obtaining a logging data deviation characteristic spectrum and a response anomaly type matrix; It should be noted that the logging pressure data sequence and the mobility data sequence are segmented by the time sliding window respectively, and the original logging data is split into several smaller time segments to form a pressure data subsequence group and a mobility data subsequence group. The role of the time sliding window is to capture the characteristics of short-term dynamic changes in the time series so that the subsequent deep network can more finely extract the characteristics in each small period of time. The pressure data subsequence group is input into the first LSTM encoder for processing. The first LSTM encoder consists of three layers of bidirectional LSTM units, and each layer of LSTM units contains 128 hidden state nodes. The use of bidirectional LSTM effectively captures the dynamic characteristics of the logging pressure data on the forward and reverse time axes, so that the model can use the past data state and enhance the understanding of the current state based on future data. After three layers of bidirectional LSTM encoding, the obtained pressure feature encoding vector reflects the complex dynamic change characteristics in the logging pressure data. Similarly, the mobility data subsequence group is input into the second LSTM encoder. The second LSTM encoder contains three layers of bidirectional LSTM units, and each layer of units also contains 128 hidden state nodes. Through the processing of the second LSTM encoder, the fluidity feature encoding vector is obtained, which effectively captures the dynamic change characteristics of the well logging fluidity data. The attention mechanism fusion operation is performed on the pressure feature encoding vector and the fluidity feature encoding vector. The attention mechanism adopts a multi-head self-attention structure, in which the number of attention heads is 8 and the dimension of each attention head is 64. When fusing pressure features and fluidity features, the multi-head attention mechanism enables the model to focus on different feature dimensions and time segments, effectively improving the expression ability of feature fusion, and obtaining a multimodal feature fusion matrix. The multimodal feature fusion matrix is ​​input into the feature extraction CNN network. The feature extraction CNN network contains four convolutional layers, and the number of convolution kernels in each layer is 64, 128, 256 and 512 respectively. The size of the convolution kernel is 3×3, and the ReLU activation function is used to introduce nonlinear characteristics. Through this CNN network, the model can extract complex features in the spatial and temporal dimensions of multimodal features. Each convolution layer gradually improves the abstract expression ability of the feature by increasing the number of convolution kernels, so that the final deep feature map can fully describe the complexity of the well logging data. The deep feature map is processed by global average pooling and maximum pooling to compress and summarize features from different angles. Global average pooling can summarize the overall trend of each feature channel, while maximum pooling can retain the most significant feature information. The two pooling results are concatenated and fused to form a feature vector sequence, so that the model can simultaneously utilize information from both overall trends and significant features. The feature vector sequence is input into the first fully connected network, which contains three layers, with 512, 256 and 128 neurons in each layer, respectively. A dropout rate of 0.5 is used to prevent the model from overfitting.Through the processing of this fully connected network, the model further nonlinearly maps the features and obtains the deviation characteristic spectrum of the logging data. The deviation characteristic spectrum is used to describe the abnormal characteristics and deviation information in the logging data, and is a quantitative description of the potential anomalies in the logging data. At the same time, the same feature vector sequence is input into the second fully connected network for processing. The fully connected network also contains three layers, with the number of neurons being 512, 256, and 7 respectively, and the last layer uses the Softmax activation function. The Softmax function can classify the response type and obtain the response anomaly type matrix. Each row in the response anomaly type matrix represents a possible anomaly type, and the values ​​of the matrix represent the probability distribution of different types of anomalies, thereby helping the model to identify and classify different anomaly patterns.

[0016] S4, according to the deviation characteristic spectrum of the logging data and the response anomaly type matrix, a multi-objective parameter optimization calculation is performed on the dense layer response parameter set, the conventional layer response parameter set and the mixed layer response parameter set to obtain a hierarchical calibration parameter sequence; Specifically, feature fusion operations are performed on the deviation characteristic spectrum of logging data and the response anomaly type matrix to obtain the comprehensive feature matrix of logging data, capturing the differences and abnormal patterns in the data. Based on the comprehensive feature matrix of logging data, the pressure response optimization calculation is performed on the dense layer response parameter set to obtain the pressure calibration parameter group of the dense layer. According to the characteristics of the dense layer, its pressure response parameters are optimized so that the equipment can accurately reflect the changes in formation pressure in the dense layer, which has poor permeability. Through targeted optimization, the accuracy and stability of the dense layer logging data are improved. Based on the dense layer pressure calibration parameter group, the fluidity compensation calculation is performed to obtain the fluidity calibration parameter group of the dense layer. Because the permeability of the dense layer is low, the change of fluidity will have a greater impact on the response characteristics, so it is necessary to compensate for the fluidity to ensure that the logging response under low permeability conditions also has a high accuracy. Based on the comprehensive feature matrix of logging data, the pressure gradient optimization operation is performed on the conventional layer response parameter set to obtain the pressure and fluidity calibration parameter group of the conventional layer. The formation properties of the conventional layer are relatively uniform, and its response characteristics depend more on the balance of pressure and fluidity. The pressure gradient needs to be optimized to ensure that the logging equipment in the conventional formation can correctly capture the impact of flow changes on pressure. At the same time, based on the comprehensive characteristic matrix of logging data, the mixed layer response parameter set is multi-parameter coupled optimization to obtain the pressure and fluidity calibration parameter group of the mixed layer. The formation characteristics of the mixed layer are relatively complex, including the characteristics of different types of formations, and its response characteristics are affected by both pressure and the coupling effect of fluidity. The goal of multi-parameter coupling optimization is to find the best balance between multiple parameters to ensure that the logging equipment can accurately reflect the actual formation conditions in the mixed layer. Through coupling optimization, the calibration parameter group of the mixed layer effectively describes its complex physical properties. The tight layer pressure calibration parameter group, tight layer fluidity calibration parameter group, conventional layer pressure and fluidity calibration parameter group, and mixed layer pressure and fluidity calibration parameter group are input into the multi-layer response optimizer for parameter coordination calculation to obtain a multi-layer calibration parameter matrix. The design purpose of the multi-layer response optimizer is to achieve the coordinated optimization of the response parameters of each formation, so as to ensure that the logging equipment can maintain the global optimal response characteristics under various complex formation conditions. The multi-layer response optimizer includes a multi-objective optimization algorithm processing unit, a parameter coordinated adjustment unit, and a response characteristic evaluation unit. Through the synergy of these units, the response parameters of multiple formations are optimized in parallel, so that the optimization results of each layer under multi-objective conditions can cooperate with each other to achieve the overall optimal effect. The inter-layer response compensation calculation of the multi-layer calibration parameter matrix is ​​performed to obtain the calibration compensation coefficient sequence. The response inconsistency between different formations is adjusted to ensure that the measurement results of each layer are consistent and comparable as a whole. The obtained calibration compensation coefficient sequence is substituted into the parameter mapping model for hierarchical configuration, and finally the hierarchical calibration parameter sequence is obtained.The parameter mapping model includes a nonlinear parameter mapping unit, a dynamic feature extraction unit and a parameter reconstruction unit. Its main function is to establish a nonlinear mapping relationship between the response parameters of the logging equipment and the calibration parameters, and to achieve accurate reconstruction of the response characteristics and parameter optimization configuration through dynamic feature extraction and parameter reconstruction.

[0017] S5, based on the layered calibration parameter sequence, performing response benchmark analysis through the formation permeability evaluation matrix, performing response gain analysis through the pressure gradient evaluation matrix, and performing response compensation analysis through the mobility attenuation evaluation matrix, to obtain a multi-layer response calibration instruction; Among them, the permeability analysis operation is performed on the dense layer response parameter set in the layered calibration parameter sequence to obtain the dense layer benchmark response sequence. Since the permeability of the dense layer is low, the response characteristics have great nonlinearity. Through the permeability analysis, the basic response characteristics of the dense layer are determined to form a benchmark reference sequence. Similarly, the permeability analysis operation is performed on the conventional layer response parameter set in the layered calibration parameter sequence to obtain the conventional layer benchmark response sequence. Since the conventional layer has relatively stable permeability, its response characteristics are usually more balanced. The purpose of the permeability analysis is to determine the standard response characteristics of the layer and provide a benchmark reference for the application of the equipment in conventional formations. The permeability analysis operation is performed on the mixed layer response parameter set to obtain the mixed layer benchmark response sequence. The mixed layer usually has the mixed characteristics of the dense layer and the conventional layer, and its permeability also shows variable characteristics. The permeability analysis can help reveal the response characteristics of different regions inside the mixed layer and generate a comprehensive benchmark response sequence. Based on the dense layer benchmark response sequence, the conventional layer benchmark response sequence and the mixed layer benchmark response sequence, the pressure measurement data is gradient calculated to obtain the pressure response gain matrix. The influence of pressure changes on the response of logging equipment under different formation conditions is evaluated. The pressure response gain matrix is ​​used to quantify this influence and is used for subsequent gain adjustment and correction operations. The pressure response gain matrix is ​​an important data reflecting the relationship between pressure and response. According to the pressure response gain matrix, the setting area coefficient is dynamically corrected and calculated to obtain the setting response correction coefficient. The significance of the setting response correction coefficient is to correct the setting behavior of the equipment so that it can adapt to the changes in different formations, especially in dense layers and mixed layers. The stability of the equipment and the accuracy of measurement are significantly improved through dynamic correction. Based on the pressure response gain matrix and the setting response correction coefficient, the fluidity compensation calculation is performed on the suction rate coefficient to obtain the fluidity attenuation correction coefficient. The purpose of fluidity attenuation correction is to ensure that the flow response of the equipment remains stable in the flow environment of different formations, especially in formations with drastic fluidity changes. The fluidity attenuation correction can effectively eliminate the measurement error caused by flow changes. The setting response correction coefficient and the fluidity attenuation correction coefficient are input into the parameter co-processor for joint optimization to obtain a multi-layer calibration control matrix. The design of the parameter co-processor includes a multi-channel parameter fusion unit, a dynamic weight allocation unit, and a parameter optimization unit. Through the collaborative work of these units, the joint optimization and dynamic adjustment of the pressure response parameters and the fluidity response parameters are realized. The function of the multi-channel parameter fusion unit is to effectively fuse different types of calibration parameters to capture the mutual influence and coupling relationship between the various parameters, while the dynamic weight allocation unit dynamically adjusts the weights of each parameter in the optimization process according to the formation conditions to ensure that the final optimization results can reflect the true response characteristics of each formation. The parameter optimization unit fine-tunes the fused parameters to obtain the optimal parameter configuration, so that the logging equipment can show good response characteristics under different formation conditions.The multi-layer calibration control matrix is ​​converted into instructions to obtain multi-layer response calibration instructions. These instructions are used to guide the response calibration operation of the logging equipment during actual operation to ensure that the equipment can be dynamically adjusted and optimized according to the characteristics of different formations. Through the comprehensive effects of multi-channel parameter fusion, multi-objective optimization and dynamic adjustment, the multi-layer response calibration instructions enable the logging equipment to obtain the global optimal response characteristics under complex formation conditions, greatly improving the accuracy and reliability of logging data.

[0018] S6, the multi-layer response calibration instruction is subjected to comparative tests under standard permeability conditions, variable pressure permeability conditions and composite permeability conditions to obtain a pressure-fluidity dual-parameter calibration curve.

[0019] Specifically, the multi-layer response calibration instruction is subjected to a standard working condition response test through the first permeability controller to obtain a first pressure response data group and a first fluidity response data group. Under the standard permeability working condition, by collecting and analyzing these data, the response characteristics of the equipment under ideal stable conditions are effectively captured. Based on the first pressure response data group and the first fluidity response data group, the first parameter calibration operation is performed on the sealing area coefficient of the logging equipment to obtain a standard working condition calibration parameter sequence. This calibration parameter sequence is used as a benchmark to define the optimal response state of the logging equipment under standard working conditions. The multi-layer response calibration instruction is subjected to a variable pressure working condition response test through the second permeability controller to obtain a second pressure response data group and a second fluidity response data group. The variable pressure working condition is usually used to simulate the rapid change of formation conditions to examine the response characteristics of the logging equipment under an unstable environment. When analyzing the second pressure and fluidity response data, the suction rate is corrected and calculated based on the previously obtained standard working condition calibration parameter sequence to obtain a variable pressure working condition compensation curve. The compensation curve is used to describe the nonlinear relationship between pressure and fluidity of the equipment in a variable pressure environment, and how to maintain the consistency of response by dynamically adjusting parameters such as the suction rate, so as to ensure that the equipment can still provide accurate measurement data in a variable pressure environment. After completing the variable pressure condition test, the multi-layer response calibration instruction is subjected to a composite condition response test through the third permeability controller to obtain a third pressure response data group and a third fluidity response data group. Composite conditions refer to the situation that the permeability of the formation is affected by multiple factors at the same time, including the interaction and coupling changes of pressure and fluidity. In view of this complex environment, based on the standard condition calibration parameter sequence and the variable pressure condition compensation curve, the third pressure response data group and the third fluidity response data group are subjected to dual parameter calibration operations to obtain a composite condition response parameter matrix. The dual parameter calibration of the composite condition helps to understand the mutual influence between different physical conditions, ensure that the logging equipment can adapt to the working environment under complex formation conditions, and maintain the accuracy and reliability of the response. Based on the composite condition response parameter matrix, the pressure response data is nonlinearly fitted by the first curve fitter, and the fluidity response data is nonlinearly fitted by the second curve fitter to obtain a dual-channel calibration feature sequence. The role of nonlinear fitting is to capture the complex patterns and trends in the response data, eliminate errors and deviations that cannot be explained by linear fitting, and make the calibration feature sequence after fitting more accurately reflect the actual response of the equipment. The dual-channel calibration feature sequence contains the response characteristics of pressure and fluidity under different working conditions. The dual-channel calibration feature sequence is input into the parameter fusion network for pressure-fluidity coupling analysis to obtain the final pressure-fluidity dual-parameter calibration curve. The design of the parameter fusion network includes a multimodal feature extraction layer, a feature interaction layer, and a parameter fusion layer. In the feature extraction layer, a dual-channel structure is used to process pressure and fluidity data separately, so as to maintain the independence of the two data characteristics while performing in-depth extraction to ensure the integrity and diversity of the features.Then, in the feature interaction layer, the dual-channel features are deeply fused to capture the complex coupling relationship between pressure and fluidity. This process can help the model understand the mutual dependence and interaction between the two under various working conditions, and form a comprehensive description of the formation characteristics. The final calibration parameters are output through the parameter fusion layer. The calibration parameters are used to describe the optimal response configuration of the equipment under various complex conditions, so that the logging equipment can accurately perform measurement tasks under different formation conditions.

[0020] In one example, dual-parameter response modeling is performed on the sealing area coefficient, suction rate coefficient and pressure measurement range of the logging equipment to obtain a pressure-fluidity coupling response model, including: performing linear mapping transformation on the sealing area coefficient and the suction rate coefficient to obtain an initial response coefficient matrix of the logging equipment, and based on the initial response coefficient matrix of the logging equipment, performing piecewise linearization processing on the pressure measurement range to obtain a pressure response function of the logging equipment; inputting the initial response coefficient matrix of the logging equipment and the pressure response function of the logging equipment into a second-order differential equation solver for operation to obtain a fluidity response function of the logging equipment; and calculating the fluidity response function of the logging equipment according to the pressure response matrix of the logging equipment. The force response function and the fluidity response function of the logging equipment are used to perform nonlinear fitting calculation on the response characteristics to obtain a dual-parameter response characteristic curve; the dual-parameter response characteristic curve is input into an adaptive filter for signal filtering to obtain a pressure-fluidity response association matrix, and a singular value decomposition operation is performed on the pressure-fluidity response association matrix to obtain a logging equipment response eigenvector; based on the logging equipment response eigenvector, a least squares fitting operation is performed on the response parameters to obtain a logging equipment coupling coefficient matrix, and the logging equipment coupling coefficient matrix is ​​substituted into the state space equation for numerical solution to obtain a pressure-fluidity coupling response model.

[0021] In this example, the setting area coefficient and the suction rate coefficient are mathematically modeled. Assume that the setting area coefficient is , the suction rate coefficient is , transform them into an initial response coefficient matrix through a linear mapping function. Let the linear mapping function be: ; in, is the initial response coefficient matrix, , , are the coefficients of the mapping, which are used to adjust the weights of the linear mapping. and The physical characteristics of the well logging equipment are obtained by , which is used to describe the initial response characteristics of the device during the measurement process. Based on the initial response coefficient matrix , the pressure measurement range is piecewise linearized to obtain the pressure response function of the logging equipment. Assuming the pressure measurement range is In order to simplify the model, the pressure measurement range is divided into linear intervals, each of which is represented by ,in In each interval, the pressure response is approximately expressed as a linear function: ; in, and They are The slope and intercept of each interval are used to describe the linear response characteristics of the pressure in this interval. Through piecewise linearization processing, the complex nonlinear pressure response can be effectively simplified into multiple linear intervals, which is convenient for subsequent calculation and analysis. and the pressure response function Input into the second-order differential equation solver for operation to obtain the fluidity response function of the logging equipment. Assuming the fluidity Response and pressure There exists a second-order differential relationship, then the relationship is expressed as the following second-order ordinary differential equation: ; in, is the acceleration term of fluidity, is the rate of change of fluidity, and is the coefficient controlling the fluidity response, is the mapping of the pressure response function to time. By solving the above second-order differential equation, we can obtain the flow response function , which describes the dynamic change of flow rate over time and is directly related to the change of pressure. Based on the pressure response function and the flow response function , perform nonlinear fitting calculation on the response characteristics to obtain the dual-parameter response characteristic curve. Assume that the fitting function is , nonlinear least squares method is used to fit the relationship between pressure and fluidity, for example: ; in, , , , , , is the fitting parameter. By fitting the pressure data and fluidity data, the dual-parameter response characteristic curve can accurately describe the comprehensive response characteristics of the logging equipment under different pressure and flow conditions. The dual-parameter response characteristic curve is input into the adaptive filter for signal filtering to obtain the pressure-fluidity response correlation matrix. The adaptive filter is used to remove high-frequency noise and unwanted signals in the measurement data, thereby improving the smoothness and accuracy of the data. Suppose the output of the adaptive filter is , the filtered pressure-fluidity response correlation matrix is ​​expressed as: ; in, is the weight coefficient of the filter, which is updated by an adaptive algorithm to minimize the error after filtering. Pressure-fluidity response correlation matrix It can reflect the coupling relationship between pressure and fluidity, and the filtered data has better robustness and consistency. Pressure-fluidity response correlation matrix Perform singular value decomposition to extract the response eigenvector of the logging equipment. Singular value decomposition decomposes the complex matrix into several eigenvectors and singular values. ,in, and are the left singular vector matrix and the right singular vector matrix respectively, is the singular value matrix. By extracting the main singular vectors, the response eigenvector of the logging equipment is obtained to describe the main response mode of the equipment. Based on the response eigenvector of the logging equipment, the least squares fitting operation is performed on the response parameters to obtain the coupling coefficient matrix of the logging equipment. Assume that the coupling coefficient matrix is , solve the following equation by the least squares method: ; in, is the input data matrix, represents the norm of the matrix. The goal of the least squares fitting operation is to find an optimal coupling coefficient matrix , so that the fitting error is minimized, thus accurately describing the coupling relationship between pressure and fluidity. The coupling coefficient matrix of the logging equipment Substitute it into the state space equation for numerical solution to obtain the pressure-fluidity coupling response model. The state space equation is used to describe the dynamic evolution of the system over time, and its form is expressed as: ; ; in, is the state vector, , , , They are the system matrix, input matrix, output matrix and direct transmission matrix, is the input signal, is the output signal. By transforming the coupling coefficient matrix Substitute the state space equation and perform numerical solution to obtain the dynamic response characteristics of the logging equipment under different pressure and fluidity conditions, forming a complete pressure-fluidity coupling response model.

[0022] In one example, according to the pressure-fluidity coupling response model, the data acquisition parameters and signal filtering parameters of the logging equipment are configured in intervals to obtain a dense layer response parameter set, a conventional layer response parameter set and a mixed layer response parameter set, including: performing segmented quantitative calculation on the sampling frequency range of the pressure-fluidity coupling response model to obtain a dynamic sampling matrix of the logging equipment, and based on the dynamic sampling matrix of the logging equipment, performing response sensitivity calculation on the sealing area coefficient to obtain a sensitivity partition sequence of the logging equipment; inputting the sensitivity partition sequence of the logging equipment into a formation classifier to perform permeability characteristic analysis to obtain a formation classification matrix of the logging equipment, and performing Gaussian filtering operation on the logging signal according to the formation classification matrix of the logging equipment to obtain a multi-layer filtering parameter sequence; and inputting the multi-layer filtering into the formation classifier to obtain a multi-layer filtering parameter sequence. The wave parameter sequence is input into the adaptive quantizer for parameter allocation to obtain the partition parameter matrix of each layer, and the fluidity response compensation operation is performed on the partition parameter matrix of each layer to obtain the multi-layer response compensation coefficient; based on the multi-layer response compensation coefficient, the layered response correction calculation is performed on the logging data to obtain the layered response correction matrix, and the layered response correction matrix is ​​substituted into the response characteristic allocation model for parameter configuration to obtain the dense layer response parameter set, the conventional layer response parameter set and the mixed layer response parameter set. The response characteristic allocation model includes: a multi-layer allocation algorithm based on formation type and measurement parameters, establishing response characteristic mapping relationships for dense layers, conventional layers and mixed layers respectively, and adopting a layered adaptive weight configuration mechanism to achieve optimal allocation of response parameters of logging equipment under different formation conditions.

[0023] In this example, the sampling frequency range is quantized in segments. Assume that the sampling frequency of the pressure-fluidity coupled response model is , whose range is , dividing the sampling frequency range into intervals, each interval has a width of By quantizing the sampling frequency in sections, the dynamic sampling matrix of the logging equipment is obtained. , whose elements Indicates The frequency interval is The sampled values ​​at the time step are expressed in matrix form as follows: ; in, Indicates different frequency ranges, is the frequency increment of each interval, and a sampling matrix containing multiple frequency bands is generated through segmented quantization , used for subsequent sensitivity calculation. Dynamic sampling matrix based on logging equipment , the response sensitivity of the sealing area coefficient is calculated to obtain the sensitivity partition sequence of the logging equipment. Assume that the sealing area coefficient is , the sensitivity calculation formula is: ; in, Indicated in The frequency interval and The sensitivity over the time step, is the sensitivity proportionality factor, which describes the effect of sampling frequency on sensitivity. The sensitivity of each element in is calculated to obtain the sensitivity partition sequence matrix of the logging equipment. . Sensitivity partition sequence matrix It can effectively reflect the sensitivity of the logging equipment to the change of the setting area in different frequency ranges. Input the formation classifier to perform permeability characteristic analysis to obtain the formation classification matrix of the logging equipment The role of the formation classifier is to determine the formation permeability characteristics of different regions by analyzing the sensitivity data. Assume that the permeability is The basis for stratigraphic classification is: ; in, and is the classification threshold, which is used to divide the dense layer, regular layer and mixed layer. In this way, the sensitivity data Divided into three categories, corresponding to different stratum types, forming a stratum classification matrix . Stratigraphic classification matrix based on logging equipment , perform Gaussian filtering on the logging signal to obtain a multi-layer filtering parameter sequence. Assume that the logging signal is , the Gaussian kernel of the filter is ,in is the filter width. The filtered signal is expressed as: ; By performing Gaussian filtering on the signals of different layers, the filter parameter sequences of each layer are obtained. The filter parameter sequences of each layer reflect the characteristics of the signal after filtering in different frequency ranges. The multi-layer filter parameter sequences are input into the adaptive quantizer for parameter allocation to obtain the partition parameter matrix of each layer. The adaptive quantizer dynamically allocates the parameters of each layer according to the characteristics of the signal after filtering. Assume that the filter parameter sequence is , the quantized parameter matrix is , the quantization process is expressed as: ; in, is the adaptive quantization function, To quantize the weights, adjust the weights , so that the quantized parameters are more consistent with the characteristics of the logging signal. Perform mobility response compensation calculation on each layer partition parameter matrix to obtain the multi-layer response compensation coefficient. Assume that the mobility response is , the compensation coefficient is , then the response after compensation is expressed as: ; in, The calculation is based on the partition parameter matrix of each layer. Through the mobility response compensation, the response deviation caused by the mobility change between different layers is effectively corrected. Based on the multi-layer response compensation coefficient, the layered response correction calculation is performed on the logging data to obtain the layered response correction matrix . Assume that the original response matrix is , the correction matrix after compensation is , the correction process is expressed as: ; Through correction, the response of the logging equipment in each layer is guaranteed to be consistent, and the measurement error caused by different formation characteristics is reduced. The layered response correction matrix is ​​substituted into the response characteristic allocation model to configure the parameters, and the dense layer response parameter set, the conventional layer response parameter set and the mixed layer response parameter set are obtained. The response characteristic allocation model includes a multi-layer allocation algorithm based on formation type and measurement parameters. Assume that the dense layer parameter set is , the conventional layer parameter set is , the mixing layer parameter set is , then the multi-layer allocation algorithm is expressed as: ; in, As the allocation function, by combining the stratigraphic classification matrix and the hierarchical response correction matrix , assigning the best response parameters to different types of formations. In this model, a hierarchical adaptive weight configuration mechanism is used to achieve the optimal allocation of response parameters of logging equipment under different formation conditions. Assume that the adaptive weight is , the final response parameter of each layer is , then the weight configuration mechanism is expressed as: ; By adjusting the weight , dynamically adjust the response parameters of each layer according to the actual formation characteristics, thereby achieving optimal configuration of the parameters of the logging equipment.

[0024] In one example, a logging pressure data sequence and a fluidity data sequence are input into a dual-channel deep neural network for dynamic feature extraction and response pattern classification to obtain a logging data deviation characteristic spectrum and a response anomaly type matrix, including: segmenting the logging pressure data sequence and the fluidity data sequence through a time sliding window to obtain a pressure data subsequence group and a fluidity data subsequence group; inputting the pressure data subsequence group into a first LSTM encoder, the first LSTM encoder includes three layers of bidirectional LSTM units, each layer of LSTM units includes 128 hidden state nodes, and obtaining a pressure feature encoding vector; inputting the fluidity data subsequence group into a second LSTM encoder, the second LSTM encoder includes three layers of bidirectional LSTM units, each layer of LSTM units includes 128 hidden state nodes, and obtaining a fluidity feature encoding vector; performing an attention mechanism fusion operation on the pressure feature encoding vector and the fluidity feature encoding vector, the attention mechanism adopts a multi-head self-attention structure , the number of attention heads is 8, the dimension of each attention head is 64, and the multimodal feature fusion matrix is ​​obtained; the multimodal feature fusion matrix is ​​input into the feature extraction CNN network, the feature extraction CNN network contains four convolutional layers, the number of convolution kernels in each layer is 64, 128, 256, 512, the convolution kernel size is 3×3, and the ReLU activation function is used to obtain the deep feature map; the deep feature map is processed by global average pooling and maximum pooling, and the pooling results are spliced ​​and fused to obtain the feature vector sequence; the feature vector sequence is input into the first fully connected network, the first fully connected network contains three layers, the number of neurons is 512, 256, 128, and the Dropout rate is 0.5 to obtain the deviation characteristic spectrum of the logging data; the feature vector sequence is input into the second fully connected network, the second fully connected network contains three layers, the number of neurons is 512, 256, 7, and the last layer uses the Softmax activation function to obtain the response anomaly type matrix.

[0025] In this example, the original logging data sequence is preprocessed. Assume that the pressure data sequence is , the flow data series is ,in Indicates the total length of the data sequence. In order to capture the dynamic change characteristics in the logging data, the data is segmented by the time sliding window method. Assume that the sliding window length is , the sliding step length is , then for the pressure data sequence, split it into multiple subsequence groups: ; Similarly, for the mobility data series , and obtain multiple mobility data subsequence groups These subsequence groups are used to capture the logging pressure and mobility characteristics in different time windows. The first LSTM encoder is input for processing. The first LSTM encoder contains three layers of bidirectional LSTM units, and each layer of LSTM units contains 128 hidden state nodes. Bidirectional LSTM can consider both forward and reverse time series information at the same time to better understand the time dependency of logging data. Assume that The hidden state of the LSTM unit of the layer is , then it is expressed as: ; Among them, LSTM Indicates layer LSTM units, and are the weight and bias parameters of this layer, is the time step Through the processing of three-layer bidirectional LSTM, the feature coding of the pressure data subsequence group can be extracted to obtain the pressure feature coding vector Similarly, the mobility data subsequences are grouped into The second LSTM encoder also contains three layers of bidirectional LSTM units, each layer of LSTM units contains 128 hidden state nodes. Through similar operations, the feature encoding vector of the flow data subsequence is obtained. , used to describe the temporal characteristics of flow data. The pressure feature encoding vector and the fluidity feature encoding vector The attention mechanism is used to perform feature fusion operations. The multi-head self-attention structure is used to achieve interaction and fusion between features. Assume that the number of heads of the multi-head attention is , the dimension of each attention head is . Multi-head self-attention is expressed as: ; in, , , They represent the query matrix, key matrix, and value matrix respectively, and softmax is the activation function used to calculate the attention weight. Through the multi-head self-attention mechanism, the multimodal feature fusion matrix F between the pressure feature encoding and the fluidity feature encoding is obtained, which describes the complex coupling relationship between pressure and fluidity. Input feature extraction CNN network to extract deep features. The feature extraction CNN network contains four convolutional layers, the number of convolution kernels in each layer is 64, 128, 256, 512 respectively, and the convolution kernel size is . Assume that the input feature map is , No. The output of the convolutional layer is: ; in, and Respectively The convolution kernel and bias of the layer, * represents the convolution operation, and ReLU is the activation function used to introduce nonlinear characteristics. Through the layer-by-layer abstraction of multi-layer convolution, the deep feature map is obtained. , which contains the deep fusion features between pressure and fluidity. Perform dual-path processing of global average pooling and maximum pooling to obtain a more compact feature representation. Global average pooling calculates the mean of each channel, while maximum pooling calculates the maximum value of each channel, which is specifically expressed as: ; in, and They are the results of global average pooling and maximum pooling, respectively. is the number of elements in the feature map. The two pooling results are concatenated and fused to obtain a feature vector sequence , this sequence is used for further feature extraction. The input is sent to the first fully connected network, which contains three layers with 512, 256, and 128 neurons respectively. The Dropout rate is 0.5 to prevent overfitting. The output of the layer is , then it is expressed as: ; Through layer-by-layer processing of the three-layer fully connected network, the deviation characteristic spectrum of the logging data is obtained , which is used to describe the abnormal characteristics and potential deviations in logging data. The input is sent to the second fully connected network for classification of the response abnormality type. The second fully connected network also contains three layers, with 512, 256 and 7 neurons respectively. The last layer uses the Softmax activation function for classification. The form of the Softmax function is: ; in, For the The probability of the class, is the output of the fully connected network. elements, is the number of categories. Through Softmax activation, we get the response anomaly type matrix , where each row represents the probability distribution of the anomaly type of the corresponding data subsequence.

[0026] In one example, according to the deviation characteristic spectrum of logging data and the response anomaly type matrix, multi-objective parameter optimization calculation is performed on the dense layer response parameter set, the conventional layer response parameter set and the mixed layer response parameter set to obtain a hierarchical calibration parameter sequence, including: performing feature fusion operation on the deviation characteristic spectrum of logging data and the response anomaly type matrix to obtain a comprehensive feature matrix of logging data, and based on the comprehensive feature matrix of logging data, performing pressure response optimization calculation on the dense layer response parameter set to obtain a dense layer pressure calibration parameter group; performing fluidity compensation calculation on the dense layer pressure calibration parameter group to obtain a dense layer fluidity calibration parameter group, and performing pressure gradient optimization operation on the conventional layer response parameter set according to the comprehensive feature matrix of logging data to obtain a conventional layer pressure and fluidity calibration parameter group; performing multi-parameter coupling optimization on the mixed layer response parameter set based on the comprehensive feature matrix of logging data to obtain a mixed layer pressure and fluidity calibration parameter group; The tight layer pressure calibration parameter group, the tight layer fluidity calibration parameter group, the conventional layer pressure fluidity calibration parameter group and the mixed layer pressure fluidity calibration parameter group are input into the multi-layer response optimizer for parameter collaborative operation to obtain a multi-layer calibration parameter matrix; the multi-layer response optimizer includes: a multi-objective optimization algorithm processing unit, a parameter collaborative adjustment unit and a response characteristic evaluation unit, which realizes the global optimal response characteristics of the logging equipment under complex formation conditions by optimizing the response parameters of multiple formations in parallel; the multi-layer calibration parameter matrix is ​​subjected to inter-layer response compensation calculation to obtain a calibration compensation coefficient sequence, and the calibration compensation coefficient sequence is substituted into the parameter mapping model for hierarchical configuration to obtain a hierarchical calibration parameter sequence; the parameter mapping model includes: a nonlinear parameter mapping unit, a dynamic feature extraction unit and a parameter reconstruction unit, which are used to establish a mapping relationship between the response parameters of the logging equipment and the calibration parameters to realize accurate reconstruction of the response characteristics and parameter optimization.

[0027] In this example, the deviation characteristic spectrum of the logging data and the response anomaly type matrix are fused. Suppose the deviation characteristic spectrum of the logging data is , the response exception type matrix is ,in represents the number of data samples, and The feature dimensions represent the deviation feature and the anomaly type respectively. By means of feature concatenation, the two matrices are concatenated along the feature dimension to obtain the comprehensive feature matrix of the logging data.

[0028] ; in, Indicates that and By splicing along the feature dimension, the deviation feature and the response anomaly feature are integrated to obtain a comprehensive feature description. , the pressure response optimization calculation is performed on the dense layer response parameter set to obtain the dense layer pressure calibration parameter set. Assume that the pressure response function of the dense layer is ,in Indicates pressure, is the parameter vector to be optimized. In order to make the pressure response of the dense layer closer to the expected response, the loss function is minimized. To optimize: ; in, For the The pressure value of each sample, is the target pressure response value. Minimize it by gradient descent or other optimization algorithms , and obtain the pressure calibration parameter set of the dense layer . For the dense layer pressure calibration parameter set Perform mobility compensation calculation to obtain the mobility calibration parameter set of the dense layer. Assume that the mobility response function is ,in is the fluidity, is the pressure calibration parameter. Through flow compensation, the parameters are adjusted to achieve an ideal balance between flow response and pressure response. The goal of compensation calculation is to minimize the error between flow response and pressure response, and the loss function is defined as: ; By minimizing , and obtain the fluidity calibration parameter set of the dense layer , to ensure the matching between pressure and fluidity. Based on the comprehensive feature matrix of logging data , perform pressure gradient optimization calculation on the conventional layer response parameter set to obtain the pressure flow calibration parameter set of the conventional layer. Assume that the pressure gradient function of the conventional layer is ,in is the parameter vector to be optimized. The goal of gradient optimization is to make the pressure response of the conventional layer stable and consistent during the change process. The loss function is defined as: ; By minimizing , and obtain the pressure and flow calibration parameter set of the conventional layer This parameter group can ensure that the pressure gradient changes in conventional formations remain within a reasonable range and improve the accuracy of logging data. , multi-parameter coupling optimization is performed on the mixed layer response parameter set to obtain the pressure and fluidity calibration parameter set of the mixed layer. The response characteristics of the mixed layer are relatively complex, and the coupling relationship between multiple parameters needs to be considered simultaneously. Assume that the pressure and fluidity response functions of the mixed layer are ,in is the parameter vector to be optimized, and the goal of coupled optimization is to minimize the coupling error between pressure and fluidity response. The loss function is defined as: ; in, is the expected response value, by minimizing , and obtain the pressure and flow calibration parameter set of the mixed layer , thus ensuring that the response of pressure and mobility in complex formations can achieve the best matching state. , Dense layer flow calibration parameter group , Pressure and flow calibration parameter group of conventional layer And the mixed layer pressure and flow calibration parameter set Input into the multi-layer response optimizer for parameter coordination operation to obtain the multi-layer calibration parameter matrix The multi-layer response optimizer includes a multi-objective optimization algorithm processing unit, a parameter coordination adjustment unit, and a response characteristic evaluation unit. By optimizing the response parameters of multiple formations in parallel, the global optimal response characteristics of the logging equipment under complex formation conditions are achieved. Assume that the multi-objective optimization function is: ; in, For the The weight of each layer, is the loss function of the corresponding layer. By adjusting the weights of each layer, the response parameters under different formation conditions are optimized. The parameter coordination adjustment unit is used to coordinate the changes of parameters between different formations to ensure the global optimal characteristics. Perform inter-layer response compensation calculation to obtain the calibration compensation coefficient sequence . Let the calibration compensation function be , the goal of compensation calculation is to reduce the difference in inter-layer responses so that the responses of each layer are more consistent. The calculation of compensation coefficient is achieved by minimizing the following loss function: ; By solving the above loss function, we get the calibration compensation coefficient sequence It can effectively reduce the response deviation between layers. Substitute into the parameter mapping model to perform hierarchical configuration and obtain a hierarchical calibration parameter sequence. The parameter mapping model includes a nonlinear parameter mapping unit, a dynamic feature extraction unit, and a parameter reconstruction unit. Its purpose is to establish a nonlinear mapping relationship between the logging equipment response parameters and the calibration parameters. Suppose the nonlinear mapping function is , the mapping process is expressed as: ; in, It represents the final hierarchical calibration parameter sequence. The dynamic feature extraction unit is used to extract the feature differences between different formations, and the parameter reconstruction unit is used to reconstruct the parameters according to these differences to ensure that the response characteristics of the logging equipment reach the optimal state under each formation condition.

[0029] In one example, based on the layered calibration parameter sequence, a response benchmark analysis is performed in turn through the formation permeability evaluation matrix, a response gain analysis is performed through the pressure gradient evaluation matrix, and a response compensation analysis is performed through the mobility attenuation evaluation matrix to obtain a multi-layer response calibration instruction, including: performing a permeability analysis operation on the dense layer response parameter set in the layered calibration parameter sequence to obtain a dense layer benchmark response sequence; performing a permeability analysis operation on the conventional layer response parameter set in the layered calibration parameter sequence to obtain a conventional layer benchmark response sequence; performing a permeability analysis operation on the mixed layer response parameter set in the layered calibration parameter sequence to obtain a mixed layer benchmark response sequence; and performing a pressure measurement on the dense layer benchmark response sequence, the conventional layer benchmark response sequence, and the mixed layer benchmark response sequence. Gradient calculation is performed to obtain a pressure response gain matrix; according to the pressure response gain matrix, the sealing area coefficient is dynamically corrected and calculated to obtain the sealing response correction coefficient, and based on the pressure response gain matrix and the sealing response correction coefficient, the fluidity compensation operation is performed on the suction rate coefficient to obtain the fluidity attenuation correction coefficient; the sealing response correction coefficient and the fluidity attenuation correction coefficient are input into the parameter co-processor for joint optimization to obtain a multi-layer calibration control matrix, and the multi-layer calibration control matrix is ​​subjected to instruction conversion operation to obtain a multi-layer response calibration instruction; the parameter co-processor includes: a multi-channel parameter fusion unit, a dynamic weight allocation unit and a parameter optimization unit, which realizes the joint optimization and dynamic adjustment of multiple parameters of the logging equipment by co-processing the pressure response parameters and the fluidity response parameters.

[0030] In this example, the response parameters of the dense layer are modeled. Assume that the response parameter set of the dense layer is , permeability It is an important parameter to measure the fluid flow capacity of the formation. Through permeability analysis, the flow characteristics of the formation under different pressure and mobility conditions are determined. Assume that the permeability analysis function is ,in Indicates pressure, Represents permeability. Permeability analysis determines the benchmark response through the relationship between response parameters and permeability: ; in, The benchmark response sequence of the dense layer is represented by the coupling relationship between the response parameter set and the permeability, and the benchmark response characteristics of the dense layer under different pressure conditions are obtained. This benchmark response is used to describe the normal performance of the dense layer under stable pressure conditions. Similarly, for the conventional layer response parameter set Perform permeability analysis to obtain the baseline response sequence of the conventional layer The permeability of conventional layers is usually high, and the relationship between pressure and mobility is relatively linear. The permeability analysis directly reflects the response characteristics of conventional layers under different conditions. The calculation of the benchmark response sequence is expressed as: ; This benchmark response sequence is used to describe the pressure and fluidity response characteristics of conventional layers under standard permeability conditions. , the complexity of permeability analysis is relatively high, because the mixed layer usually has diverse geological characteristics and its permeability exhibits nonlinearity. The permeability analysis function is also , but multiple analyses of different regions of the mixed layer are needed to obtain a comprehensive baseline response sequence: ; Mixed Layer Reference Response Sequence It is used to describe the response characteristics of the mixed layer in different regions, taking into account the influence of different geological features on fluid flow. , conventional layer benchmark response sequence and the mixed layer reference response sequence , perform gradient calculation on the pressure measurement data to obtain the pressure response gain matrix. The purpose of gradient calculation is to determine the rate of change of pressure response of different layers. Suppose the pressure measurement data is , the gradient calculation formula is: ; in, is the pressure response gain matrix, represents the derivative of the base response with respect to pressure. , and Perform gradient calculation to obtain the corresponding pressure response gain matrix , and , these matrices are used to describe the sensitivity of each layer to pressure changes. According to the pressure response gain matrix, the setting area coefficient is dynamically corrected to obtain the setting response correction coefficient. Assume that the setting area coefficient is , the corrected setting response correction coefficient is The purpose of dynamic correction calculation is to make the setting response consistent with the pressure gain. The specific correction formula is: ; in, is the pressure response gain matrix of the corresponding layer. By multiplying the sealing area coefficient with the pressure gain, the sealing area coefficient is dynamically adjusted to match the pressure response of the current formation. Based on the pressure response gain matrix and the sealing response correction coefficient, the fluidity compensation operation is performed on the suction rate coefficient to obtain the fluidity attenuation correction coefficient. Assume that the suction rate coefficient is , the fluidity decay correction factor is , the compensation formula is: ; By combining the suction rate coefficient with the setting response correction coefficient and the pressure gain, a correction coefficient that adapts to the current pressure and flow conditions is obtained to ensure that the flow characteristics of the equipment remain stable under different pressures and flow conditions. and the mobility decay correction factor The input parameters are jointly optimized with the co-processor to obtain a multi-layer calibration control matrix The parameter co-processor includes a multi-channel parameter fusion unit, a dynamic weight allocation unit and a parameter optimization unit. By jointly optimizing the pressure response parameters and fluidity response parameters of different formations, the global response characteristics of the logging equipment under complex formation conditions are optimized. The main function of the multi-channel parameter fusion unit is to fuse parameters from different formations to capture the mutual influence between the layers. Assume that the correction coefficients of the dense layer, conventional layer and mixed layer are , and , the fusion process is expressed as: ; in, , , The weights assigned to the dynamic weight allocation unit are used to adjust the contribution of each layer parameter in the fusion process. The dynamic adjustment of weights ensures that the fusion results adapt to the changes under different formation conditions. The parameter optimization unit further optimizes the fused parameters globally to ensure that the optimal response characteristics can be obtained under all formation conditions. Let the loss function be , which is used to measure the response error under the current parameter setting. The optimization goal is to minimize the loss function: ; By globally optimizing the fused parameters, the multi-layer calibration control matrix is ​​obtained. , which is used to describe the optimal response settings of each layer under different conditions. For the multi-layer calibration control matrix Perform command conversion operations to obtain multi-layer response calibration commands. Convert the optimized parameters into control commands that can be executed by the measuring equipment to adjust the response characteristics of the equipment and ensure the measurement accuracy under different formation conditions. Suppose the command conversion function is , then the multi-layer response calibration instruction is expressed as: ; in, It represents the final generated sequence of control instructions, through which the logging equipment dynamically adjusts its response parameters according to the current formation conditions.

[0031] In one example, a multi-layer response calibration instruction is subjected to comparative tests under standard permeability conditions, variable pressure permeability conditions, and composite permeability conditions, respectively, to obtain a pressure-fluidity dual-parameter calibration curve, including: subjecting the multi-layer response calibration instruction to a standard condition response test through a first permeability controller to obtain a first pressure response data group and a first fluidity response data group, and performing a first parameter calibration operation on the sealing area coefficient based on the first pressure response data group and the first fluidity response data group to obtain a standard condition calibration parameter sequence; subjecting the multi-layer response calibration instruction to a variable pressure condition response test through a second permeability controller to obtain a second pressure response data group and a second fluidity response data group, and performing a suction rate correction calculation on the second pressure response data group and the second fluidity response data group based on the standard condition calibration parameter sequence to obtain a variable pressure condition compensation curve; subjecting the multi-layer response calibration instruction to a composite permeability condition test through a third permeability controller to obtain a first pressure response data group and a first fluidity response data group, and performing a first parameter calibration operation on the sealing area coefficient based on the first pressure response data group and the ... standard condition calibration parameter sequence to obtain a standard condition calibration parameter sequence; subjecting the multi-layer response calibration instruction to a composite permeability condition test through a third permeability controller to obtain a first pressure response data group and a second fluidity response data group, and performing a first parameter calibration operation on the sealing area coefficient based on the first pressure response data group and the first fluidity response data group, and performing a first parameter calibration operation on the sealing area coefficient based on the standard condition calibration parameter sequence to obtain a standard condition calibration parameter sequence A combined operating condition response test is performed to obtain a third pressure response data group and a third fluidity response data group, and a dual-parameter calibration operation is performed on the third pressure response data group and the third fluidity response data group according to the standard operating condition calibration parameter sequence and the variable pressure operating condition compensation curve to obtain a composite operating condition response parameter matrix; based on the composite operating condition response parameter matrix, a nonlinear fitting is performed on the pressure response data through a first curve fitter, and a nonlinear fitting is performed on the fluidity response data through a second curve fitter to obtain a dual-channel calibration feature sequence, and the dual-channel calibration feature sequence is input into a parameter fusion network for pressure-fluidity coupling analysis to obtain a pressure-fluidity dual-parameter calibration curve; the parameter fusion network includes: a multimodal feature extraction layer, a feature interaction layer and a parameter fusion layer, wherein the feature extraction layer adopts a dual-channel structure to process pressure and fluidity data respectively, the feature interaction layer realizes deep fusion of dual-channel features, and the parameter fusion layer outputs the final calibration parameters.

[0032] In this example, the logging environment under standard working conditions is defined. Standard working conditions generally refer to the response test of the logging equipment under constant permeability and pressure conditions to determine its performance under ideal conditions. Suppose the multi-layer response calibration instruction received by the logging equipment under the control of the standard permeability controller is This instruction is used to drive the adjustment of various parameters of the logging equipment so as to perform measurements under standard conditions. By sampling the response of the logging equipment, the first pressure response data group is obtained. and the first fluidity response data set ,in represents the number of sampling points, and Describe the pressure response and flow rate response of the equipment under standard working conditions respectively. Based on the first pressure response data set and the first fluidity response data set , perform the first parameter calibration operation on the sealing area coefficient to obtain the calibration parameter sequence of the standard working condition. Assume that the sealing area coefficient is , the relationship between the pressure response and the fluidity response is expressed as a calibration equation of the following form: ; in, is the calibration function, and is a constant term, and Represents pressure and fluidity respectively. Under standard working conditions, by fitting the above calibration equation, the error between pressure response and fluidity response is minimized to obtain the optimal sealing area coefficient. , forming a standard working condition calibration parameter sequence . The multi-layer response calibration instruction The second permeability controller is used to perform a variable pressure response test to simulate different pressure change conditions and obtain a second pressure response data set. and the second mobility response data set Variable pressure conditions refer to formation environments where pressure changes over time. Through the permeability controller, the logging conditions are dynamically adjusted to capture the response characteristics of the equipment under unstable pressure. Calibration parameter sequence based on standard conditions , for the second pressure response data set and the second mobility response data set The suction rate correction calculation is performed to obtain the compensation curve for variable pressure conditions. The goal of the suction rate correction is to adjust the flow response to match the change in pressure response. Suppose the suction rate is , the correction function is: ; in, is the corrected suction rate, is the compensation function, The calibration parameters for standard working conditions are obtained by compensating the pressure and flow data under variable pressure conditions to obtain the compensation curve for variable pressure conditions. . The multi-layer response calibration instruction The third pressure response data set is obtained by performing a composite operating condition response test through the third permeability controller. and the third fluidity response data set Composite conditions refer to complex environments where formation conditions change in permeability, pressure, and mobility. Under composite conditions, the response capability of logging equipment can be more comprehensively evaluated by controlling the changes in permeability and pressure. Calibration parameter sequence based on standard conditions And transformer working condition compensation curve , for the third pressure response data set and the third fluidity response data set Perform dual parameter calibration to obtain the composite operating condition response parameter matrix The response characteristics of pressure and flow are corrected at the same time to ensure that the device can provide accurate measurements under complex conditions. The calibration equation is expressed as: ; in, is the calibration parameter to be optimized. By minimizing the error between the measured response and the expected response, the optimal calibration parameter matrix under the composite working condition is obtained. Based on the composite operating condition response parameter matrix , the pressure response data is nonlinearly fitted by the first curve fitter, and the fluidity response data is nonlinearly fitted by the second curve fitter to obtain the dual-channel calibration characteristic sequence. The purpose of nonlinear fitting is to capture the complex patterns and trends in the pressure and fluidity responses. Assume that the pressure fitting function is , the fluidity fitting function is , the fitting process is expressed as: ; in, and They are the fitting parameters of pressure and fluidity, respectively. By performing nonlinear fitting on pressure and fluidity, more accurate response characteristics are obtained to form a dual-channel calibration feature sequence . Set the dual channel calibration feature sequence The input parameter fusion network is used for pressure-fluidity coupling analysis to obtain the pressure-fluidity dual parameter calibration curve. The parameter fusion network consists of a multimodal feature extraction layer, a feature interaction layer, and a parameter fusion layer. The feature extraction layer uses a dual-channel structure to process pressure and fluidity data separately to maintain the independence and integrity of these two physical quantities. Assume that the pressure feature is , the flow characteristics are , the output of the feature extraction layer is expressed as: ; in, and They are the feature extraction functions of pressure and fluidity, respectively. The function of the feature interaction layer is to achieve deep fusion of dual-channel features to capture the coupling relationship between pressure and fluidity. The fusion process is expressed as: ; in, is a feature interaction function used to capture the complex interaction between pressure and fluidity. The fused features are finally optimized through the parameter fusion layer to output the calibration parameters , the process is expressed as: ; in, is the output function of the fusion layer. Through the fused calibration parameters, the pressure-fluidity dual-parameter calibration curve is obtained. , which describes the optimal response characteristics of the logging equipment under different pressure and fluidity conditions.

[0033] Reference Figure 2 This embodiment provides a response characteristic calibration device for a well logging device, comprising: Modeling module 1 is used to perform dual-parameter response modeling on the setting area coefficient, suction rate coefficient and pressure measurement range of the logging equipment to obtain a pressure-fluidity coupling response model; Configuration module 2, used to configure the data acquisition parameters and signal filtering parameters of the logging equipment in different intervals according to the pressure-fluidity coupling response model, and obtain a dense layer response parameter set, a conventional layer response parameter set and a mixed layer response parameter set; Classification module 3, used for inputting the logging pressure data sequence and the mobility data sequence into a dual-channel deep neural network for dynamic feature extraction and response mode classification, and obtaining the logging data deviation characteristic spectrum and response anomaly type matrix; The calculation module 4 is used to perform multi-objective parameter optimization calculation on the dense layer response parameter set, the conventional layer response parameter set and the mixed layer response parameter set according to the deviation characteristic spectrum of the logging data and the response anomaly type matrix, and obtain a layered calibration parameter sequence; Analysis module 5, used for performing response benchmark analysis through formation permeability evaluation matrix, response gain analysis through pressure gradient evaluation matrix and response compensation analysis through mobility attenuation evaluation matrix based on the layered calibration parameter sequence, to obtain multi-layer response calibration instructions; The test module 6 is used to perform comparative tests on the multi-layer response calibration instructions under standard permeability conditions, variable pressure permeability conditions and composite permeability conditions to obtain a pressure-fluidity dual-parameter calibration curve.

[0034] In this embodiment, for the specific implementation of each unit in the above device embodiment, please refer to the above method embodiment, which will not be repeated here.

[0035] Reference Figure 3 In an embodiment of the present invention, a computer device is also provided. The computer device may be a server, and its internal structure may be as follows: Figure 3As shown. The computer device includes a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. Among them, the processor designed by the computer is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0036] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0037] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0038] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided by the present invention and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM.

[0039] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0040] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for calibrating the response characteristics of a well logging device, characterized in that: The following steps are involved: The double-parameter response modeling of the logging equipment's sealing area coefficient, suction rate coefficient and pressure measurement range was carried out to obtain the pressure-fluidity coupling response model. According to the pressure-fluidity coupling response model, data acquisition parameters and signal filtering parameters of the logging equipment are configured in different intervals to obtain a dense layer response parameter set, a conventional layer response parameter set and a mixed layer response parameter set; The logging pressure data sequence and the mobility data sequence are input into a dual-channel deep neural network for dynamic feature extraction and response mode classification to obtain the logging data deviation characteristic spectrum and response anomaly type matrix; According to the logging data deviation characteristic spectrum and the response anomaly type matrix, a multi-objective parameter optimization calculation is performed on the dense layer response parameter set, the conventional layer response parameter set and the mixed layer response parameter set to obtain a layered calibration parameter sequence; Based on the layered calibration parameter sequence, a response benchmark analysis is performed through a formation permeability evaluation matrix, a response gain analysis is performed through a pressure gradient evaluation matrix, and a response compensation analysis is performed through a mobility attenuation evaluation matrix, to obtain a multi-layer response calibration instruction; The multi-layer response calibration instructions are subjected to comparative tests under standard permeability conditions, variable pressure permeability conditions and composite permeability conditions to obtain a pressure-fluidity dual-parameter calibration curve.

2. The response characteristic calibration method of the logging equipment according to claim 1, characterized in that: The dual-parameter response modeling of the setting area coefficient, the suction rate coefficient and the pressure measurement range of the logging equipment is performed to obtain a pressure-fluidity coupling response model, including: Performing linear mapping transformation on the setting area coefficient and the suction rate coefficient to obtain an initial response coefficient matrix of the logging equipment, and based on the initial response coefficient matrix of the logging equipment, performing piecewise linearization processing on the pressure measurement range to obtain a pressure response function of the logging equipment; Inputting the initial response coefficient matrix of the well logging equipment and the pressure response function of the well logging equipment into a second-order differential equation solver for calculation to obtain a fluidity response function of the well logging equipment; According to the pressure response function of the well logging equipment and the fluidity response function of the well logging equipment, a nonlinear fitting calculation is performed on the response characteristics to obtain a dual-parameter response characteristic curve; Inputting the dual-parameter response characteristic curve into an adaptive filter for signal filtering processing to obtain a pressure-fluidity response correlation matrix, and performing a singular value decomposition operation on the pressure-fluidity response correlation matrix to obtain a logging equipment response eigenvector; Based on the response characteristic vector of the logging equipment, a least square fitting operation is performed on the response parameters to obtain a coupling coefficient matrix of the logging equipment, and the coupling coefficient matrix of the logging equipment is substituted into the state space equation for numerical solution to obtain a pressure-fluidity coupling response model.

3. The response characteristic calibration method of the logging equipment according to claim 2, characterized in that: According to the pressure-fluidity coupling response model, the data acquisition parameters and signal filtering parameters of the logging equipment are configured in different intervals to obtain a dense layer response parameter set, a conventional layer response parameter set and a mixed layer response parameter set, including: Performing segmented quantitative calculation on the sampling frequency range of the pressure-fluidity coupling response model to obtain a dynamic sampling matrix of the logging equipment, and performing response sensitivity calculation on the setting area coefficient based on the dynamic sampling matrix of the logging equipment to obtain a sensitivity partition sequence of the logging equipment; Inputting the well logging equipment sensitivity partition sequence into a formation classifier to perform permeability characteristic analysis to obtain a well logging equipment formation classification matrix, and performing Gaussian filtering operation on the well logging signal according to the well logging equipment formation classification matrix to obtain a multi-layer filtering parameter sequence; Inputting the multi-layer filtering parameter sequence into an adaptive quantizer for parameter allocation to obtain a partition parameter matrix of each layer, and performing a fluidity response compensation operation on the partition parameter matrix of each layer to obtain a multi-layer response compensation coefficient; Based on the multi-layer response compensation coefficient, a layered response correction calculation is performed on the logging data to obtain a layered response correction matrix, and the layered response correction matrix is ​​substituted into the response characteristic allocation model for parameter configuration to obtain a dense layer response parameter set, a conventional layer response parameter set and a mixed layer response parameter set. The response characteristic allocation model includes: a multi-layer allocation algorithm based on formation type and measurement parameters, establishing response characteristic mapping relationships for dense layers, conventional layers and mixed layers respectively, and adopting a layered adaptive weight configuration mechanism to achieve optimal allocation of response parameters of logging equipment under different formation conditions.

4. The response characteristic calibration method of the logging equipment according to claim 3, characterized in that: The well logging pressure data sequence and the fluidity data sequence are input into a dual-channel deep neural network for dynamic feature extraction and response mode classification to obtain a well logging data deviation characteristic spectrum and a response anomaly type matrix, including: Segmenting the logging pressure data sequence and the mobility data sequence respectively through a time sliding window to obtain a pressure data subsequence group and a mobility data subsequence group; Inputting the pressure data subsequence group into a first LSTM encoder, wherein the first LSTM encoder comprises three layers of bidirectional LSTM units, each layer of LSTM units comprises 128 hidden state nodes, to obtain a pressure feature encoding vector; Inputting the mobility data subsequence group into a second LSTM encoder, wherein the second LSTM encoder comprises three layers of bidirectional LSTM units, each layer of LSTM units comprises 128 hidden state nodes, to obtain a mobility feature encoding vector; Performing an attention mechanism fusion operation on the pressure feature coding vector and the fluidity feature coding vector, wherein the attention mechanism adopts a multi-head self-attention structure, the number of attention heads is 8, and the dimension of each attention head is 64, to obtain a multimodal feature fusion matrix; Input the multimodal feature fusion matrix into a feature extraction CNN network, wherein the feature extraction CNN network comprises four convolutional layers, the number of convolution kernels in each layer is 64, 128, 256, and 512 respectively, the convolution kernel size is 3×3, and a ReLU activation function is used to obtain a deep feature map; Performing global average pooling and maximum pooling dual-path processing on the depth feature map, and concatenating and fusing the pooling results to obtain a feature vector sequence; Input the feature vector sequence into a first fully connected network, wherein the first fully connected network comprises three layers, the number of neurons is 512, 256, and 128 respectively, and a dropout rate of 0.5 is used to obtain a deviation feature spectrum of the well logging data; The feature vector sequence is input into the second fully connected network, which includes three layers, and the number of neurons is 512, 256, and 7 respectively. The last layer uses the Softmax activation function to obtain the response anomaly type matrix.

5. The response characteristic calibration method of the logging equipment according to claim 4, characterized in that: The method of performing multi-objective parameter optimization calculation on the dense layer response parameter set, the conventional layer response parameter set and the mixed layer response parameter set according to the logging data deviation characteristic spectrum and the response anomaly type matrix to obtain a layered calibration parameter sequence includes: Performing feature fusion operation on the well logging data deviation characteristic spectrum and the response anomaly type matrix to obtain a comprehensive feature matrix of the well logging data, and performing pressure response optimization calculation on the dense layer response parameter set based on the comprehensive feature matrix of the well logging data to obtain a dense layer pressure calibration parameter group; Performing mobility compensation calculation on the dense layer pressure calibration parameter group to obtain a dense layer mobility calibration parameter group, and performing pressure gradient optimization calculation on the conventional layer response parameter set according to the comprehensive characteristic matrix of the logging data to obtain a conventional layer pressure mobility calibration parameter group; Based on the comprehensive characteristic matrix of the logging data, multi-parameter coupling optimization is performed on the mixed layer response parameter set to obtain a mixed layer pressure and fluidity calibration parameter group; The dense layer pressure calibration parameter group, the dense layer fluidity calibration parameter group, the conventional layer pressure fluidity calibration parameter group and the mixed layer pressure fluidity calibration parameter group are input into a multi-layer response optimizer for parameter collaborative operation to obtain a multi-layer calibration parameter matrix; the multi-layer response optimizer comprises: a multi-objective optimization algorithm processing unit, a parameter collaborative adjustment unit and a response characteristic evaluation unit, which realizes the global optimal response characteristic of the logging equipment under complex formation conditions by optimizing the response parameters of multiple formations in parallel; An interlayer response compensation calculation is performed on the multi-layer calibration parameter matrix to obtain a calibration compensation coefficient sequence, and the calibration compensation coefficient sequence is substituted into a parameter mapping model for hierarchical configuration to obtain a hierarchical calibration parameter sequence; the parameter mapping model includes: a nonlinear parameter mapping unit, a dynamic feature extraction unit and a parameter reconstruction unit, which are used to establish a mapping relationship between the response parameters of the logging equipment and the calibration parameters, so as to achieve accurate reconstruction of the response characteristics and parameter optimization.

6. The response characteristic calibration method of the logging equipment according to claim 5, characterized in that: Based on the layered calibration parameter sequence, a response benchmark analysis is performed in sequence through a formation permeability evaluation matrix, a response gain analysis is performed through a pressure gradient evaluation matrix, and a response compensation analysis is performed through a mobility attenuation evaluation matrix to obtain a multi-layer response calibration instruction, including: Performing a permeability analysis operation on the dense layer response parameter set in the layered calibration parameter sequence to obtain a dense layer reference response sequence; performing a permeability analysis operation on the conventional layer response parameter set in the layered calibration parameter sequence to obtain a conventional layer reference response sequence; performing a permeability analysis operation on the mixed layer response parameter set in the layered calibration parameter sequence to obtain a mixed layer reference response sequence; Based on the dense layer reference response sequence, the conventional layer reference response sequence and the mixed layer reference response sequence, gradient calculation is performed on the pressure measurement data to obtain a pressure response gain matrix; According to the pressure response gain matrix, a setting area coefficient is dynamically corrected and calculated to obtain a setting response correction coefficient, and based on the pressure response gain matrix and the setting response correction coefficient, a flow rate compensation operation is performed on a suction rate coefficient to obtain a flow rate attenuation correction coefficient; The setting response correction coefficient and the fluidity attenuation correction coefficient are input into a parameter co-processor for joint optimization to obtain a multi-layer calibration control matrix, and an instruction conversion operation is performed on the multi-layer calibration control matrix to obtain a multi-layer response calibration instruction; the parameter co-processor includes: a multi-channel parameter fusion unit, a dynamic weight allocation unit and a parameter optimization unit, which realizes the joint optimization and dynamic adjustment of multiple parameters of the logging equipment by co-processing pressure response parameters and fluidity response parameters.

7. The response characteristic calibration method of the logging equipment according to claim 6, characterized in that: The multi-layer response calibration instruction is subjected to comparative tests under standard permeability conditions, variable pressure permeability conditions and composite permeability conditions to obtain a pressure-fluidity dual-parameter calibration curve, including: The multi-layer response calibration instruction is subjected to a standard working condition response test through a first permeability controller to obtain a first pressure response data group and a first fluidity response data group, and based on the first pressure response data group and the first fluidity response data group, a first parameter calibration operation is performed on the setting area coefficient to obtain a standard working condition calibration parameter sequence; The multi-layer response calibration instruction is subjected to a variable pressure condition response test through a second permeability controller to obtain a second pressure response data group and a second fluidity response data group, and a suction rate correction calculation is performed on the second pressure response data group and the second fluidity response data group based on the standard condition calibration parameter sequence to obtain a variable pressure condition compensation curve; The multi-layer response calibration instruction is subjected to a composite working condition response test through a third permeability controller to obtain a third pressure response data group and a third fluidity response data group, and a dual parameter calibration operation is performed on the third pressure response data group and the third fluidity response data group according to the standard working condition calibration parameter sequence and the variable pressure working condition compensation curve to obtain a composite working condition response parameter matrix; Based on the composite working condition response parameter matrix, the pressure response data is nonlinearly fitted by a first curve fitter, and the fluidity response data is nonlinearly fitted by a second curve fitter to obtain a dual-channel calibration feature sequence, and the dual-channel calibration feature sequence is input into a parameter fusion network for pressure-fluidity coupling analysis to obtain a pressure-fluidity dual-parameter calibration curve; the parameter fusion network includes: a multimodal feature extraction layer, a feature interaction layer and a parameter fusion layer, wherein the feature extraction layer adopts a dual-channel structure to process pressure and fluidity data respectively, the feature interaction layer realizes deep fusion of dual-channel features, and the parameter fusion layer outputs the final calibration parameters.

8. A response characteristic calibration device for a well logging device, characterized in that: For implementing the steps of the method according to any one of claims 1 to 7, the device comprises: A modeling module is used to perform dual-parameter response modeling on the setting area coefficient, suction rate coefficient and pressure measurement range of the logging equipment to obtain a pressure-fluidity coupling response model; A configuration module, for performing interval configuration of data acquisition parameters and signal filtering parameters of the logging equipment according to the pressure-fluidity coupling response model, to obtain a dense layer response parameter set, a conventional layer response parameter set, and a mixed layer response parameter set; A classification module is used to input the logging pressure data sequence and the mobility data sequence into a dual-channel deep neural network for dynamic feature extraction and response mode classification to obtain the logging data deviation characteristic spectrum and response anomaly type matrix; A calculation module, configured to perform multi-objective parameter optimization calculation on the dense layer response parameter set, the conventional layer response parameter set and the mixed layer response parameter set according to the logging data deviation characteristic spectrum and the response anomaly type matrix, to obtain a hierarchical calibration parameter sequence; An analysis module, for performing response benchmark analysis through a formation permeability evaluation matrix, response gain analysis through a pressure gradient evaluation matrix, and response compensation analysis through a mobility attenuation evaluation matrix based on the layered calibration parameter sequence, to obtain a multi-layer response calibration instruction; The test module is used to perform comparative tests on the multi-layer response calibration instructions under standard permeability conditions, variable pressure permeability conditions and composite permeability conditions to obtain a pressure-fluidity dual-parameter calibration curve.

9. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images