Horizontal well multiphase flow profile inversion method based on multisource data fusion

Abstract

The invention relates to the technical field of oil and gas field development, and discloses a horizontal well multiphase flow profile inversion method based on multisource data fusion, which comprises the following steps: S1, laying optical fibers and acquiring data; s2, data preprocessing; s3, obtaining empirical parameters and a multi-source database; s4, judging a fluid scene and constructing a solver model; s5, iterative solution and phase fraction inversion are carried out, according to the empirical parameters and the multi-source database, the solver model constructed in the step S4 is adopted, solution is carried out in different phase number scenes, and an inversion index result is obtained; and S6, visual output and abnormal flow detection are carried out, and the inversion index result obtained in the step S5 is visually presented in the depth direction of the shaft and is used for assisting in judging abnormal phenomena. According to the method, the multiphase flow thermodynamic model is constructed based on the state equation, the water-liquid ratio and the gas-oil ratio are dynamically inverted through the iterative optimization algorithm in combination with the DAS sound velocity data, and high-precision characterization of multiphase flow characteristics is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of oil and gas field development, and in particular to a horizontal well multiphase flow profile inversion method based on multi-source data fusion. Background Art

[0002] During oil and gas well development, multiphase fluids such as water, oil, and gas often exist within the wellbore. Failure to accurately identify and monitor their distribution along the wellbore can lead to reduced well productivity or the risk of abnormal flow (such as water or gas channeling). Traditional monitoring methods, which mostly rely on downhole multiphase flowmeters or periodic logging, suffer from the following shortcomings:

[0003] 1. The measurement position is discrete, and only local area data can be obtained, making it difficult to achieve continuous monitoring of the entire well section;

[0004] 2. Well logging requires interrupting production, affecting economic benefits and lacking timeliness;

[0005] 3. Insufficient adaptability and inversion accuracy to complex wellbore conditions (high gas-oil ratio, high water-liquid ratio, etc.).

[0006] In recent years, Distributed Acoustic Sensing (DAS), with its advantages of full-well coverage, high resolution, and real-time performance, has become an emerging tool for multiphase flow monitoring. However, many challenges remain in effectively integrating the acoustic velocity information collected by DAS with accurate fluid thermodynamic models and using optimization algorithms to perform real-time inversion of multiphase flow composition (such as the water-to-liquid ratio and the gas-to-oil ratio).

[0007] During the oil and gas production process in horizontal wells, the distribution of multiphase fluids (including water, oil, and gas) within the wellbore is crucial for production management and reservoir development. However, the limitations of traditional monitoring technologies make it difficult to monitor the composition of multiphase flow in real time and with high resolution. Traditional downhole multiphase flowmeters and pressure gauges can only provide fluid information at discrete points and cannot cover the continuous monitoring needs of the entire wellbore area. At the same time, although logging technology can provide gas-liquid interface tracking capabilities, it is expensive and requires production to be interrupted during logging, which directly affects economic benefits.

[0008] There is currently no better solution for working conditions such as high gas-oil ratio and complex wellbore. Summary of the Invention

[0009] In order to solve one or more of the above technical problems, the present invention combines DAS data and pressure-volume-temperature state equation data to provide a horizontal well multiphase flow inversion method based on multi-source data fusion. By real-time and accurate monitoring of the three-phase flow state of water, oil, and gas in the wellbore, and using the pressure-volume-temperature model to accurately describe the thermophysical properties of the fluid, the multiphase flow composition in the wellbore (including the water-liquid ratio and the gas-oil ratio) is efficiently inverted. The method of the present invention realizes dynamic monitoring of the entire wellbore section and has the advantages of high resolution, real-time performance, and adaptability to complex flow conditions. The method of the present invention provides reliable support for optimizing wellbore production management, improving reservoir development efficiency, and timely identification of abnormal flows (such as water channeling and gas leakage).

[0010] The present invention provides the following technical solutions:

[0011] A horizontal well multiphase flow profile inversion method based on multi-source data fusion includes the following steps:

[0012] S1: Fiber optic deployment and data acquisition

[0013] An optical fiber is laid along the outer casing of the horizontal well and connected to the surface DAS and DTS demodulation system to achieve real-time acquisition of acoustic waves throughout the wellbore and obtain raw DAS and DTS data.

[0014] S2: Data preprocessing

[0015] Through noise reduction, calibration, segmentation processing, feature extraction and quality assessment, the raw DAS and DTS data are subjected to bandpass filtering for noise reduction, trend removal, DC removal and sound wave velocity calculation, and a relatively smooth waterfall distribution that can represent the local area is output to obtain the pre-processed measured data;

[0016] S3: Obtaining empirical parameters and multi-source databases

[0017] Combining the measured data with empirical parameters in an existing pressure-volume-temperature database to construct a multi-source database, wherein the empirical parameters include uniform pressure-volume-temperature data, a variable water-oil ratio, and a variable gas-oil ratio; the multi-source database includes the pressure-volume-temperature database, measured bottom hole pressure and temperature, surface production data, and empirical curves;

[0018] S4: Fluid scene identification and solver model construction

[0019] After identifying the fluid scenario, a corresponding solver model is constructed according to the number of fluid phases in different scenarios. The solver model types include single-phase, two-phase and three-phase;

[0020] S5: Iterative solution and phase fractional inversion

[0021] Based on the empirical parameters and the multi-source database, the solver model constructed in step S4 is used to perform a solution in different phase number scenarios to obtain inversion index results, wherein the inversion index results include a real-time sound velocity profile, flow pattern identification results, and segmented production capacity / injection rate; for two-phase flow, the state equation model results output by the solver model include liquid holdup distribution and phase flow rate; for three-phase flow, the state equation model results output by the solver model include water-liquid ratio, gas-oil ratio and their distribution along the wellbore;

[0022] S6: Visualization output and abnormal flow detection

[0023] The inversion index result obtained in the above step S5 is visualized in the wellbore depth direction to assist in judging abnormal phenomena.

[0024] Preferably, step S1 of optical fiber deployment and data acquisition includes the following steps:

[0025] S11: Fiber optic deployment

[0026] Single-mode optical fiber is used to measure DAS signals, while multimode optical fiber is used to measure DTS signals. Both optical fibers are coupled outside the casing along the wellbore, covering the entire horizontal section of the wellbore and close to the pay zone. A linear arrangement is used to record the sound velocity signal within the wellbore using high-frequency sampling. The two optical fibers are connected to surface DTS and DAS demodulation equipment, respectively. The surface equipment sends laser pulses to the optical fibers and analyzes the scattering effects in the echo signals.

[0027] S12: Data Collection

[0028] The surface equipment sends laser pulses to the optical fiber and analyzes the scattering characteristics in the echo signal to generate real-time sound velocity distribution data of the wellbore fluid;

[0029] The sound velocity signal inside the wellbore is continuously monitored through optical fiber to capture the fluid sound wave propagation characteristics and wellbore dynamic behavior. The collected data includes sound velocity signals, environmental parameters and other key production data. The vibration signal recorded by the DAS demodulation equipment is stored in the form of a time-depth two-dimensional distribution, reflecting the flow state of the fluid in the wellbore. The pressure data in the wellbore is recorded by the pressure sensor, and the temperature is recorded by the DTS demodulator, providing environmental parameters for subsequent sound velocity inversion.

[0030] Preferably, step S2 specifically includes the following steps:

[0031] S21: Data Noise Reduction

[0032] In the preprocessing of DAS data, an adaptive filtering algorithm is used to separate the effective signal from the background noise in the time and space domain, and the filter parameters and structure are automatically adjusted according to the changes in the input signal to perform filtering. The filtering process includes the calculation of the error signal, the adjustment of the filter coefficients, and the update of the filter coefficients. The error signal calculation is to calculate the error signal between the target signal and the filter output signal, and the error is used to adjust the filter coefficients. The filter coefficient adjustment is to update the filter coefficients according to the error signal, with the goal of minimizing the energy of the error signal. The filter coefficient update uses the least mean square algorithm. At the same time, the moving window technology is used to further optimize the noise reduction effect and significantly improve the signal quality.

[0033] S22: Data normalization

[0034] Convert the original sound velocity signal into a standardized format and perform amplitude normalization on the original DAS signal to eliminate the response inconsistency between different fiber segments;

[0035] S23: Data segmentation, signal feature extraction and analysis

[0036] To capture the dynamic changes along the wellbore, the data is divided into multiple spatial segments for independent processing. For each spatial segment, the average sound velocity value is calculated, and key signal features are extracted from the DAS data. A two-dimensional Fourier transform is performed on the DAS data to convert the time-depth domain data into the frequency-wavenumber domain. Then, through frequency-wavenumber analysis, a positive and negative wavenumber peak diagram of the sound velocity distribution is generated to obtain detailed information on the changes in sound velocity over time and space. The measured well section wave velocity or vibration characteristic distribution is obtained through frequency-wavenumber analysis. Combined with the theoretical sound velocity calculated by the state equation, matching / minimization error iteration is performed. The DAS data is processed through a low-pass filter to obtain LF-DAS data, which can track key features in multiphase flow, including transient events and fluid velocity, thereby improving the dynamic characteristic description of the data.

[0037] Preferably, step S4 specifically includes the following steps:

[0038] S41: Flow direction definition

[0039] The flow direction is defined according to the production operation mode of the wellbore. The flow direction includes the production flow direction and the injection flow direction. The production flow direction refers to the process of fluid flowing from the bottom of the wellbore to the wellhead; the injection flow direction refers to the process of fluid being injected from the wellhead to the bottom of the wellbore.

[0040] S42: Fluid scene discrimination

[0041] Based on the collected sound velocity characteristics, pressure and temperature conditions, and flow direction, the flow scenario is identified. The flow scenarios include single-phase, two-phase, and three-phase scenarios. The single-phase flow scenario includes the flow of pure oil, pure water, or pure gas; the two-phase flow scenario includes oil-water or oil-gas flow; and the three-phase flow scenario includes the coexistence of oil, gas, and water.

[0042] S43: Solver model construction

[0043] The solver model is used to calculate the sound velocity distribution in different fluid scenarios. In single-phase flow scenarios, the solver model assumes uniform fluid properties and calculates the sound velocity distribution. In two-phase flow scenarios, the solver model combines the equation of state to describe the change in sound velocity at the oil-water or oil-gas interface. In three-phase flow scenarios, the solver model considers the interaction between the oil, gas, and water phases and uses dynamic thermodynamic models to calculate complex sound velocity distributions.

[0044] S44: Experience parameter input

[0045] The empirical parameters are combined with the sound velocity data and input into the solver model for calculation;

[0046] S45: Expected Output

[0047] The solver model generates multiple outputs, including water-to-liquid ratio distribution, gas-to-oil ratio distribution, and liquid holdup profile. The water-to-liquid ratio distribution reflects the ratio of water to oil phases through a dynamic curve along the wellbore depth, providing an important basis for identifying water breakthrough. The gas-to-oil ratio distribution reveals changes in the ratio of gas to liquid phases, helping to identify high gas-to-oil ratio sections and abnormal gas flow behavior. The liquid holdup profile provides data support for optimizing production processes by evaluating the spatial proportion of liquid at different wellbore depths.

[0048] Preferably, step S5 specifically includes the following steps:

[0049] S51: Modeling the equation of state

[0050] The mathematical expression of the state equation is:

[0051]

[0052] Where: P is pressure; T is temperature; V is molar volume; R is the gas constant under specific temperature and pressure conditions; a, b are parameters related to fluid properties, which are calculated using the critical temperature Tc and critical pressure Pc of the fluid as follows:

[0053] a=0.45724(R 2 Tc 2 / Pc)×α(Tr,ω) (2)

[0054] b=0.07780(RTc / Pc) (3)

[0055] Where α(Tr,ω) is a function of the fluid eccentricity factor ω(omega), Tr = T / Tc, Tr is the dimensionless temperature;

[0056] S52: Calculation of the speed of sound

[0057] The calculation formula of sound speed c is as follows:

[0058]

[0059] Where: c is the speed of sound; ρ is the density; P is the pressure; s is the entropy;

[0060] S53: Calculation of the sound velocity of mixed fluids

[0061] The calculation of the mixed sound speed in multiphase flow scenarios is based on the following formula:

[0062]

[0063] In the above formula: is the sound velocity of the mixed fluid; α1 is the volume fraction of component 1; is the sound velocity of component 1; α2 is the volume fraction of component 2; is the sound velocity of component 2; when it is a three-phase system of oil, gas and water, the gas phase is regarded as component 1 and the oil and water as component 2. After solving, the influence of component 1 is eliminated and the oil and water are separated, with the oil as component 1 and the water as component 2.

[0064] S54: Iterative calculation

[0065] The input data for the iterative calculation include DAS sound velocity data, pressure-volume-temperature experimental data, and initial guess values;

[0066] S55: Dynamic error calculation

[0067] In each iteration, the algorithm calculates the sound speed c measured by DAS. DAS The sound speed c predicted by the equation of state EoS The error between them is calculated and the error expression is:

[0068] E=c DAS -c EoS (6)

[0069] The error E is the deviation between the water-to-liquid ratio W and gas-to-oil ratio G in the current model parameters and the actual wellbore fluid composition;

[0070] S56: Parameter Optimization

[0071] Optimize W and G by gradient descent method, the goal is to minimize the error E, the optimization formula is:

[0072]

[0073] Among them, α and β are learning rates, which represent the adjustment amplitude of the parameters in each iteration, and n is the number of iterations. Through multiple iterations, the model gradually converges to the optimal parameters;

[0074] S57: Phase Fractional Tracking Algorithm

[0075] Through dynamic optimization and iterative convergence of the phase fraction tracking algorithm, DAS data and state equations are combined to achieve the distribution inversion of multiphase fluids;

[0076] S58: Calculate the theoretical speed of sound c using the pressure-volume-temperature database and the equation of state model EoS This process combines the state equation and Equation 5. Through the error evaluation and feedback mechanism, the difference between the DAS sound velocity data and the theoretical sound velocity is calculated. The estimated values ​​of the water-liquid ratio and the gas-oil ratio are adjusted through the gradient optimization algorithm to gradually reduce the error.

[0077] S59: Determine whether the current model meets the accuracy requirements through convergence check. If the error reaches the set threshold, output the final result, including the dynamic distribution of water-liquid ratio and gas-oil ratio; if the error exceeds the threshold, return to adjust the model parameters and iterate again;

[0078] S510: Model Results and Evaluation

[0079] When the algorithm converges, the final output results include water-to-liquid ratio, gas-to-oil ratio distribution, and sound velocity profile, reflecting the dynamic changes and distribution characteristics of the fluid in the wellbore.

[0080] Preferably, in step S53, the speed of sound data measured by the DAS is combined with the pressure-volume-temperature parameters to optimize the model parameters, specifically:

[0081] In single-phase flow scenarios, linear regression analysis of DAS sound velocity data and pressure-volume-temperature database parameters is used to calibrate key parameters in the state equation, including critical temperature and critical pressure;

[0082] In multiphase flow scenarios, the pressure-volume-temperature parameters are dynamically adjusted using an error minimization algorithm; the estimated value of the fluid component ratio is optimized by fitting the sound velocity calculated by the model with the actual measurement data.

[0083] Compared with the prior art, the present invention has the following beneficial effects:

[0084] (1) Combination of pressure-volume-temperature equation of state and dynamic optimization algorithm

[0085] The present invention constructs an accurate multiphase flow thermodynamic model based on the equation of state, combines DAS sound velocity data, and dynamically inverts the water-liquid ratio and gas-oil ratio through an iterative optimization algorithm to achieve high-precision characterization of multiphase flow characteristics.

[0086] (2) Adaptability to multiple scenarios and working conditions

[0087] The present invention develops a highly adaptable solver model for single-phase flow, two-phase flow and three-phase flow, which can accurately describe the thermal physical properties of fluids under complex working conditions such as high gas-oil ratio and water breakthrough. It has strong adaptability and a wide range of applications.

[0088] (3) Real-time identification of abnormal flow behavior

[0089] Through dynamic analysis of the sound velocity distribution, the present invention can identify abnormal behaviors such as water breakthrough and gas leakage in the wellbore in real time, and combine with the early warning mechanism to support production safety and ensure the stability of wellbore operation.

[0090] (4) Real-time dynamic monitoring and high-resolution coverage

[0091] The present invention uses DAS technology to obtain high-resolution sound velocity distribution data accurate to 1 meter. Combined with state equation modeling, it can track the gas-liquid interface and multiphase flow characteristics in real time, realizing continuous dynamic monitoring of the entire wellbore section, and making up for the deficiency of traditional methods that can only provide discrete point information.

[0092] (5) Optimize production parameters and improve economic benefits

[0093] This method accurately inverts the water-to-liquid ratio, gas-to-oil ratio, and liquid holdup, providing a scientific basis for optimizing production parameters. For example, by dynamically adjusting pump speed, water injection rate, or gas injection rate, production efficiency can be significantly improved. Furthermore, it avoids the production losses caused by traditional logging techniques requiring production interruptions, thereby improving economic benefits.

[0094] (6) Early identification of abnormal flow behavior

[0095] This technology can accurately identify abnormal flow behavior within the wellbore, such as water breakthrough, gas leakage, and dynamic changes in areas with high gas-oil ratios. Through real-time monitoring and analysis, these anomalies can be effectively identified at an early stage, reducing the risk of safety incidents (such as blowouts) and ensuring smooth production operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0096] Figure 1 The present invention provides a flowchart of a method for inverting multiphase flow profiles in horizontal wells based on multi-source data fusion.

[0097] Figure 2 Schematic diagram of the S4 fluid scene discrimination and solver model construction provided in an embodiment of the present invention.

[0098] Figure 3 This is a result diagram provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0099] The present invention provides a method for real-time monitoring and inversion of multiphase flow (oil, gas, water, etc.) within a horizontal wellbore by combining multi-source data, including acoustic velocity data extracted from distributed acoustic sensing (DAS) and pressure-volume-temperature equation of state data. By collecting acoustic velocity data within the wellbore in real time, combined with fluid thermodynamic models, dynamic optimization algorithms, and empirical parameters, the distribution ratio of water, oil, and gas within the wellbore is accurately inverted, and a production profile is generated, providing a scientific basis for dynamic monitoring and production optimization of horizontal wells. This method can assist in oilfield production management and early identification of abnormal wellbore flow, and has the characteristics of high resolution, real-time performance, and high precision.

[0100] The technical route of the present invention is as follows Figure 1 As shown in the figure, it runs through the complete implementation path from flow direction definition to final result output. By clarifying the relationship between flow direction, scenario definition, solver model, empirical parameter input and expected output, a clear technical implementation framework is formed.

[0101] The present invention is described in detail below with reference to the embodiments and accompanying drawings. However, it should be understood that the embodiments and accompanying drawings are merely exemplary descriptions of the present invention and do not constitute any limitation on the scope of protection of the present invention. All reasonable variations and combinations within the scope of the inventive concept of the present invention fall within the scope of protection of the present invention.

[0102] Example 1

[0103] according to Figure 1 As shown in the flow chart, this embodiment fully utilizes the characteristic of DAS that can obtain the acoustic velocity distribution of the entire wellbore, and combines the state equation that has a good description of the thermodynamic properties of the fluid to perform dynamic inversion of the multiphase flow characteristic quantities. The inversion steps are as follows:

[0104] S1: Fiber optic deployment and data acquisition

[0105] Optical fiber is laid along the outer tubing of the horizontal well casing and connected to the ground DAS demodulation system to achieve real-time acquisition of acoustic waves throughout the wellbore, obtaining raw DAS and DTS data, which facilitates quality control of the collected data in the next step.

[0106] S2: Data preprocessing

[0107] The original DAS and DTS data are subjected to bandpass filtering, de-noising, de-trending, and DC removal operations to complete the pre-processing steps, and the output is a relatively smooth waterfall chart distribution that can represent the local area;

[0108] S3: Obtaining empirical parameters and multi-source databases

[0109] When embedding the pressure-volume-temperature state equation in the solver, it is necessary to build a complete multi-source database to support it, including uniform pressure, volume and temperature mode; variable water-oil ratio and variable gas-oil ratio mode; bottom hole pressure, temperature and flow measurement points.

[0110] Empirical parameters include uniform pressure-volume-temperature data or variable water-oil ratio, variable gas-oil ratio, etc.

[0111] The multi-source database reflects the fusion of multi-source data, including the pressure-volume-temperature database, bottom hole pressure and temperature measurements, surface production data, experience curves, etc.

[0112] S4: Fluid scene identification and solver model construction

[0113] Build corresponding solvers for the number of fluid phases in different scenarios, including single-phase / two-phase / three-phase / four-phase.

[0114] In single-phase flow scenarios, the model is relatively simple; in multiphase flow scenarios, it is necessary to consider the interaction between oil, gas, and water phases and the mixed sound velocity formula.

[0115] S5: Iterative solution and phase fraction (component ratio) inversion

[0116] Based on empirical parameters and a multi-source database, the solver in step S4 is used to solve the problem. Under different phase number scenarios, the core results of the state equation model output by the solver are slightly different. For example, for two-phase flow: liquid holdup distribution, phase flow rate, etc.

[0117] For three-phase flow: water-liquid ratio, gas-oil ratio and their distribution along the wellbore;

[0118] For four-phase flow: add solid phase on the basis of three-phase flow;

[0119] For each phase flow, these metrics can be further output, including real-time acoustic velocity profiles, flow pattern identification, and segmented production capacity / injection rate, yielding multiple inversion metrics. Core metrics such as water-to-liquid ratio distribution, gas-to-oil ratio distribution, and liquid holdup provide a basis for wellbore multiphase flow monitoring and production optimization.

[0120] S6: Visualization output and abnormal flow detection

[0121] The inversion index results obtained in the above step S5 are visualized in the wellbore depth direction to assist in judging certain abnormal phenomena (such as water breakthrough and gas leakage).

[0122] Combined with historical or real-time data, it provides a basis for wellbore production management, water (gas) injection strategy adjustment, threshold alarm, etc.

[0123] More specifically, step S1 of optical fiber deployment and data collection includes the following steps:

[0124] S11: Fiber optic deployment

[0125] Fiber optic layout is a prerequisite for DTS and DAS technologies, and its design directly impacts the quality of data acquisition and the long-term stability of the system. In this embodiment, single-mode fiber is used to measure DAS signals, and multimode fiber is used to measure DTS signals to improve signal coherence, ensuring a longer measurement distance and higher resolution. The optical fiber is coupled to the outside of the tubing along the wellbore, covering the entire horizontal section of the wellbore and maximizing proximity to the producing layer. A linear layout ensures that the optical fiber is evenly distributed along the wellbore wall, allowing high-frequency sampling to record the sound velocity signal within the wellbore. The optical fiber is connected to the surface DTS and DAS demodulation equipment via a highly sealed connector. The surface equipment is responsible for transmitting laser pulses to the optical fiber and analyzing the scattering characteristics of the echo signal to generate real-time sound velocity distribution data for the wellbore fluid. The exterior of the optical fiber is coated with a high-temperature, high-pressure, and corrosion-resistant coating to withstand the extreme environmental conditions within the wellbore, ensuring the stability and reliability of the system during long-term operation.

[0126] S12: Data Collection

[0127] Data acquisition continuously monitors the sound velocity signal inside the wellbore through optical fiber, focusing on capturing the fluid acoustic wave propagation characteristics and the dynamic behavior of the wellbore. The main contents of the acquisition include sound velocity signals, environmental parameters and other key production data. The vibration signals recorded by the DAS system are stored in the form of a two-dimensional time-depth distribution, which fully reflects the flow state of the fluid in the wellbore. The pressure in the wellbore is recorded by a special sensor, and the temperature is recorded by the DTS demodulator, providing the necessary environmental parameters for subsequent sound velocity inversion. In response to the monitoring needs of different scenarios, the system's laser pulse width, frequency and time-space sampling resolution can be dynamically adjusted. For example, in complex multiphase flow scenarios, the sampling frequency needs to be increased to capture detailed changes.

[0128] S2: Data preprocessing

[0129] Data quality is controlled through data preprocessing. The reliability of DTS and DAS data directly impacts the accuracy of subsequent inversion calculations. This embodiment improves signal quality through rational hardware design and data processing strategies. The system monitors signal fluctuations during acquisition in real time, automatically rejects abnormal data, selects an appropriate scale length, and uses multi-point averaging to reduce measurement errors.

[0130] Data preprocessing in step S2 is essential for converting raw DAS and DTS signals into high-quality input data. This embodiment systematically improves signal quality and reliability through processes such as noise reduction, calibration, segmented processing, feature extraction, and quality assessment, providing a solid data foundation for subsequent multiphase flow inversion and fractional phase tracking algorithms. Its specific implementation includes the following steps:

[0131] S21: Data Noise Reduction

[0132] Noise reduction is a critical step in DAS data preprocessing. Due to the complex multiphase flow within the wellbore and the influence of the external environment, the collected signals are often interspersed with interference signals such as mechanical vibration, fluid leakage, and random environmental noise. To improve the signal-to-noise ratio, adaptive filtering algorithms are used. These filters can separate the effective signal from background noise in the spatiotemporal domain and automatically adjust the filter parameters and structure based on changes in the input signal. This technique automatically adjusts the filter coefficients based on the statistical characteristics of the input and output signals using a least mean square algorithm to achieve optimal filtering characteristics. This technique can be used to suppress non-stationary noise in fiber optic data sets.

[0133] The implementation steps are as follows:

[0134] 1. Error signal calculation: Calculate the error between the target signal and the filter output signal. The error signal is used to adjust the filter coefficients.

[0135] 2. Filter coefficient adjustment: The filter coefficients are updated according to the error signal, with the goal of minimizing the energy of the error signal.

[0136] 3. Filter coefficient update: The coefficient update is based on the least mean square algorithm.

[0137] At the same time, moving window technology is used to further optimize the noise reduction effect, thereby significantly improving signal quality.

[0138] S22: Data normalization

[0139] Signal calibration is a crucial step in data preprocessing, aiming to eliminate systematic errors and nonlinear biases and convert the raw acoustic velocity signal into a standardized format. Amplitude normalization is performed on the raw DAS signal to eliminate response inconsistencies between different fiber segments.

[0140] S23: Data segmentation, signal feature extraction and analysis

[0141] To capture dynamic changes along the wellbore, the data is divided into multiple spatial segments and processed independently. For each segment, the average acoustic velocity is calculated and its temporal and spatial variations are analyzed. Furthermore, characteristic parameters of key events within the segment, such as interface transitions and changes in multiphase flow patterns, are extracted to provide important input for subsequent model calculations.

[0142] Extract key signal features from DAS data. A two-dimensional Fourier transform is performed on the DAS data, converting the time-depth domain data to the frequency-wavenumber domain. Frequency-wavenumber analysis is then used to generate a peak-to-peak diagram of the sound velocity distribution, detailing how the sound velocity varies over time and space. Frequency-wavenumber analysis is used to determine the measured wave velocity or vibration signature distribution in the well section. This is combined with the theoretical sound velocity calculated using the equation of state to perform matching / error minimization iterations. Low-pass filtering is used to process the DAS data to obtain LF-DAS data, which can track key features in multiphase flow, including transient events and fluid velocity, further enhancing the dynamic characterization of the data.

[0143] The specific steps of obtaining empirical parameters and constructing a multi-source database in step 3 are to combine the measured data (DAS + bottom hole sensor + surface production, etc.) with the empirical parameters in the pressure-volume-temperature database (critical pressure, critical temperature, eccentricity factor, etc.) to obtain the constructed multi-source database.

[0144] Empirical parameters include uniform pressure-volume-temperature data and variable water-oil ratio, variable gas-oil ratio, etc.

[0145] The multi-source database reflects the fusion of multi-source data, including the pressure-volume-temperature database, bottom hole pressure and temperature measurements, surface production data, experience curves, etc.

[0146] S4: Fluid scene identification and solver model construction

[0147] First, the single-phase, two-phase, and three-phase flow scenarios are identified based on the collected sound velocity characteristics, pressure and temperature conditions, production / injection direction, and other information. Based on the previous production data, the production or injection scenario is obtained, and the flow scenario is determined to be single-phase, two-phase, or three-phase. Applicable composite thermodynamic state equations (or "solver models") are established for different scenarios. For example, the phase distribution of oil and gas, the mixed density of oil, water, and gas, and the sound velocity formula are distinguished. Figure 2 Specifically, the method includes the following steps:

[0148] S41: Flow direction definition

[0149] The definition of flow direction is the starting point of the technical logic process, which directly affects the subsequent scenario classification and the construction of the solver model. According to the production operation mode of the wellbore, the flow direction is mainly divided into production flow direction and injection flow direction. The production flow direction refers to the process of fluid flowing from the bottom of the wellbore to the wellhead. It is often used in oil and gas production operations. It mainly focuses on the distribution of multiphase fluids in the wellbore and the impact of dynamic changes on production. The injection flow direction refers to the process of fluid injection from the wellhead to the bottom of the wellbore. It is usually used for water injection or gas injection operations to evaluate the diffusion behavior of the injected fluid and the wellbore response. The definition of flow direction provides a directional basis for subsequent scenario classification and data analysis, and at the same time determines the acquisition range and analysis boundary of the DAS sound velocity signal.

[0150] S42: Fluid scene discrimination

[0151] Fluid scenarios are classified into three categories based on the complexity of the fluid composition: single-phase flow, two-phase flow, and three-phase flow, each applicable to different production and injection scenarios. Single-phase flow scenarios include the flow of pure oil, pure water, or pure gas, where the sound velocity distribution is uniform and does not vary significantly, and usually occurs in the early stages of production or water injection operations. In two-phase flow scenarios, such as oil-water or oil-gas flow, the sound velocity shows a significant jump at the phase interface, reflecting the physical properties at the multiphase interface. In three-phase flow scenarios, oil, gas, and water coexist, the flow characteristics are complex, and the sound velocity dynamics are affected by multiple factors, often occurring in sections with high gas-oil ratios or high water-liquid ratios.

[0152] S43: Solver model construction

[0153] The solver model calculates the sound velocity distribution for various flow scenarios. In single-phase flow, the model assumes uniform fluid properties, making the sound velocity distribution calculation relatively simple. In two-phase flow, the model combines the equation of state with Equation 5, described later, to describe the sound velocity variations at the oil-water or oil-gas interface. In three-phase flow, the model considers the interactions between the oil, gas, and water phases, utilizing a dynamic thermodynamic model to calculate the complex sound velocity distribution.

[0154] S44: Experience parameter input

[0155] Empirical parameter input primarily comes from experimental data in the pressure-volume-temperature database. This data, including thermodynamic parameters such as density, pressure, and temperature, is used to describe the characteristics of the fluid at different depths and conditions. Furthermore, depending on the specific fluid scenario, the pressure-volume-temperature parameters can be selected to have a uniform or variable distribution pattern. In single-phase flow scenarios, it is assumed that the fluid characteristics remain uniform in the wellbore, and a uniform distribution pattern is applicable. In multiphase flow scenarios, however, it is necessary to consider the dynamic changes of parameters with depth or time, and a variable distribution pattern is adopted. These input parameters, combined with DAS sound velocity data, provide reliable boundary conditions for state equation model calculations.

[0156] S45: Expected Output

[0157] Through model calculations, this embodiment generates multiple key outputs, including water-to-liquid ratio distribution, gas-to-oil ratio distribution, and liquid holdup profile. The water-to-liquid ratio distribution, through a dynamic curve along the wellbore depth, reflects the ratio of water to oil, providing an important basis for identifying water breakthrough. The gas-to-oil ratio distribution reveals the changes in the ratio of gas to liquid, helping to identify high gas-to-oil ratio sections and abnormal gas flow behavior. The liquid holdup profile, by assessing the spatial proportion of liquid at different wellbore depths, provides essential data support for optimizing production processes.

[0158] The above technical process completes a complete technical chain from data acquisition, scenario definition, model solution, to result output. The above technical roadmap clearly demonstrates the logical relationship between each link and its role in the overall technical system, providing a scientific and comprehensive implementation framework for accurate inversion of wellbore multiphase flow.

[0159] Step S5, iterative solution and phase fraction (component ratio) inversion, is used to construct an equation of state model. This equation of state is used to describe the thermodynamic properties of the fluid, primarily for calculating the velocity of sound and other physical parameters of multiphase flow in the wellbore. This embodiment achieves dynamic multiphase flow modeling by combining the equation of state, a pressure-volume-temperature database, and real-time DAS data. Specifically, the following steps are included:

[0160] S51: Modeling of Equations of State

[0161] The state equation used in this embodiment is suitable for thermodynamic modeling of complex oil and gas mixtures, and its mathematical expression is:

[0162]

[0163] Where: P - pressure; T - temperature; V - molar volume; R - gas constant under specific temperature and pressure conditions; a, b - parameters related to fluid properties, calculated from the critical temperature Tc and critical pressure Pc of the fluid.

[0164] a=0.45724(R 2 Tc 2 / Pc)×α(Tr,ω) (2)

[0165] b=0.07780(RTc / Pc) (3)

[0166] Among them, α(Tr,ω) is a function that takes into account the fluid eccentricity factor ω(omega), Tr = T / Tc, Tr is the dimensionless temperature;

[0167] The advantage of the state equation is that it can better simulate the behavior of oil and gas fluids under high pressure and high temperature conditions.

[0168] S52: Calculation of the speed of sound

[0169] The calculation of the sound velocity c is based on the isentropic condition of the state equation, which describes the relationship between the fluid sound velocity, density and pressure. The formula is as follows:

[0170]

[0171] Where: c-speed of sound; ρ-density; P-pressure; s-entropy.

[0172] In multiphase fluid scenarios, the equation of state combines pressure, temperature, and component ratios to calculate the sound velocity of each component and ultimately the mixed sound velocity distribution. This means that, assuming entropy (s) remains constant, the partial derivative of pressure with respect to density determines the sound velocity of small disturbances in the medium.

[0173] S53: Calculation of the sound velocity of mixed fluids

[0174] The calculation of mixed sound velocity in multiphase flow scenarios is based on Formula 5, which comprehensively considers the acoustic characteristics of different fluid components. The formula is as follows:

[0175]

[0176] In formula 5: - the sound velocity of the mixed fluid; α1- the volume fraction of component 1; - the sound velocity of component 1; α2- the volume fraction of component 2; -Sound velocity of component 2.

[0177] When the solution is oil, gas and water, the gas phase is regarded as component 1 and the oil and water are regarded as component 2. After solving, the influence of component 1 is eliminated and the oil and water are separated, with the oil as component 1 and the water as component 2.

[0178] The choice of this model fully considers the interaction between gas, oil and water phases, and can accurately describe the changes in sound velocity under different flow modes.

[0179] Step S5 involves data input and optimization of the pressure-volume-temperature database. The pressure-volume-temperature database is a crucial data source for equation-of-state (EOS) modeling. It contains experimentally calibrated fluid thermodynamic parameters, such as density, saturation pressure, and gas-to-oil ratio. This data provides reliable boundary conditions for EOS calculations and supports the construction of a dynamic velocity of sound model. To improve the applicability of the EOS, the present invention combines velocity of sound data measured by the DAS with the pressure-volume-temperature parameters to optimize the model parameters.

[0180] In a single-phase flow scenario, this example uses linear regression analysis of DAS sound velocity data and pressure-volume-temperature database parameters to calibrate key parameters in the equation of state, including critical temperature and critical pressure. This step ensures that the equation of state model accurately reflects the thermodynamic behavior of the single-phase fluid.

[0181] In multiphase flow scenarios, an error minimization algorithm is used to dynamically adjust the pressure-volume-temperature parameters. By fitting the model-calculated sound velocity with actual measured data, the estimated fluid component ratios are optimized. This approach can adapt to dynamic changes in complex flow conditions and provide higher accuracy for sound velocity calculations in multiphase flow scenarios.

[0182] S54: Iterative calculation

[0183] The input data mainly includes DAS sound velocity data, pressure-volume-temperature experimental data, and initial guess values. The sound velocity data calculated by DAS is a dynamic signal, and the real-time sound velocity distribution obtained by the DAS system is stored in the form of a depth-time two-dimensional matrix. These data directly reflect the local dynamic characteristics of the fluid in the wellbore and provide an important physical basis for phase fraction inversion. The pressure-volume-temperature experimental data contains pressure, temperature, fluid density, and volume parameters of each component. These parameters are obtained through experimental calibration, ensuring the accuracy and reliability of the state equation model calculation. In addition, the initial conditions of the algorithm include initial estimates of the water-to-liquid ratio and the gas-to-oil ratio. These parameters serve as the starting point for iterative optimization.

[0184] S55: Dynamic error calculation

[0185] In each iteration, the algorithm calculates the sound speed (c DAS ) and the sound speed predicted by the equation of state (c EoS ) is calculated. The error expression is:

[0186] E=c DAS -c EoS (6)

[0187] The error E reflects the deviation between the water-to-liquid ratio W and gas-to-oil ratio G in the current model parameters and the actual wellbore fluid composition, and is the core basis for subsequent optimization.

[0188] S56: Parameter Optimization

[0189] Optimize W and G by gradient descent method, the goal is to minimize the error E. The optimization formula is:

[0190]

[0191] Where α and β are learning rates, indicating the adjustment of parameters in each iteration, and n is the number of iterations. Through multiple iterations, the model gradually converges to the optimal parameters.

[0192] S57: Phase Fractional Tracking Algorithm

[0193] The phase fraction tracking algorithm combines DAS data with the equation of state through dynamic optimization and iterative convergence to achieve multiphase fluid distribution inversion. This process clearly demonstrates the complete path from data input to final output, covering all links. First, by combining the wellbore structure and fluid properties, flow direction and possible multiphase scenarios are defined. For example, typical flow patterns such as high gas-oil ratio scenarios, water breakthrough, or uniformly distributed two-phase flow can be identified, providing clear physical boundary conditions for model calculations.

[0194] S58: Calculate the theoretical speed of sound c using the pressure-volume-temperature database and the equation of state model EoS This process combines the equation of state and Equation 5 to accurately describe the mixed sound velocity characteristics of multiphase fluids, providing the core computational basis for inversion. After completing the theoretical sound velocity calculation, an error evaluation and feedback mechanism is used to calculate the difference between the DAS sound velocity data and the theoretical sound velocity. A gradient optimization algorithm is then used to adjust the estimated water-to-liquid ratio and gas-to-oil ratio, gradually reducing the error.

[0195] S59: Determine whether the current model meets the accuracy requirements through convergence check. If the error reaches the set threshold, the final result is output, including the dynamic distribution of the water-liquid ratio and the gas-oil ratio; if the error exceeds the threshold, it returns to adjust the model parameters and iterates again. The "stable solution" module in the optimization process is a key node, which is used to determine whether the current solution meets the expected accuracy, thereby determining whether the algorithm continues to iterate. In addition, when it is found that the water-liquid ratio and gas-oil ratio distribution have abnormal jumps or critical values ​​in space, some kind of alarm / judgment will be triggered, which may be gas or water channeling. Based on this complete process, this algorithm can accurately invert the dynamic characteristics of wellbore multiphase flow and provide reliable data support for production optimization.

[0196] S510: Model Results and Evaluation

[0197] When the algorithm converges, the final output includes the water-to-liquid ratio, gas-to-oil ratio distribution, and acoustic velocity profile. These results intuitively reflect the dynamic changes and distribution characteristics of the fluid in the wellbore.

[0198] The output of the state equation model in step S5 is based on the construction and optimization method of the above steps. The state equation model can provide key characteristic distribution and dynamic change information of the multiphase fluid.

[0199] These outputs include:

[0200] 1) The sound velocity distribution is a direct output of the equation of state, reflecting the variation of sound velocity along the depth of the wellbore. The sound velocity distribution curve provides an important basis for analyzing the flow characteristics of the fluid in the wellbore.

[0201] 2) The water-to-liquid ratio and gas-to-oil ratio are key parameters describing the composition of multiphase fluids. Equation of state models can dynamically predict the distribution of these parameters at different wellbore depths, helping to identify fluid flow anomalies.

[0202] 3) Liquid holdup is a key indicator in fluid profile analysis. The liquid holdup curve generated by the model reflects the proportion of liquid in the wellbore, providing data support for optimizing production processes.

[0203] These output results not only provide accurate basic data for multiphase flow inversion, but also directly support the decision-making process of production optimization, greatly improving the analysis ability of wellbore fluid flow behavior.

[0204] The iteration and phase fraction tracking in step S5 adopt dynamic optimization and phase fraction tracking algorithms.

[0205] Step 6: Visualization of output and abnormal flow detection The output results in the previous steps are visualized.

[0206] This embodiment uses the data collected in a certain oil well and applies the horizontal well multiphase flow profile inversion method based on multi-source data fusion provided by this embodiment to form the following Figure 3 The interpretation results are shown:

[0207] The first column shows the fracture stages, the second column shows the perforation clusters, the third column shows the cumulative water and oil production distribution curves for all producing intervals (all fracture stages or all perforation clusters), the fourth column shows the water and oil production distribution histograms for each fracture stage, and the fifth column shows the water and oil production distribution histograms for each perforation cluster. Blue represents water, and green represents oil.

[0208] The above embodiments are merely preferred embodiments of the present invention, and the scope of protection of the present invention is not limited to the above embodiments. All technical solutions that fall within the scope of protection of the present invention are within the scope of protection of the present invention. It should be noted that improvements and modifications that can be made by a person skilled in the art without departing from the principles of the present invention are also considered to be within the scope of protection of the present invention.

Claims

1. A method for inverting multiphase flow profiles in horizontal wells based on multi-source data fusion, characterized by: The steps include: S1: Fiber optic deployment and data acquisition An optical fiber is laid along the outer casing of the horizontal well and connected to the surface DAS and DTS demodulation system to achieve real-time acquisition of acoustic waves throughout the wellbore and obtain raw DAS and DTS data. S2: Data preprocessing Through noise reduction, calibration, segmentation processing, feature extraction and quality assessment, the raw DAS and DTS data are subjected to bandpass filtering for noise reduction, trend removal, DC removal and sound wave velocity calculation, and a relatively smooth waterfall distribution that can represent the local area is output to obtain the pre-processed measured data; S3: Obtaining empirical parameters and multi-source databases Combining the measured data with empirical parameters in an existing pressure-volume-temperature database to construct a multi-source database, wherein the empirical parameters include uniform pressure-volume-temperature data, a variable water-oil ratio, and a variable gas-oil ratio; the multi-source database includes the pressure-volume-temperature database, measured bottom hole pressure and temperature, surface production data, and empirical curves; S4: Fluid scene identification and solver model construction After identifying the fluid scenario, a corresponding solver model is constructed according to the number of fluid phases in different scenarios. The solver model types include single-phase, two-phase and three-phase; S5: Iterative solution and phase fractional inversion Based on the empirical parameters and the multi-source database, the solver model constructed in step S4 is used to perform a solution in different phase number scenarios to obtain inversion index results, wherein the inversion index results include a real-time sound velocity profile, flow pattern identification results, and segmented production capacity / injection rate; for two-phase flow, the state equation model results output by the solver model include liquid holdup distribution and phase flow rate; for three-phase flow, the state equation model results output by the solver model include water-liquid ratio, gas-oil ratio and their distribution along the wellbore; S6: Visualization output and abnormal flow detection The inversion index result obtained in the above step S5 is visualized in the wellbore depth direction to assist in judging abnormal phenomena.

2. The horizontal well multiphase flow profile inversion method based on multi-source data fusion according to claim 1 is characterized by: Step S1: Fiber optic deployment and data collection, including the following steps: S11: Fiber optic deployment Single-mode optical fiber is used to measure DAS signals, while multimode optical fiber is used to measure DTS signals. Both optical fibers are coupled outside the casing along the wellbore, covering the entire horizontal section of the wellbore and close to the pay zone. A linear arrangement is used to record the sound velocity signal within the wellbore using high-frequency sampling. The two optical fibers are connected to the DTS and DAS demodulation equipment on the surface, respectively. S12: Data Collection The surface equipment sends laser pulses to the optical fiber and analyzes the scattering characteristics in the echo signal to generate real-time sound velocity distribution data of the wellbore fluid; The sound velocity signal inside the wellbore is continuously monitored through optical fiber to capture the fluid sound wave propagation characteristics and wellbore dynamic behavior. The collected data includes sound velocity signals, environmental parameters and other key production data. The vibration signal recorded by the DAS demodulation equipment is stored in the form of a time-depth two-dimensional distribution, reflecting the flow state of the fluid in the wellbore. The pressure data in the wellbore is recorded by the pressure sensor, and the temperature is recorded by the DTS demodulator, providing environmental parameters for subsequent sound velocity inversion.

3. The horizontal well multiphase flow profile inversion method based on multi-source data fusion according to claim 2 is characterized by: Step S2 specifically includes the following steps: S21: Data Noise Reduction In the preprocessing of DAS data, an adaptive filtering algorithm is used to separate the effective signal from the background noise in the time and space domain, and the filter parameters and structure are automatically adjusted according to the changes in the input signal to perform filtering. The filtering process includes the calculation of the error signal, the adjustment of the filter coefficients, and the update of the filter coefficients. The error signal calculation is to calculate the error signal between the target signal and the filter output signal, and the error is used to adjust the filter coefficients. The filter coefficient adjustment is to update the filter coefficients according to the error signal, with the goal of minimizing the energy of the error signal. The filter coefficient update uses the least mean square algorithm. At the same time, the moving window technology is used to further optimize the noise reduction effect and significantly improve the signal quality. S22: Data normalization Convert the original sound velocity signal into a standardized format and perform amplitude normalization on the original DAS signal to eliminate the response inconsistency between different fiber segments; S23: Data segmentation, signal feature extraction and analysis To capture the dynamic changes along the wellbore, the data is divided into multiple spatial segments for independent processing. For each spatial segment, the average sound velocity value is calculated, and key signal features are extracted from the DAS data. A two-dimensional Fourier transform is performed on the DAS data to convert the time-depth domain data into the frequency-wavenumber domain. Then, through frequency-wavenumber analysis, a positive and negative wavenumber peak diagram of the sound velocity distribution is generated to obtain detailed information on the changes in sound velocity over time and space. The measured well section wave velocity or vibration characteristic distribution is obtained through frequency-wavenumber analysis. Combined with the theoretical sound velocity calculated by the state equation, matching / minimization error iteration is performed. The DAS data is processed through a low-pass filter to obtain LF-DAS data, which can track key features in multiphase flow, including transient events and fluid velocity, thereby improving the dynamic characteristic description of the data.

4. The horizontal well multiphase flow profile inversion method based on multi-source data fusion according to claim 3 is characterized by: Step S4 specifically includes the following steps: S41: Flow direction definition The flow direction is defined according to the production operation mode of the wellbore. The flow direction includes the production flow direction and the injection flow direction. The production flow direction refers to the process of fluid flowing from the bottom of the wellbore to the wellhead; the injection flow direction refers to the process of fluid being injected from the wellhead to the bottom of the wellbore. S42: Fluid scene discrimination Based on the collected sound velocity characteristics, pressure and temperature conditions, and flow direction, the flow scenario is identified. The flow scenarios include single-phase, two-phase, and three-phase scenarios. The single-phase flow scenario includes the flow of pure oil, pure water, or pure gas; the two-phase flow scenario includes oil-water or oil-gas flow; and the three-phase flow scenario includes the coexistence of oil, gas, and water. S43: Solver model construction The solver model is used to calculate the sound velocity distribution in different fluid scenarios. In single-phase flow scenarios, the solver model assumes uniform fluid properties and calculates the sound velocity distribution. In two-phase flow scenarios, the solver model combines the equation of state to describe the change in sound velocity at the oil-water or oil-gas interface. In three-phase flow scenarios, the solver model considers the interaction between the oil, gas, and water phases and uses dynamic thermodynamic models to calculate complex sound velocity distributions. S44: Experience parameter input The empirical parameters are combined with the sound velocity data and input into the solver model for calculation; S45: Expected Output The solver model generates multiple outputs, including water-to-liquid ratio distribution, gas-to-oil ratio distribution, and liquid holdup profile. The water-to-liquid ratio distribution reflects the ratio of water to oil phases through a dynamic curve along the wellbore depth, providing an important basis for identifying water breakthrough. The gas-to-oil ratio distribution reveals changes in the ratio of gas to liquid phases, helping to identify high gas-to-oil ratio sections and abnormal gas flow behavior. The liquid holdup profile provides data support for optimizing production processes by evaluating the spatial proportion of liquid at different wellbore depths.

5. The horizontal well multiphase flow profile inversion method based on multi-source data fusion according to claim 4 is characterized by: Step S5 specifically includes the following steps: S51: Modeling the equation of state The mathematical expression of the state equation is: Where: P is pressure; T is temperature; V is molar volume; R is the gas constant under specific temperature and pressure conditions; a and b are parameters related to fluid properties, which are calculated using the following formula: a=0.45724 (R 2 Tc 2 / Pc) × α(Tr, ω) (2) b = 0.07780 (RTc / Pc) (3) Where α(Tr,ω) is a function of the fluid eccentricity factor ω(omega), Tr = T / Tc, Tr is the dimensionless temperature; S52: Calculation of the speed of sound The calculation formula of sound speed c is as follows: Where: c is the speed of sound; ρ is the density; P is the pressure; s is the entropy; S53: Calculation of the sound velocity of mixed fluids The calculation of the mixed sound speed in multiphase flow scenarios is based on the following formula: In the above formula: is the sound velocity of the mixed fluid; α1 is the volume fraction of component 1; is the sound velocity of component 1; α2 is the volume fraction of component 2; is the sound velocity of component 2; when it is a three-phase system of oil, gas and water, the gas phase is regarded as component 1 and the oil and water as component 2. After solving, the influence of component 1 is eliminated and the oil and water are separated, with the oil as component 1 and the water as component 2. S54: Iterative calculation The input data for the iterative calculation include DAS sound velocity data, pressure-volume-temperature experimental data, and initial guess values; S55: Dynamic error calculation In each iteration, the algorithm calculates the sound speed c measured by DAS. DAS The sound speed c predicted by the equation of state EoS The error between them is calculated and the error expression is: E=c DAS -c EoS (6) The error E is the deviation between the water-to-liquid ratio W and gas-to-oil ratio G in the current model parameters and the actual wellbore fluid composition; S56: Parameter Optimization Optimize W and G by gradient descent method, the goal is to minimize the error E, the optimization formula is: Among them, α and β are learning rates, which represent the adjustment amplitude of the parameters in each iteration, and n is the number of iterations. Through multiple iterations, the model gradually converges to the optimal parameters; S57: Phase Fractional Tracking Algorithm Through dynamic optimization and iterative convergence of the phase fraction tracking algorithm, DAS data and state equations are combined to achieve the distribution inversion of multiphase fluids; S58: Calculate the theoretical speed of sound c using the pressure-volume-temperature database and the equation of state model EoS This process combines the state equation and Equation 5. Through the error evaluation and feedback mechanism, the difference between the DAS sound velocity data and the theoretical sound velocity is calculated. The estimated values ​​of the water-liquid ratio and the gas-oil ratio are adjusted through the gradient optimization algorithm to gradually reduce the error. S59: Determine whether the current model meets the accuracy requirements through convergence check. If the error reaches the set threshold, output the final result, including the dynamic distribution of water-liquid ratio and gas-oil ratio; if the error exceeds the threshold, return to adjust the model parameters and iterate again; S510: Model Results and Evaluation When the algorithm converges, the final output results include water-to-liquid ratio, gas-to-oil ratio distribution, and sound velocity profile, reflecting the dynamic changes and distribution characteristics of the fluid in the wellbore.

6. The method for inverting multiphase flow profiles in horizontal wells based on multi-source data fusion according to claim 5, characterized in that: In step S53, the speed of sound data measured by the DAS is combined with the pressure-volume-temperature parameters to optimize the model parameters, specifically: In single-phase flow scenarios, linear regression analysis of DAS sound velocity data and pressure-volume-temperature database parameters is used to calibrate key parameters in the state equation, including critical temperature and critical pressure; In multiphase flow scenarios, the pressure-volume-temperature parameters are dynamically adjusted using an error minimization algorithm; The estimated values ​​of the fluid component ratios are optimized by fitting the sound speed calculated by the model with the actual measured data.

Images