A device and method for inverting the leakage orifice size of a tubular tube based on multi-point acoustic response.
The multi-point acoustic response tube leakage aperture inversion device solves the problems of monitoring distance sensitivity and loss of absolute amplitude information in the existing technology, realizes stable aperture inversion under different operating conditions, reduces errors and improves detection accuracy.
Patent Information
- Application Number
- CN202611428626.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-15
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2046-09-15
AI Technical Summary
Existing acoustic leak detection technologies are sensitive to monitoring distance and leak conditions. Single-point responses are easily affected, and multi-point responses are difficult to form stable aperture inversion results. Furthermore, absolute amplitude information is easily lost during normalization, and extrapolation from fixed geometric models is inaccurate.
A tube leakage aperture inversion device based on multi-point acoustic response is adopted. By fixing the leakage location and array geometry, and combining the high-pressure gas supply and differential pressure control system, an acoustic sensor array and array synchronous data acquisition device are used to construct a multi-point acoustic response matrix. Unified dimension conversion and filtering are performed to retain absolute amplitude information. The final aperture is output by multi-window robust fusion.
It achieves stable and accurate aperture inversion under different working conditions, reduces the impact of monitoring distance and local disturbances, retains aperture-related absolute amplitude information, avoids errors caused by inconsistent model parameters, and provides a single and stable aperture result.
Smart Images

Figure CN122921494B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of oil and gas development technology, and in particular to a tubing leakage orifice inversion device and method based on multi-point acoustic response. Background Technology
[0002] Oil and gas production tubing is a crucial fluid channel connecting downhole reservoirs to surface production and transportation systems. Deep, ultra-deep, and high-pressure gas well production tubing is subjected to high temperatures, high pressures, high flow rates, and complex media over long periods, making it susceptible to perforation, cracks, or irregular leakage channels due to corrosion, erosion, fatigue, thread seal degradation, and localized mechanical damage. Tubing leakage can cause abnormal pressure inside and outside the tubing, production losses, and annular pressure. The leakage orifice diameter, or the area equivalent leakage orifice diameter that characterizes the size of a non-circular opening, is an important parameter for evaluating the severity of leakage.
[0003] Methods such as wellhead pressure and flow analysis, temperature logging, and pressure logging can be used to detect anomalies or assist in localization, but they are easily affected by changes in formation supply capacity, wellbore flow conditions, and production regimes, and have limited ability to continuously and quantitatively characterize the diameter of small leaks. Acoustic detection has the advantages of high sensitivity, non-invasiveness, and suitability for dynamic monitoring. When the leaking medium passes through the leaking orifice under the action of the tubing-annular pressure difference, it generates high-speed jets, vortices, and pressure pulsations. The resulting acoustic response contains leak-scale information in its amplitude, waveform, and frequency distribution.
[0004] Existing acoustic leak detection technologies mostly employ single-point acoustic responses for leak identification, location, or severity assessment. Single-point responses are susceptible to variations in monitoring distance, propagation attenuation, near-field jet disturbances, leak pressure differential, sensor installation status, and background noise. The same aperture may produce different acoustic amplitudes and energy distributions under different pressure differentials or monitoring distances, making it difficult to generate stable, continuous aperture inversion results. Even with multiple sensors, if each channel is processed separately, or if the channel order, array position, and processing parameters are inconsistent between the calibration and testing phases, the spatial correspondence between multi-point responses will be disrupted.
[0005] In acoustic signal data processing, root mean square normalization or standardization is often used to reduce amplitude scale differences between different recordings. However, variations in the leakage aperture itself cause changes in the acoustic absolute amplitude and channel energy. If only the normalized signal is retained, the absolute amplitude information related to the aperture will be weakened. On the other hand, continuous acoustic recordings generate multiple window predictions after passing through a sliding window. If only a single window result is output or all windows are simply averaged, it is easily affected by transient interference and abnormal windows, and cannot form a single stable aperture result with applicable range conditions.
[0006] Furthermore, changes in the location of the leak or the geometry of the sensor array alter the propagation distances and channel correspondences between each acoustic channel and the leak source. Applying this model directly to other leak locations or array configurations, calibrated only for one set of geometric configurations, leads to unverified extrapolations.
[0007] Therefore, it is necessary to establish a tubular leakage orifice inversion device and method based on fixed leakage location and fixed array geometry, capable of calibrating known orifice diameters under graded differential pressure conditions, retaining orifice diameter-related absolute amplitude information, solidifying complete models and processing parameters, and performing configuration and operating condition applicability verification and multi-window robust fusion during the test phase. Summary of the Invention
[0008] The purpose of this invention is to address the aforementioned deficiencies in the existing technology by providing a tube leakage aperture inversion device and method based on multi-point acoustic response. This addresses the problems of single-point acoustic response being sensitive to monitoring distance and leakage conditions, easy loss of aperture-related absolute amplitude information during normalization, inconsistencies between input structure and processing parameters in the calibration and test stages, the use of fixed geometric models for extrapolation of other configurations, and the inability of multiple window predictions to form a single stable output.
[0009] The present invention mentions a tubing leakage aperture inversion device based on multi-point acoustic response, the technical solution of which includes a computer system, wherein it also includes a high-pressure gas supply and differential pressure control system, a tubing leakage experimental system, an acoustic sensor array, and an array synchronous data acquisition unit; The high-pressure gas supply and differential pressure control system includes a high-pressure gas source, a tubing-side inlet pressure regulating unit, a tubing-side pressure stabilizing unit, a tubing-side pressure detection unit, and a tubing-side check valve. The output end of the high-pressure gas source is connected to one end of the simulated production tubing through pipelines and the tubing-side inlet pressure regulating unit, the tubing-side pressure stabilizing unit, the tubing-side pressure detection unit, and the tubing-side check valve. The tubing leakage test system includes a simulated production tubing, an annular cylinder sleeved outside the simulated production tubing, and replaceable leakage hole modules set at fixed axial and fixed circumferential positions on the simulated production tubing. A first sealing end cap is installed at one end of the simulated production tubing, and a second sealing end cap is installed at the other end. An experimental device support frame is installed on the lower side of the simulated production tubing. The acoustic sensor array comprises components fixedly arranged along the axial direction of the simulated production tubing and numbered according to a predetermined channel sequence. N One acoustic sensor, N ≥3; The array synchronous data acquisition unit synchronously acquires the acoustic responses of N acoustic sensors and synchronously obtains the internal pressure and annular pressure of the simulated production tubing.
[0010] Preferably, the above-mentioned high-pressure gas supply and differential pressure control system further includes an annular back pressure regulating unit, an annular pressure stabilizing unit, an annular pressure detection unit, a leaking gas mass flow meter, and a tubing-side outlet back pressure regulating unit. The other end of the simulated production tubing is connected to the tubing-side outlet back pressure regulating unit via a pipeline. The end of the annular cylinder is connected to the annular back pressure regulating unit, the annular pressure stabilizing unit, the annular pressure detection unit, and the leaking gas mass flow meter via a pipeline.
[0011] Preferably, the aforementioned tubular side pressure detection unit, annular side pressure detection unit, and leaking gas mass flow meter are respectively connected to the array synchronous data acquisition unit via signal lines. The leaking gas mass flow meter is used to determine the stability of the calibration record and verify the leakage status of a known orifice.
[0012] Preferably, the replaceable leakage hole module includes an internal threaded connector, a leakage hole bolt plug, a through leakage hole, a sealing ring, and a positioning end face. The internal threaded connector is fixedly connected to the wall of the simulated production tubing and extends into the inner cavity of the annular cylinder. After the leakage hole bolt plug is screwed into the internal threaded connector, the installation depth is limited by the positioning end face. The sealing ring is used to prevent leakage from the thread gap. A through leakage hole is provided in the leakage hole bolt plug to connect the simulated production tubing and the annular cylinder, which is used to simulate tubing leakage.
[0013] Preferably, the above-mentioned leakage hole bolt plugs are of various models, and the through leakage holes of different models of leakage hole bolt plugs have different hole diameters or different cross-sectional shapes.
[0014] Preferably, the acoustic sensor array includes a first acoustic sensor, a second acoustic sensor, a third acoustic sensor, a fourth acoustic sensor, a fifth acoustic sensor, and a sixth acoustic sensor. The first acoustic sensor, the second acoustic sensor, the third acoustic sensor, the fourth acoustic sensor, the fifth acoustic sensor, and the sixth acoustic sensor are installed on the side wall of the annular cylinder and numbered in a predetermined channel sequence, and are connected to the array synchronization data acquisition unit through signal lines.
[0015] Preferably, the above-mentioned computer system adopts a computer aperture inversion analysis system, which includes a steady state determination and synchronous acquisition triggering unit, an acoustic response preprocessing and window segmentation unit, a multi-point acoustic response matrix construction unit, an amplitude information extraction and retention unit, a unified calibration sample library and model training parameter storage unit, and a test sample, range verification and multi-window fusion output unit. Each unit is implemented by the processor executing program instructions in the memory, and cooperates with the multi-channel acoustic response and pressure condition parameters acquired by the array synchronous data acquisition device.
[0016] The inversion method of the tubular leakage aperture inversion device based on multi-point acoustic response mentioned in this invention includes the following process: S1, First, install the replaceable leak hole module, install the acoustic sensor array and predetermined channel sequence, and fix the geometric configuration markings; wherein, the fixed geometric configuration markings include at least the center coordinates of the leak hole, N The coordinates of each acoustic sensor, the sequence of each acoustic channel, the correspondence between each acoustic channel and the monitoring distance, the sensor calibration coefficients, and the acquisition gain; when the fixed geometric configuration changes, the aperture inversion model corresponding to the original fixed geometric configuration is stopped from being called and a recalibration prompt is output; S2, install leakage hole bolt plugs with different orifice diameters or different cross-sectional shapes in sequence to form multiple tubing-annular leakage differential pressure levels, and simultaneously collect multi-channel acoustic response and pressure condition parameters after meeting the calibration stability criteria; S3 performs unified dimension conversion, filtering, synchronization window segmentation, and channel-by-channel mean removal on the multi-channel acoustic response to construct a multi-point acoustic response matrix for each time window. X k ; S4, Calculate the overall root mean square magnitude R k and the root mean square amplitude eigenvectors of each channel E k ,use R k right X k Normalize and retain R k and E k ; S5 associates the normalized multi-point acoustic response matrix, the overall root mean square amplitude, the channel root mean square amplitude eigenvector, the pressure condition parameters, and the known aperture labels to form a unified calibration sample; S6 divides the training subset, validation subset, and test subset according to independent continuous records, trains the aperture inversion model, and saves the model, signal processing parameters, fixed geometric configuration identifier, calibration scope, and multi-window fusion rules. S7. Install the bolt plug of the leak hole to be tested with unknown hole diameter or hole diameter to be determined after damage at the same fixed position, verify the fixed geometric configuration, and when the stability criterion to be tested is met and the pressure condition is within the calibration applicable range, call the parameters saved in step S6 and construct multiple window models to be tested according to the same processing flow as steps S3 and S4. S8 obtains multiple window aperture prediction values, removes out-of-range prediction values and abnormal window prediction values, robustly fuses the remaining prediction values, and outputs the final leakage aperture, the number of effective windows, and the validity status of the result.
[0017] Preferably, in step S5, the first kThe aperture inversion model input for each time window is I k ={ , R k , E k , C k},in To normalize the multi-point acoustic response matrix, R k This represents the overall root mean square magnitude. E k This is the eigenvector of the channel root mean square magnitude. C k It is a vector of operating parameters including the internal pressure of the tubing, the annulus pressure, and the tubing-annulus leakage pressure difference; R k , E k and C k All transformations are performed using scaling parameters determined only by a subset of the training data; I k This represents the aperture inversion model input corresponding to the k-th time window.
[0018] Preferably, in step S8, the median and median absolute deviation are calculated for the predicted window aperture values within the calibrated aperture range. Predicted window aperture values that deviate from the median by more than a preset multiple of the median absolute deviation are identified as abnormal predicted window aperture values. The indices of the remaining windows are then used to form a valid window index set. V When the number of valid windows reaches a preset minimum, a set is used. V The median of the corresponding predicted window aperture values is used as the final leakage aperture. When the pressure condition to be measured exceeds the range of the calibrated pressure condition, the fixed geometric configuration to be measured is inconsistent with the saved fixed geometric configuration identifier, the state to be measured does not meet the stability criterion, the number of effective windows is less than the preset minimum number, or the overall root mean square amplitude, channel root mean square amplitude or pressure parameter exceeds the calibration applicable range, the extrapolated value of the model will not be used as a valid inversion result, and a prompt for re-acquisition or recalibration will be output.
[0019] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention, by fixing the axial and circumferential leakage installation positions, fixing the acoustic array geometry and channel-distance binding relationship, ensures that calibration records with different known apertures and the test records have a consistent spatial configuration, reducing response differences caused by changes in leakage location and channel misalignment; 2. This invention uses the synchronous acoustic responses at multiple fixed monitoring distances in combination within the same time window, enabling aperture inversion to utilize multi-channel temporal waveforms and spatial propagation information simultaneously, thereby reducing the impact of single-point local jet disturbances, propagation attenuation, and background noise on the results; 3. While using overall root mean square amplitude normalization to unify the input scale, this invention independently retains the overall root mean square amplitude and the root mean square amplitude of each channel before normalization, taking into account relative waveform characteristics, spatial amplitude distribution and aperture-related absolute amplitude information. 4. This invention saves the model, signal processing parameters, fixed geometric configuration identifier, scale transformation parameters, calibration applicable range and fusion rules as a unified parameter package, which can be directly called in the test stage, avoiding inconsistencies in input structure or processing parameters between the calibration stage and the test stage; 5. This invention divides the model dataset into independent and continuous records to avoid overlapping windows of the same record entering different data subsets, which would lead to an overestimation of the evaluation. Through out-of-range judgment, abnormal window removal, and robust fusion of multiple windows, a single final aperture result with effective states can be formed. 6. This invention uses the mass flow rate of leaked gas for calibration stability and anomaly record verification, instead of using it as input to the acoustic aperture inversion model. This allows the model output to still be obtained from the acoustic response and pressure conditions, while improving the reliability of the calibration samples. Attached Figure Description
[0020] Figure 1 This is a schematic diagram of the device structure of the present invention; Figure 2 This is an enlarged schematic diagram of the installation of the replaceable leak hole module; Figure 3 This is a flowchart of a computer aperture inversion analysis system; In the diagram: High-pressure gas source 101, inlet pressure regulating unit 102, pressure stabilizing unit 103, pressure detection unit 104, check valve 105, back pressure regulating unit 106, pressure stabilizing unit 107, pressure detection unit 108, leaking gas mass flow meter 109, outlet back pressure regulating unit 110, first sealing end cap 201, simulated production tubing 202, annular cylinder 203, replaceable leaking hole module 204, experimental device support frame 205, second sealing end cap 206, first acoustic sensor 301, second acoustic sensor 302, third acoustic sensor 303, fourth acoustic sensor 304, fifth acoustic sensor 305, sixth acoustic sensor 306, array synchronous data acquisition unit 401, computer aperture inversion analysis system 402; 2041 Internal threaded connector, 2042 Leakage hole bolt plug, 2043 Through leakage hole, 2044 Sealing ring, 2045 Positioning end face. Detailed Implementation
[0021] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.
[0022] Example 1, referring to Figures 1-3 The present invention mentions a tubing leakage aperture inversion device based on multi-point acoustic response, which includes a computer system, a high-pressure gas supply and differential pressure control system, a tubing leakage experimental system, an acoustic sensor array, and an array synchronous data acquisition unit 401. The high-pressure gas supply and differential pressure control system includes a high-pressure gas source 101, a column-side inlet pressure regulating unit 102, a column-side pressure stabilizing unit 103, a column-side pressure detection unit 104, and a column-side one-way valve 105. The output end of the high-pressure gas source 101 is connected to one end of the simulated production column 202 through pipelines and the column-side inlet pressure regulating unit 102, the column-side pressure stabilizing unit 103, the column-side pressure detection unit 104, and the column-side one-way valve 105. The tubing leakage test system includes a simulated production tubing 202, an annular cylinder 203 sleeved outside the simulated production tubing 202, and replaceable leakage hole modules 204 set at fixed axial and fixed circumferential positions on the simulated production tubing 202. A first sealing end cap 201 is installed at one end of the simulated production tubing 202, and a second sealing end cap 206 is installed at the other end. An experimental device support frame 205 is installed on the lower side of the simulated production tubing 202. The acoustic sensor array comprises components fixedly arranged along the axial direction of the simulated production column 202 and numbered according to a predetermined channel sequence. N One acoustic sensor, N ≥3; The array synchronous data acquisition unit 401 synchronously acquires the acoustic responses of N acoustic sensors and synchronously obtains the internal pressure and annular pressure of the simulated production column 202.
[0023] The aforementioned high-pressure gas supply and differential pressure control system further includes an annular back pressure regulating unit 106, an annular pressure stabilizing unit 107, an annular pressure detection unit 108, a leaking gas mass flow meter 109, and a tubing-side outlet back pressure regulating unit 110. The other end of the simulated production tubing 202 is connected to the tubing-side outlet back pressure regulating unit 110 via a pipeline. The end of the annular cylinder 203 is connected to the annular back pressure regulating unit 106, the annular pressure stabilizing unit 107, the annular pressure detection unit 108, and the leaking gas mass flow meter 109 via a pipeline.
[0024] The aforementioned tubular side pressure detection unit 104, annular side pressure detection unit 108, and leak gas mass flow meter 109 are respectively connected to the array synchronous data acquisition unit 401 via signal lines. The leak gas mass flow meter 109 is used to determine the stability of the calibration record and verify the leakage status of the known orifice.
[0025] Reference Figure 2 The replaceable leakage hole module 204 mentioned in this invention includes an internal threaded connector 2041, a leakage hole bolt plug 2042, a through leakage hole 2043, a sealing ring 2044, and a positioning end face 2045. The internal threaded connector 2041 is fixedly connected to the wall of the simulated production tubing 202 and extends into the inner cavity of the annular cylinder 203. After the leakage hole bolt plug 2042 is screwed into the internal threaded connector 2041, the installation depth is limited by the positioning end face 2045. The sealing ring 2044 is used to prevent leakage from the thread gap. A through leakage hole 2043 is provided in the leakage hole bolt plug 2042 to connect the simulated production tubing 202 and the annular cylinder 203 to simulate tubing leakage.
[0026] Among them, the above-mentioned leakage hole bolt plug 2042 adopts various models, and the through leakage through hole 2043 of different models of leakage hole bolt plug 2042 adopts different hole diameters or different cross-sectional shapes.
[0027] The aforementioned acoustic sensor array includes a first acoustic sensor 301, a second acoustic sensor 302, a third acoustic sensor 303, a fourth acoustic sensor 304, a fifth acoustic sensor 305, and a sixth acoustic sensor 306. The first acoustic sensor 301, the second acoustic sensor 302, the third acoustic sensor 303, the fourth acoustic sensor 304, the fifth acoustic sensor 305, and the sixth acoustic sensor 306 are installed on the side wall of the annular cylinder 203 and numbered in a predetermined channel sequence, and are connected to the array synchronization data acquisition unit 401 via signal lines.
[0028] Reference Figure 3 The computer system mentioned in this invention employs a computer aperture inversion analysis system 402. The computer aperture inversion analysis system 402 includes a stable state determination and synchronous acquisition triggering unit, an acoustic response preprocessing and window segmentation unit, a multi-point acoustic response matrix construction unit, an amplitude information extraction and retention unit, a unified calibration sample library and model training parameter storage unit, and a test sample, range verification and multi-window fusion output unit. Each unit is implemented by the processor executing program instructions in the memory, and cooperates with the multi-channel acoustic response and pressure condition parameters acquired by the array synchronous data acquisition unit 401.
[0029] The inversion method of the tubular leakage aperture inversion device based on multi-point acoustic response mentioned in this invention includes the following process: S1, First, install the replaceable leak hole module 204, install the acoustic sensor array and predetermined channel sequence, and fix the geometric configuration markings; wherein, the fixed geometric configuration markings include at least the center coordinates of the leak hole, N The coordinates of each acoustic sensor, the sequence of each acoustic channel, the correspondence between each acoustic channel and the monitoring distance, the sensor calibration coefficients, and the acquisition gain; when the fixed geometric configuration changes, the aperture inversion model corresponding to the original fixed geometric configuration is stopped from being called and a recalibration prompt is output; S2, install leakage hole bolt plugs 2042 with different orifice diameters or different cross-sectional shapes in sequence to form multiple tubing-annulus leakage differential pressure levels, and simultaneously collect multi-channel acoustic response and pressure condition parameters after meeting the calibration stability criteria; Specifically, while maintaining the installation position of the leak hole bolt plug 2042, the center position of the through leak hole 2043, and the leakage direction, the leak hole bolt plugs 2042 with different known hole diameters are replaced sequentially. The tubing-annulus leakage pressure difference is: (1), In the formula, p t To simulate the internal pressure of production tubing 202, For the annular pressure, Δ p For the tubing-annulus leakage pressure differential, multiple independent continuous records were collected for each combination of orifice diameter and leakage pressure differential. Within a preset stability assessment period, the stability of the internal pressure of the tubing, the annular pressure, and the calibrated leak gas mass flow rate are assessed. For any state variable z, its relative volatility is calculated based on the sampled values within the preset stability assessment period using the following formula: (2), In the formula, Z j Let M be the j-th sampled value of state variable Z within the preset stability determination time period, and M be the number of sampling points within the preset stability determination time period. r Z Let Z be the relative volatility of the state variable. When the relative volatility of the internal pressure of the tubing, the annular pressure, and the mass flow rate of the leaked gas are not greater than the corresponding preset stability thresholds, the continuous record is determined as a valid calibration record. During the effective stable period, the array synchronous data acquisition unit 401 synchronously acquires the acoustic response, internal pressure of the tubing and annular pressure of N acoustic channels according to a unified sampling clock, and synchronously records the mass flow rate of the leaking gas during the calibration phase; wherein, the mass flow rate of the leaking gas is only used for stable state determination, abnormal calibration record identification and known aperture leakage state verification, and is not used as the input parameter of the aperture inversion model. For a non-circular through-hole 2043, when it is necessary to represent its geometric opening size on a uniform scale, the area-equivalent leakage orifice diameter is calculated based on the known leakage cross-sectional area A: (3), In the formula, d e To obtain the area equivalent leakage aperture, for leakage holes with different cross-sectional shapes, a model parameter package containing corresponding hole shape calibration samples is used, or a separate aperture inversion model is trained for the corresponding hole shape. The aperture inversion model trained only by circular leakage hole samples is not directly called. S3 performs unified dimension conversion, filtering, synchronization window segmentation, and channel-by-channel mean removal on the multi-channel acoustic response to construct a multi-point acoustic response matrix for each time window. X k ; Specifically, based on the calibration coefficients of each acoustic sensor and the acquisition gain of the corresponding acoustic channel, the raw signal acquired by the array synchronous data acquisition unit 401 is converted into an acoustic response of uniform type and uniform dimensions; the acoustic response of each channel is filtered using preset filtering parameters, and the N acoustic channels are segmented into synchronous windows according to the same window start point, window length L, and window movement step size H; within each time window, the window mean of each acoustic channel is subtracted to eliminate DC components and channel baseline offset; as a specific implementation, the acoustic sampling frequency is set to 20kHz, a bandpass filter of 100~9000Hz is used, the number of acoustic channels N=6, the window length L=2048, and the window movement step size H=512; According to the predetermined acoustic channel order, the preprocessed acoustic responses of N acoustic channels within the Kth time window are constructed into a multi-point acoustic response matrix according to equation (4): (4), In the formula, X k This is the multi-point acoustic response matrix corresponding to the k-th time window; x i,k ( j ) is the first i The preprocessed acoustic response values of each acoustic channel at the j-th sampling point within the k-th time window; N is the number of acoustic channels; L is the window length; i =1,2,..., N , j =1,2,..., L Each row of the matrix corresponds to an acoustic response sequence at a fixed monitoring location, and each column corresponds to the synchronous acoustic response of N acoustic channels at the same sampling time; R represents the set of real numbers; S4, Calculate the overall root mean square magnitude R kand the root mean square amplitude eigenvectors of each channel E k , adopt R k right X k Normalize and retain R k and E k ; The overall root mean square amplitude of all acoustic channels within the k-th time window R k Calculate according to the following formula: (5), Calculate the root mean square amplitude of each acoustic channel separately: (6), (7), In the formula, E i,k Let be the root mean square amplitude of the i-th acoustic channel within the k-th time window. E k This refers to the root mean square amplitude feature vectors of the channels constructed according to a predetermined channel order. The multi-point acoustic response matrix is normalized using the overall root mean square amplitude: (8), In the formula, ε A preset correction amount greater than zero is used to prevent the denominator from being zero or too small. As one specific implementation method... ε Take 10 -8 Normalized multipoint acoustic response matrix Used to characterize the relative waveform features and spatial response differences between channels; before normalization R k and E k The additional amplitude feature is retained separately to avoid the loss of absolute amplitude information related to the leakage aperture during the normalization process. S5 associates the normalized multi-point acoustic response matrix, the overall root mean square amplitude, the channel root mean square amplitude eigenvector, the pressure condition parameters, and the known aperture labels to form a unified calibration sample; The pressure condition parameter vector corresponding to the kth time window is: (9), In the formula, and These are the average values of the internal pressure and annular pressure of the tubing within the k-th time window, respectively. The aperture inversion model input and unified calibration sample for the k-th time window are represented as follows: (10) In the formula, To normalize the multi-point acoustic response matrix, R k This represents the overall root mean square magnitude. E k This is the eigenvector of the channel root mean square magnitude. C k It is a vector of operating parameters including the internal pressure of the tubing, the annulus pressure, and the tubing-annulus leakage pressure difference; R k , E k and C k All transformations are performed using scaling parameters determined only by a subset of the training data; I k S represents the aperture inversion model input corresponding to the k-th time window; k y represents the uniformly calibrated sample corresponding to the k-th time window; k Labels for known leaking orifice diameters; for circular leaking through-holes, y k y is the actual geometric diameter; for a non-circular leakage through-hole with a known cross-sectional shape, y k The area equivalent leakage orifice diameter; The standardized samples are simultaneously associated with continuous record numbers, aperture type, and fixed geometric configuration identifiers for subsequent data subset partitioning and model parameter package management. S6 divides the training subset, validation subset, and test subset according to independent continuous records, trains the aperture inversion model, and saves the model, signal processing parameters, fixed geometric configuration identifier, calibration scope, and multi-window fusion rules. Specifically, an independent continuous record is used as the smallest grouping unit for data subset partitioning. All calibration samples obtained by the sliding window for the same continuous record are only assigned to one subset of the training subset, validation subset, and test subset, avoiding overlapping or highly related time windows from entering the training subset and test subset at the same time. Calculated only from the training subset R k , E k and C k The mean and standard deviation are used to perform scaling transformation; the verification subset, test subset and test sample all directly call the scaling transformation parameters determined by the training subset, without recalculation; As one specific implementation, the aperture inversion model includes a multi-point acoustic feature extraction module, an auxiliary parameter fusion module, and a regression output module. The multi-point acoustic feature extraction module takes a 6×2048 normalized multi-point acoustic response matrix as the main input, and sequentially sets a one-dimensional convolutional layer with 32 output channels and a kernel length of 15, a one-dimensional convolutional layer with 64 output channels and a kernel length of 9, and a one-dimensional convolutional layer with 128 output channels and a kernel length of 5. After each convolutional layer, a batch normalization layer and a ReLU activation layer are set. After the first two convolutional layers, a max pooling layer is set, and after the third convolutional layer, a global average pooling layer is set to obtain 128-dimensional multi-point acoustic features. The overall root mean square amplitude after scaling, the 6-dimensional channel root mean square amplitude feature vector, and the 3-dimensional pressure condition parameter vector are combined to form a 10-dimensional auxiliary parameter. This parameter is then passed through a fully connected layer with 32 neurons and a ReLU activation layer to obtain a 32-dimensional auxiliary feature. This 32-dimensional auxiliary feature is then connected to a 128-dimensional multi-point acoustic feature, and sequentially passed through a fully connected layer with 64 neurons, a ReLU activation layer, a Dropout layer with an inactivation rate of 0.2, and a linear output layer with 1 neuron to obtain a continuous numerical form of the predicted window aperture value. The model training preferably uses the Huber loss function and the Adam optimizer, with an initial learning rate of 1×10⁻⁶. -3 The batch size is set to 64, and the maximum number of training epochs is set to 200. Training stops when the mean absolute error of the validation subset fails to decrease for 20 consecutive epochs, and the model parameters are saved when the mean absolute error of the validation subset is minimized. The aperture inversion model can also use a multi-scale one-dimensional convolutional network, a time-domain-frequency domain dual-branch network, or other regression models capable of handling multi-channel time series, but the model input structure remains I. k ; After training, the model structure and weights, fixed geometric configuration identifiers, acoustic channel order, signal conversion parameters, filtering parameters, window length L, window movement step size H, mean removal rule, and normalization constant are set. ε The scaling transformation parameters, calibration orifice diameter range, calibration pressure operating condition range, overall root mean square amplitude and calibration distribution range of root mean square amplitude of each channel, anomaly window criteria and multi-window fusion rules are all saved as a model parameter package; when training separately for different orifice shapes, the model parameter package corresponding to the corresponding orifice shape is saved separately. S7. Install the bolt plug 2042 of the leak hole to be tested with unknown hole diameter or hole diameter to be determined after damage at the same fixed position, verify the fixed geometric configuration, and when the stability criterion to be tested is met and the pressure condition is within the calibration applicable range, call the parameters saved in step S6 and construct multiple window models to be tested according to the same processing flow as steps S3 and S4. Specifically, during the testing phase, the installation position of the leakage hole, the coordinates of each acoustic sensor, the sequence of acoustic channels, the installation coupling method, and the acquisition gain are kept consistent with those during the calibration phase. First, the fixed geometric configuration identifier corresponding to the test state is obtained and compared with the fixed geometric configuration identifier saved in the model parameter package. If the two are inconsistent, the original aperture inversion model is stopped and a recalibration prompt is output. According to the relative volatility calculation method described in step S2, the stability of the internal pressure and annular pressure of the test string is determined; after the fixed geometric configuration and stable state verification are passed, it is determined whether the internal pressure, annular pressure and leakage pressure difference of the test string are within the calibration pressure range saved in the model parameter package. Then, the conversion parameters, filtering parameters, window length, window movement step size, mean removal rule, normalization constant, and scaling transformation parameters saved in step S6 are called, and multiple test windows are constructed according to the same process as in the calibration stage; the test data is not used to recalculate the scaling transformation parameters; after the window processing is completed, it is further determined whether the overall root mean square magnitude and channel root mean square magnitude of each test window are within the corresponding calibration applicable range; Input the q-th test window that has passed the verification into the model to obtain the predicted window aperture value: (11), In the formula, , θ The model parameters saved in step S6 Let q be the predicted aperture value for the q-th test window; S8 obtains multiple window aperture prediction values, removes out-of-range prediction values and abnormal window prediction values, robustly fuses the remaining prediction values, and outputs the final leakage aperture, number of effective windows, and result validity status. For the Ω predicted aperture values obtained under the same stable test state, the first step is to filter them according to the calibration aperture range stored in the model parameter package, and form a preliminary candidate index set by selecting the window indices that fall within the calibration aperture range: (12) In the formula, d min and d max These are the lower and upper limits of the calibration aperture corresponding to the model, respectively; when Ω is empty, the result is directly judged as invalid. Calculate the median m and median absolute deviation (MAD) of the preliminary candidate window predictions: (13) Outlier windows are removed based on the absolute deviation of the median, and the window indices that satisfy the following formula are combined to form the set of valid window indices V: (14) In the formula, λ To preset the anomaly detection coefficient, ε m This is a correction amount greater than zero, used to prevent the anomaly detection threshold from failing when the absolute deviation of the median is zero; as one specific implementation method, λ Take 3; The effective window number and the final leakage orifice diameter are as follows: (15) When the number of valid windows M is not less than the preset minimum number of valid windows M min At that time, As the final leakage aperture; as a specific implementation method, M min Set the value to 10. When the pressure condition to be measured exceeds the calibration range, the fixed geometric configuration is inconsistent, the state to be measured is unstable, the overall root mean square amplitude or the root mean square amplitude of any channel exceeds the calibration applicable range, the preliminary candidate set is empty, or the number of effective windows is insufficient, the model prediction value will not be used as a valid inversion result, and a prompt for re-acquisition or recalibration will be output. The final output should include at least the final leakage orifice diameter. The output includes the number of effective windows (M), the verification results of fixed geometry configuration, the verification results of pressure operating conditions range, and the validity status of the results. For non-circular leakage holes, the final output value represents the area equivalent leakage hole diameter and simultaneously indicates the hole shape category corresponding to the called model parameter package.
[0030] For the predicted aperture values within the calibrated aperture range, calculate the median and median absolute deviation. Window aperture predicted values that deviate from the median by more than a preset multiple of the median absolute deviation are identified as abnormal window aperture predicted values. The indices of the remaining windows are then used to form a set of valid window indexes. V When the number of valid windows reaches a preset minimum, a set is used. V The median of the corresponding predicted window aperture values is used as the final leakage aperture. When the pressure condition to be measured exceeds the range of the calibrated pressure condition, the fixed geometric configuration to be measured is inconsistent with the saved fixed geometric configuration identifier, the state to be measured does not meet the stability criterion, the number of effective windows is less than the preset minimum number, or the overall root mean square amplitude, channel root mean square amplitude or pressure parameter exceeds the calibration applicable range, the extrapolated value of the model will not be used as a valid inversion result, and a prompt for re-acquisition or recalibration will be output.
[0031] The above description is merely a partial preferred embodiment of the present invention. Any person skilled in the art can modify the above-described technical solutions or modify them into equivalent technical solutions. Therefore, any simple modifications or equivalent transformations made based on the technical solutions of the present invention fall within the scope of protection claimed by the present invention.
Claims
1. A tubular leakage orifice inversion device based on multi-point acoustic response, comprising a computer system, characterized in that: It also includes a high-pressure gas supply and differential pressure control system, a tubing leakage test system, an acoustic sensor array, and an array synchronous data acquisition unit (401). The high-pressure gas supply and differential pressure control system includes a high-pressure gas source (101), a column-side inlet pressure regulating unit (102), a column-side pressure stabilizing unit (103), a column-side pressure detection unit (104), and a column-side check valve (105). The output end of the high-pressure gas source (101) is connected to one end of the simulated production column (202) through pipelines and the column-side inlet pressure regulating unit (102), the column-side pressure stabilizing unit (103), the column-side pressure detection unit (104), and the column-side check valve (105). The tubing leakage test system includes a simulated production tubing (202), an annular cylinder (203) sleeved outside the simulated production tubing (202), and replaceable leakage hole modules (204) set at fixed axial and fixed circumferential positions on the simulated production tubing (202). A first sealing end cap (201) is installed at one end of the simulated production tubing (202), and a second sealing end cap (206) is installed at the other end. An experimental device support frame (205) is installed on the lower side of the simulated production tubing (202). The acoustic sensor array comprises components fixedly arranged along the axial direction of the simulated production column (202) and numbered according to a predetermined channel sequence. N One acoustic sensor, N ≥3; The array synchronous data acquisition unit (401) synchronously acquires the acoustic responses of N acoustic sensors and synchronously obtains the internal pressure and annular pressure of the simulated production tubing (202); The replaceable leakage hole module (204) includes an internal threaded connector (2041), a leakage hole bolt plug (2042), a through leakage hole (2043), a sealing ring (2044), and a positioning end face (2045). The internal threaded connector (2041) is fixedly connected to the wall of the simulated production tubing (202) and extends into the inner cavity of the annular cylinder (203). After the leakage hole bolt plug (2042) is screwed into the internal threaded connector (2041), the installation depth is limited by the positioning end face (2045). The sealing ring (2044) is used to prevent leakage from the thread gap. A through leakage hole (2043) is provided in the leakage hole bolt plug (2042) to connect the simulated production tubing (202) and the annular cylinder (203) to simulate tubing leakage. The computer system adopts a computer aperture inversion analysis system (402), which includes a stable state determination and synchronous acquisition triggering unit, an acoustic response preprocessing and window segmentation unit, a multi-point acoustic response matrix construction unit, an amplitude information extraction and retention unit, a unified calibration sample library and model training parameter storage unit, and a test sample, range verification and multi-window fusion output unit. Each unit performs its function by executing program instructions in the memory by the processor, and coordinates with the multi-channel acoustic response and pressure condition parameters collected by the array synchronous data acquisition unit (401).
2. The tubular leakage aperture inversion device based on multi-point acoustic response according to claim 1, characterized in that: The high-pressure gas supply and differential pressure control system also includes an annular back pressure regulating unit (106), an annular pressure stabilizing unit (107), an annular pressure detection unit (108), a leaking gas mass flow meter (109), and a tubing-side outlet back pressure regulating unit (110). The other end of the simulated production tubing (202) is connected to the tubing-side outlet back pressure regulating unit (110) via a pipeline. The end of the annular cylinder (203) is connected to the annular back pressure regulating unit (106), the annular pressure stabilizing unit (107), the annular pressure detection unit (108), and the leaking gas mass flow meter (109) via a pipeline.
3. The tubular leakage aperture inversion device based on multi-point acoustic response according to claim 2, characterized in that: The tubular side pressure detection unit (104), the annular side pressure detection unit (108), and the leak gas mass flow meter (109) are connected to the array synchronous data acquisition unit (401) via signal lines. The leak gas mass flow meter (109) is used to determine the stability of the calibration record and verify the leakage status of the known orifice.
4. The tubular leakage aperture inversion device based on multi-point acoustic response according to claim 3, characterized in that: The leak hole bolt plug (2042) is available in various models, and the through leak hole (2043) of different models of leak hole bolt plug (2042) has different hole diameters or different cross-sectional shapes.
5. The tubular leakage aperture inversion device based on multi-point acoustic response according to claim 4, characterized in that: The acoustic sensor array includes a first acoustic sensor (301), a second acoustic sensor (302), a third acoustic sensor (303), a fourth acoustic sensor (304), a fifth acoustic sensor (305), and a sixth acoustic sensor (306). The first acoustic sensor (301), the second acoustic sensor (302), the third acoustic sensor (303), the fourth acoustic sensor (304), the fifth acoustic sensor (305), and the sixth acoustic sensor (306) are installed on the side wall of the annular cylinder (203) and numbered in a predetermined channel sequence, and are connected to the array synchronization data acquisition unit (401) through signal lines.
6. An inversion method using the multi-point acoustic response-based tubular leakage aperture inversion device as described in claim 5, characterized in that: The process includes the following: S1, First, install the replaceable leak hole module (204), install the acoustic sensor array and predetermined channel sequence, and fix the geometric configuration markings; wherein, the fixed geometric configuration markings include at least the center coordinates of the leak hole, N The coordinates of each acoustic sensor, the sequence of each acoustic channel, the correspondence between each acoustic channel and the monitoring distance, the sensor calibration coefficients, and the acquisition gain; when the fixed geometric configuration changes, the aperture inversion model corresponding to the original fixed geometric configuration is stopped from being called and a recalibration prompt is output; S2, install leakage hole bolt plugs (2042) with different orifice diameters or different cross-sectional shapes in sequence, and form multiple tubing-annular leakage differential pressure levels. After meeting the calibration stability criteria, synchronously collect multi-channel acoustic response and pressure condition parameters. S3 performs unified dimension conversion, filtering, synchronization window segmentation, and channel-by-channel mean removal on the multi-channel acoustic response to construct a multi-point acoustic response matrix for each time window. X k ; S4, Calculate the overall root mean square magnitude R k and the root mean square amplitude eigenvectors of each channel E k ,use R k right X k Normalize and retain R k and E k ; S5 associates the normalized multi-point acoustic response matrix, the overall root mean square amplitude, the channel root mean square amplitude eigenvector, the pressure condition parameters, and the known aperture labels to form a unified calibration sample; S6 divides the training subset, validation subset, and test subset according to independent continuous records, trains the aperture inversion model, and saves the model, signal processing parameters, fixed geometric configuration identifier, calibration scope, and multi-window fusion rules. S7, Install the bolt plug (2042) of the leak hole to be tested with unknown hole diameter or hole diameter to be determined after damage at the same fixed position, verify the fixed geometric configuration, and when the stability criterion to be tested is met and the pressure condition is within the calibration applicable range, call the parameters saved in step S6 and construct multiple window models to be tested according to the same processing flow as steps S3 and S4. S8 obtains multiple window aperture prediction values, removes out-of-range prediction values and abnormal window prediction values, robustly fuses the remaining prediction values, and outputs the final leakage aperture, the number of effective windows, and the validity status of the result.
7. The inversion method of the tubular leakage aperture inversion device based on multi-point acoustic response according to claim 6, characterized in that: In step S5, the first k The aperture inversion model input for each time window is I k ={ , R k , E k , C k },in To normalize the multi-point acoustic response matrix, R k This represents the overall root mean square magnitude. E k This is the eigenvector of the channel root mean square magnitude. C k It is a vector of operating parameters including the internal pressure of the tubing, the annulus pressure, and the tubing-annulus leakage pressure difference; R k , E k and C k All transformations are performed using scaling parameters determined only by a subset of the training data; I k This represents the aperture inversion model input corresponding to the k-th time window.
8. The inversion method of the tubular leakage aperture inversion device based on multi-point acoustic response according to claim 7, characterized in that: In step S8, the median and median absolute deviation are calculated for the predicted window aperture values within the calibrated aperture range. Predicted window aperture values that deviate from the median by more than a preset multiple of the median absolute deviation are identified as abnormal predicted window aperture values. The indices of the remaining windows are then used to form a set of valid window indexes. V When the number of valid windows reaches a preset minimum, a set is used. V The median of the corresponding predicted window aperture values is used as the final leakage aperture. When the pressure condition to be measured exceeds the range of the calibrated pressure condition, the fixed geometric configuration to be measured is inconsistent with the saved fixed geometric configuration identifier, the state to be measured does not meet the stability criterion, the number of effective windows is less than the preset minimum number, or the overall root mean square amplitude, channel root mean square amplitude or pressure parameter exceeds the calibration applicable range, the extrapolated value of the model will not be used as a valid inversion result, and a prompt for re-acquisition or recalibration will be output.
Citation Information
Patent Citations
System and method for detecting multiple leaks in pipelines using hydraulic transients and machine learning
US20260185894A1
System and method for simulation of combustible ice exploitation leakage and for quantitative inversion of environmental parameters
WO2022001095A1
