Joint inversion method of wellbore fluid production profile based on DTS and DAS data

Abstract

This invention discloses a joint inversion method for wellbore fluid production profiles based on DTS and DAS data, relating to the technical field of oil and gas production monitoring. The key technical points are: establishing a DTS forward model and a DAS forward model, respectively; establishing a multi-objective optimization model based on minimizing the combined error between the DTS and DAS model calculation data and the measured data; and performing a joint inversion of the DTS and DAS signals using a multi-objective optimization algorithm to obtain the optimal parameters for the wellbore fluid production profile, thereby improving the accuracy of downhole injection and production profile interpretation.

Description

Technical Field

[0001] The present invention relates to the technical field of oil and gas production monitoring, and in particular to a joint inversion method for a wellbore fluid production profile based on DTS and DAS data. Background Art

[0002] Distributed temperature sensing (DTS) and distributed acoustic sensing (DAS) technologies are widely used to provide continuous and precise monitoring of reservoir conditions and have become an essential component of real-time downhole monitoring. Accurately inverting DTS and DAS data is crucial for identifying key injection pathways, optimizing the utilization of remaining oil reserves, and accurately predicting production metrics. However, the task of inverting subsurface properties and fluid dynamics from DAS and DTS data is complex, and existing inversion algorithms often exhibit inefficiency and lack robustness when dealing with forward models with large numbers of parameters and high computational costs.

[0003] With the rapid development of artificial intelligence (AI) technology, numerous machine learning methods have been applied to reservoir development. The integration of machine learning and intelligent algorithms has achieved significant progress in inverting DTS and DAS data. However, existing research often relies solely on DTS or DAS data, ignoring the potential connections between DAS and DTS data. Simultaneous inversion of both types of data has not yet been achieved, and the accuracy of interpreting downhole injection and production profiles remains to be improved. Summary of the Invention

[0004] The present invention provides a joint inversion method for wellbore fluid production profile based on DTS and DAS data, so as to solve the problem of insufficient diagnostic accuracy of wellbore water injection profile inversion based on single DAS or DTS data.

[0005] The present invention is achieved through the following technical solutions:

[0006] A first aspect of the present invention provides a joint inversion method for wellbore fluid production profile based on DTS and DAS data, comprising:

[0007] Establishing a DTS forward model and a DAS forward model respectively, wherein the DTS forward model is used to simulate the temperature and pressure distribution in the wellbore, and the DAS forward model is used to simulate the flow distribution in the wellbore;

[0008] A multi-objective optimization model is established based on the goal of minimizing the joint error between the error between the DTS calculated data and the measured data and the error between the DAS calculated data and the measured data, wherein the DTS calculated data is calculated according to the DTS forward model, and the DAS calculated data is calculated according to the DAS forward model;

[0009] The multi-objective optimization model is solved by a multi-objective optimization algorithm to obtain the optimal parameters of the wellbore fluid production profile.

[0010] The present invention establishes a forward model of distributed temperature sensing (DTS) and distributed acoustic sensing (DAS) signals, and combines it with a multi-objective optimization problem to establish a joint inversion multi-objective optimization model for DTS and DAS data for solving the liquid production profile. This model takes into account the mutual influence between low-frequency acoustic signals and temperature signals, and combines it with a multi-objective optimization algorithm to simultaneously invert DTS and DAS data. Compared with the inversion of a single DTS or DAS signal, this model improves the diagnostic accuracy of the water injection profile of horizontal water injection wells in the reservoir, thereby optimizing oilfield development strategies and increasing oil and gas recovery rates.

[0011] Furthermore, the DTS forward model includes a wellbore flow model and a wellbore thermal model; the wellbore flow model is expressed as:

[0012]

[0013] Where P is the wellbore pressure, x is the wellbore depth, ρ w is the density of the fluid in the wellbore, f is the friction coefficient of the wellbore wall, v w is the velocity of injected water in the wellbore, θ is the wellbore inclination angle, g is the acceleration of gravity, and R is the flow coefficient;

[0014] The wellbore thermal model is expressed as:

[0015]

[0016] Where T is the temperature of the reservoir wall, K JT is the Joule-Thomson coefficient, β is the correction coefficient, ρ is the wellbore density, C p is the specific heat capacity of the wellbore, α w is the total heat transfer coefficient of injected water from the wellbore to the well wall, and T1 is the initial temperature.

[0017] Furthermore, the method of establishing the DAS forward model includes:

[0018] S3-1, establish a quantitative model of DAS signal and flow rate, expressed as:

[0019]

[0020] Among them, RMS f [i,m] is the DAS signal strength generated by the pipe flow at time m and wellbore section i, RMS p [i,m] is the DAS signal intensity generated along the reservoir direction at time m and wellbore section i, θ1 and θ2 are unknown coefficients, Q i is the pipe flow rate at wellbore section i, q i is the flow rate in the reservoir direction, κi is the flow coefficient in the reservoir direction;

[0021] S3-2, establish the parameter optimization objective function, expressed as:

[0022]

[0023] Among them, RMS true [i,m] is the DAS measured signal strength at time m and wellbore section i, Q is the production profile vector, Q trial is the given value of the production profile dimension;

[0024] S3-3, based on the given production profile vector, solve the parameter optimization objective function to obtain the values ​​of the unknown coefficients θ1 and θ2, and substitute the solved θ1 and θ2 into the DAS signal and flow rate quantitative model to obtain the DAS forward model.

[0025] Furthermore, when solving the parameter optimization objective function, the undetermined coefficients θ1 and θ2 are randomly perturbed by a simulated annealing algorithm, a particle swarm optimization algorithm or a genetic algorithm, wherein θ1 is [e -6 ,e -1 ] range, θ2 is randomly perturbed in [e -8 ,e -4 ] until the values ​​of θ1 and θ2 satisfy the optimization objective function.

[0026] Furthermore, the multi-objective optimization model is expressed as:

[0027] minimize(F(x)=(f1(x), f2(x)));

[0028] Where f1(x) is the error between the DAS calculated data and the DAS measured data at the wellbore depth x, f2(x) is the error between the DTA calculated data and the DTA measured data at the wellbore depth x, and F(x) is the joint error of f1(x) and f2(x).

[0029] Furthermore, solving the multi-objective optimization model by a multi-objective optimization algorithm includes:

[0030] S6-1, Initialize particle swarm: Each particle represents a set of DAS and DTS fitting parameters, and randomly initialize the position x of a group of particles in the search space i (0) and velocity v i (0), where i = 1, 2, …, N, N is the number of particles;

[0031] S6-2, calculate the fitness of each particle, based on the Pareto dominance criterion, using Pareto rank and crowding distance to evaluate fitness;

[0032] S6-3, update the optimal position p of each particle i and the global optimal position g best :

[0033] p i =x i (t),if f j (x i ) <f j (p i (t-1));

[0034] g best =x i (t),if f j (x i ) <f j (g best (t-1));

[0035] Among them, p i represents the updated position of the i-th particle, x i (t) represents the position of the i-th particle in the t-th iteration, f j (x i ) indicates that the i-th particle is at its current position x i The j-th objective function value at j (P i (t-1)) represents the position of the i-th particle at its individual optimal position p i (t-1), that is, the j-th objective function value at the best position in the t-1-th iteration, g best represents the global optimal position of the particle swarm, f j (g best (t-1)) represents the global optimal position g best (t-1), that is, the j-th objective function value at the global optimal position in the t-1-th iteration;

[0036] S6-4, update the particle's velocity and position:

[0037] v i (t+1)=ωv i (t)+c1r1(p i (t)-x i (t))+c2r2(p i (t)-x i (t));

[0038] x i (t+1)=x i (t)+v i (t+1);

[0039] Among them, vi (t+1) is the velocity of the i-th particle in the t+1-th iteration, v i (t) is the velocity of the i-th particle in the t-th iteration, ω is the inertia weight, c1 and c2 are learning factors, r1 and r2 are random numbers in the interval [0,1], p i (t) is the optimal position of the i-th particle in the t-th iteration, x i (t+1) is the position of the i-th particle in the t+1-th iteration, x i (t) is the position of the i-th particle in the t-th iteration;

[0040] S6-5, combined with the NSGA-Ⅱ algorithm, introduces new solutions through crossover and mutation operations and updates the particle swarm;

[0041] S6-6, using the DAS forward model and the DTS forward model to evaluate the fitness of the particles in the updated particle swarm, and selecting the non-dominated solution set to update the Pareto front;

[0042] S6-7, repeat steps S6-2 to S6-6 until the maximum number of iterations is reached, and output the final Pareto front as the non-dominated solution set of DAS and DTS fitting parameters.

[0043] Furthermore, before solving the problem, the flow parameters in the fitting parameters are subjected to dimensionality reduction processing through the autoencoder to obtain low-dimensional flow features. When solving the problem, the low-dimensional flow features are used for iterative calculation, and the feature dimension of the low-dimensional flow features is less than or equal to 64.

[0044] Furthermore, the autoencoder adopts a bidirectional long short-term memory network as a basic framework, including an encoder and a decoder; flow samples with different production rates distributed along the wellbore space are generated using Perlin noise as a training set, and the encoder and decoder are jointly trained. The trained encoder is used to perform dimensionality reduction processing on the flow parameters.

[0045] The second aspect of the present invention provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for joint inversion of wellbore fluid production profiles based on DTS and DAS data as described in any one of the first aspects of the present invention is implemented.

[0046] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the joint inversion method for wellbore fluid production profile based on DTS and DAS data as described in any one of the first aspects of the present invention.

[0047] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0048] By combining the DTS and DAS forward models, a multi-objective optimization model for joint inversion is established, and a multi-objective optimization algorithm is used to solve the DTS and DAS fitting parameters. Due to the integration of the mutual relationship between DTS and DAS data, the profile interpretation accuracy is improved compared with the single DTS or DAS inversion.

[0049] A DTS forward model simulating the temperature and pressure distribution in the wellbore and a DAS forward model simulating the flow distribution in the wellbore are proposed. Through joint inversion, downhole injection and production profile information considering different signals is derived. An innovative and effective method is provided for the interpretation of the horizontal well water absorption profile, and a solution is proposed for the joint quantitative use of DTS and DAS data.

[0050] A solution method combining the NSGA-Ⅱ algorithm and hybrid multi-objective particle swarm optimization is proposed. New solutions are introduced through the hybrid strategy to prevent the algorithm from falling into local optimality, ensure the diversity and good distribution of the solution set, and select the best fitting parameters based on the optimization results.

[0051] Combined with deep learning, the flow is reduced from the original dimension to a feature space of 64 dimensions or even lower dimensions. In the subsequent steps, the optimization algorithm is solved for the feature space, which speeds up the inversion speed and improves the optimization robustness. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required for use in the examples. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be considered as limiting the scope. A person of ordinary skill in the art can also derive other relevant drawings based on these drawings without inventive effort. In the drawings:

[0053] Figure 1 This is a DTS and DAS joint inversion flow chart according to an embodiment of the present invention;

[0054] Figure 2 1 is a diagram of an autoencoder network structure according to an embodiment of the present invention;

[0055] Figure 3 is a selected DAS data graph for inversion according to an embodiment of the present invention;

[0056] Figure 4 is a selected DTS data graph for inversion according to an embodiment of the present invention;

[0057] Figure 5 is a result of a Pareto front multi-objective optimization in an embodiment of the present invention;

[0058] Figure 6 A schematic diagram of a DAS data decomposition into flow along a wellbore and flow in a reservoir according to an embodiment of the present invention;

[0059] Figure 7 This is a comparison chart of real DAS data and calculated DAS data according to an embodiment of the present invention;

[0060] Figure 8 This is a comparison chart of real DTS data and calculated DTS data according to an embodiment of the present invention;

[0061] Figure 9 is a schematic diagram of a profile interpretation method based on DAS data according to an embodiment of the present invention;

[0062] Figure 10 It is a schematic diagram of a profile interpretation method dominated by DTS data according to an embodiment of the present invention. DETAILED DESCRIPTION

[0063] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with examples and drawings. The exemplary embodiments of the present invention and their descriptions are only used to explain the present invention and are not intended to limit the present invention.

[0064] It should be noted that the terms "including" and "having" and any variations thereof in the specification and claims of the present invention and the above-mentioned drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to other steps or units inherent in the device.

[0065] The terms used in the various embodiments of the application are only used to describe the purpose of specific embodiments and are not intended to limit the various embodiments of the application. As used herein, the singular form is intended to also include the plural form, unless the context clearly indicates otherwise. Unless otherwise limited, all terms used here (including technical terms and scientific terms) have the same meaning as the meaning generally understood by those of ordinary skill in the art of the application. The terms (such as the terms defined in the dictionary generally used) will be interpreted as having the same meaning as the contextual meaning in the relevant technical field and will not be interpreted as having idealized meaning or too formal meaning, unless clearly defined in the various embodiments of the application.

[0066] An embodiment of the present invention provides a joint inversion method for wellbore fluid production profile based on DTS and DAS data, which is used to simultaneously invert distributed temperature sensing (DTS) and distributed acoustic sensing (DAS) data to improve the injection profile diagnostic accuracy of horizontal injection wells in carbonate reservoirs, optimize oilfield development strategies, and improve oil and gas recovery rates.

[0067] See also Figure 1 The DTS and DAS joint inversion flowchart shown here proposes an innovative and efficient methodology that significantly improves the diagnostic accuracy of water injection profiles and oil and gas recovery. The core innovations of this invention include the construction of a forward model, a DAS signal quantification model, an auto-encoder flow distribution model, and a hybrid multi-objective optimization algorithm.

[0068] This paper first proposes a forward model for DTS and DAS signals, establishes a multi-objective model for the joint inversion of DTS and DAS data to determine water absorption profiles, and provides a solution. Based on this process, the present invention provides an innovative and effective method for interpreting horizontal well water absorption profiles and offers a solution for the combined quantitative use of DTS and DAS data.

[0069] One implementation method includes the following steps:

[0070] S1, establish DTS forward model and DAS forward model respectively;

[0071] S2: A multi-objective optimization model is established based on the goal of minimizing the combined error between the DTS calculated data and the measured data, and the error between the DAS calculated data and the measured data. The DTS calculated data is obtained from the DTS forward model, and the DAS calculated data is obtained from the DAS forward model.

[0072] S3, solves the multi-objective optimization model of S2 through a multi-objective optimization algorithm to obtain the optimal parameters of the wellbore fluid production profile.

[0073] Among them, the DTS forward model is used to simulate the temperature and pressure distribution in the wellbore, including the wellbore flow model and the wellbore thermal model, and the DAS forward model is used to simulate the flow distribution in the wellbore.

[0074] Furthermore, based on the principles of conservation of mass, momentum and energy, an injected water wellbore flow model and a wellbore thermal model are established. The flow of the fluid includes flow along the wellbore axis and radial flow into the wellbore flow model.

[0075] 1. Wellbore flow model.

[0076] According to the law of conservation of mass and momentum, the pressure gradient equation of the injected water wellbore is expressed as:

[0077]

[0078] Where P is the wellbore pressure, x is the wellbore depth, ρ w is the density of the fluid in the wellbore, f is the friction coefficient of the wellbore wall, v wis the velocity of injected water in the wellbore, θ is the wellbore inclination angle, g is the acceleration of gravity, and R is the flow coefficient;

[0079] 2. Wellbore thermal model.

[0080] Under steady-state conditions, the steady-state wellbore temperature expression is established as:

[0081]

[0082] Where T is the temperature of the reservoir wall, K JT is the Joule-Thomson coefficient, β is the correction coefficient, ρ is the wellbore density, C p is the specific heat capacity of the wellbore, α w is the total heat transfer coefficient of injected water from the wellbore to the wellbore wall, ρ w is the density of the fluid in the wellbore, v w is the velocity of injected water in the wellbore, R is the flow coefficient, and T1 is the initial temperature.

[0083] Finally, the fourth-order Runge-Kutta format is used to solve the system of equations consisting of the pressure gradient equation and the temperature gradient equation in one go. The solution result is the temperature profile in the wellbore, that is, the temperature profile calculated by the forward model, which is used to compare with the actual monitored DTS data to realize the inversion of reservoir parameters.

[0084] Methods for establishing DAS forward models include:

[0085] S3-1, establish a quantitative model of DAS signal and flow rate, which can be expressed as:

[0086]

[0087]

[0088] Among them, RMS f [i,m] is the DAS signal strength generated by the pipe flow at time m and wellbore section i, RMS p [i,m] is the DAS signal intensity generated along the reservoir direction at time m and wellbore section i, θ1 and θ2 are unknown coefficients, Q i is the pipe flow rate, q i is the flow rate in the reservoir direction, κ i is the flow coefficient in the reservoir direction.

[0089] Specifically, the DAS signal is obtained by intercepting the relatively stable DAS data S of the horizontal section of the wellbore within a certain period of time at the end of the injection test. The data S is processed with the same parameters at each depth of the wellbore. By adjusting the length parameter of the window function, the frequency resolution is increased. The frequency resolution is generally higher than 0.5 Hz. The two-dimensional time-frequency representation of the acoustic wave data of the wellbore at the i-th depth is obtained:

[0090]

[0091] Where X[i,m,k] is the spectrum representation at borehole depth index i, time index m, and frequency index k; N is the time window length; S[n] is the discrete DAS signal, n is the time index of the discrete point; ω[nm] is the window function of length N, centered at time m, e -j2πkn / N It is the complex exponential representation of frequency k, where k = 0, 1, …, N.

[0092] Considering that the response of DAS to fluid is mainly reflected in low frequency, the low frequency interval k1-k2 is selected for each depth index i, such as the amplitude of 1-50 Hz, and the corresponding RMS (Root Mean Square) is calculated without being limited to this interval. The calculation formula is as follows:

[0093]

[0094] Wherein, K is the number of frequency points in the frequency interval, K=k2-k1+1.

[0095] S3-2, establish parameter optimization objective function,

[0096] Considering the minimum error between the calculated value and the actual value: RMS true [i,m]=RMS f [i,m]+RMS p [i,m], where RMS true [i,m] is obtained by processing the measured signal data, and the parameter optimization objective function is established, which is expressed as:

[0097]

[0098] Among them, RMS true [i,m] is the DAS measured signal strength at time m and wellbore section i, Q is the production profile vector, Q trial is the given value of the production profile dimension.

[0099] S3-3, based on the given production profile vector, solve the parameter optimization objective function to obtain the values ​​of the unknown coefficients θ1 and θ2, and substitute the solved θ1 and θ2 into the DAS signal and flow rate quantitative model to obtain the DAS forward model.

[0100] Furthermore, in the process of solving the objective function, based on the given production profile vector Q=Q trial , the optimization problem can be solved by using simulated annealing algorithm, particle swarm optimization algorithm or genetic algorithm, and the variables θ1 and θ2 are perturbed within the response frequency band amplitude range, such as θ1 in [e -6 ,e -1 ] range, θ2 is randomly perturbed in [e -8 ,e -4 ] range, but is not limited to this. The interval range may need to be selected according to the amplitude of the selected response segment. Substitute the variables θ1 and θ2 into the objective function and update the parameters according to the corresponding algorithm until the values ​​of θ1 and θ2 satisfy the requirement that the calculated values ​​are equal to the measured values ​​or are within a certain error range, thus obtaining the unknown parameters θ1 and θ2.

[0101] Finally, the response to DAS under the current production profile is calculated using the solved parameters θ1 and θ2, which is the DAS forward model. This model is denoted as DAS calculated (Q).

[0102] In the present invention, the goal of inversion is to find the minimum value of the fitting error between the original DTS and DAS data and the forward model data. The fitting error between the original DAS data and the forward model data can be expressed as:

[0103] f1(x)=min(DAS true -DAS calculated ) 2

[0104] Among them, DAS true It is measured data, DAS calculated It is the model calculation data.

[0105] The fitting error between the original DTS data and the forward model data can be expressed as:

[0106] f2(x)=min(DTS true -DTS calculated ) 2

[0107] Among them, DTS true It is measured data, DTS calculated It is the model calculation data.

[0108] The joint inversion model is described as follows:

[0109] minimize(F(x)=(f1(x),f2(x)))

[0110] Where f1(x) is the error between the DAS calculated data and the measured DAS data at wellbore depth x, f2(x) is the error between the DTA calculated data and the measured DTA data at wellbore depth x, and F(x) is the objective function vector, which is the joint error of f1(x) and f2(x), expressed as the vector form of the two errors. The constraint is x∈Ω, where Ω is the feasible region of the decision space, which is limited by the physical conditions of the wellbore.

[0111] Furthermore, the present invention adopts the hybrid multi-objective particle swarm optimization (HMPSO) algorithm to solve the multi-objective optimization model. In HMPSO, a group of particles (candidate solutions) move in the search space, adjust their positions based on their own experience and the experience of their neighbors, and gradually approach the optimal solution. Multi-objective optimization involves optimizing multiple conflicting objective functions at the same time. The Pareto front is usually used to represent the non-dominated solution set. On this basis, combined with the non-dominated sorting genetic algorithm II (NSGA-Ⅱ), new solutions are introduced through crossover and mutation operations to prevent the algorithm from falling into local optimality. The optimization steps are as follows:

[0112] S6-1, Initialize particle swarm: Each particle represents a set of DAS and DTS fitting parameters, and randomly initialize the position x of a group of particles in the search space i (0) and velocity v i (0), where i = 1, 2, …, N, and N is the number of particles.

[0113] S6-2, calculate the fitness of each particle Fitness(x i ), based on the Pareto dominance criterion, using the Pareto level (P l ) and congestion distance (C d ) to evaluate fitness.

[0114] S6-3, update the optimal position p of each particle i and the global optimal position g best :

[0115] p i =x i (t),if f j (x i ) <f j (p i (t-1))

[0116] g best =x i (t),if f j (xi ) <f j (g best (t-1))

[0117] Among them, p i represents the updated position of the i-th particle, x i (t) represents the position of the i-th particle in the t-th iteration, f j (x i ) indicates that the i-th particle is at its current position x i The j-th objective function value at j (P i (t-1)) represents the position of the i-th particle at its individual optimal position p i (t-1), that is, the j-th objective function value at the best position in the t-1-th iteration, g best represents the global optimal position of the particle swarm, f j (g best (t-1)) shows the global optimal position g best (t-1) is the jth objective function value at the global optimal position in the t-1th iteration. The objective function value of each particle is the fitting error represented by each particle. The fitness function takes the inverse of the fitting error. The smaller the error, the higher the fitness.

[0118] According to the fitness evaluation results, the most suitable group of particles is selected in the current iteration step as the parent of the next iteration for the next optimization iteration.

[0119] S6-4, update the particle's velocity and position:

[0120] v i (t+1)=ωv i (t)+c1r1(p i (t)-x i (t))+c2r2(p i (t)-x i (t))

[0121] x i (t+1)=x i (t)+v i (t+1)

[0122] Among them, v i (t+1) is the velocity of the i-th particle in the t+1-th iteration, v i (t) is the velocity of the i-th particle in the t-th iteration, ω is the inertia weight, c1 and c2 are learning factors, r1 and r2 are random numbers in the interval [0,1], p i (t) is the optimal position of the i-th particle in the t-th iteration, xi (t+1) is the position of the i-th particle in the t+1-th iteration, x i (t) is the position of the i-th particle at the t-th iteration.

[0123] S6-5, combined with the NSGA-Ⅱ algorithm, introduces new solutions through crossover and mutation operations and updates the particle swarm.

[0124] S6-6, the DAS forward model and the DTS forward model are used to evaluate the fitness of the particles in the updated particle swarm, and the non-dominated solution set is selected to update the Pareto front to ensure the diversity and good distribution of the solution set.

[0125] S6-7, repeat steps S6-2 to S6-6 until the maximum number of iterations is reached, and output the final Pareto front PF* as the non-dominated solution set of DAS and DTS fitting parameters.

[0126] Finally, according to actual needs and actual reservoir conditions, the solution in the final Pareto front PF* is selected as the fitting parameter for the joint inversion of DTS and DAS.

[0127] Furthermore, before solving the problem, the flow parameters in the fitting parameters are reduced in dimensionality through the autoencoder to obtain low-dimensional flow features. When solving the problem, the low-dimensional flow features are used for iterative calculation.

[0128] In the above-mentioned solution process using the HMPSO algorithm, the fitting parameters of DAS and DTS are the feature vectors of the reservoir parameters after compression and dimensionality reduction by the neural network, that is, the latent variable features output by the autoencoder, which are the feature vectors of the reservoir parameters after compression and dimensionality reduction by the neural network.

[0129] In the present invention, the flow rate at each depth point is actually the object that needs to be solved, but the spatial dimension of the flow solution is too large (often thousands of dimensions). Therefore, the present invention introduces an autoencoder to reduce the flow rate from the original dimension to a feature space of 64 dimensions or even lower dimensions. In subsequent steps, an optimization algorithm is performed on the feature space to speed up the inversion speed and improve the optimization robustness.

[0130] Furthermore, considering that the traffic should show a certain but limited spatial continuity, which means that there is a sequential relationship between adjacent grid points, the bidirectional long short-term memory network (Bi-LSTM) framework is adopted as the basic architecture of the Auto-encoder. Figure 2 As shown in the autoencoder network structure diagram, the auto-encoder is divided into two main modules: encoder and decoder.

[0131] The encoder and decoder parts of the Auto-encoder are used to establish the mapping x = f encoder (q) and Q = f decoder (x), where x is the latent variable output by the encoder and Q is the traffic data.

[0132] When solving multi-objective problems, the forward model of DTS and DAS is used to calculate the DTS and DAS data, and the HMPSO multi-objective optimization algorithm is used in the feature space x to calculate and measure the DTS and DAS for error minimization fitting solution.

[0133] During model training, flow rate samples with different production rates distributed along the wellbore space are generated using Perlin noise as a training set. The encoder and decoder are trained jointly. The trained encoder is used to perform dimensionality reduction processing on the flow rate parameters.

[0134] To train a robust auto-encoder, a large number of samples are required. Using simple random noise to generate flow rates at various wellbore depths is not feasible because noise can severely disrupt continuity, resulting in meaningless samples. To increase the diversity of samples and spatial relationships, the present invention uses Perlin noise to generate samples with varying but reasonable spatial distributions of production rates along the wellbore.

[0135] Through the above-mentioned joint inversion method, the present invention conducts a joint inversion experiment of wellbore fluid production profile based on DTS and DAS data.

[0136] 1. First, establish the DTS and DAS forward models.

[0137] This step is based on the principles of conservation of mass, momentum, and energy. Using these laws, the pressure gradient and temperature distribution within the wellbore are calculated, resulting in a theoretical model for DTS data. By processing the DAS data and performing spectral analysis on the DAS signal using a short-time Fourier transform (STFT), the amplitude characteristics of the low-frequency range are extracted to capture the motion of the fluid within the wellbore and establish a quantitative relationship between the DAS signal and flow velocity. Figure 3 Figure 2 shows selected DAS data used for inversion. Figure 4 Figure 2 shows selected DTS data used for inversion, where the short-time Fourier transformed DAS data are used for subsequent model optimization.

[0138] 2. Establish a multi-objective optimization model.

[0139] After completing the establishment of the DTS and DAS forward models, the two types of data, DTS and DAS, are combined and jointly inverted using a multi-objective optimization algorithm.

[0140] 3. Application of encoder-decoder neural network.

[0141] In order to effectively process high-dimensional traffic data, the Auto-encoder model is introduced to compress high-dimensional traffic distribution data into a low-dimensional feature space, reducing the computational complexity of the optimization solution.

[0142] 4. Apply multi-objective optimization algorithm.

[0143] After completing the construction of the forward model and flow distribution model of DTS and DAS data, the hybrid multi-objective particle swarm optimization algorithm (HMPSO) is used to jointly invert the DTS and DAS data to minimize the data fitting error.

[0144] During the inversion process, the fitting errors of DTS and DAS data are calculated, and the minimum value of the error is sought in the optimization algorithm. Figure 5 The Pareto front in Figure 1 shows the results of a multi-objective optimization, with each point representing a different set of inversion solutions. This optimization process incorporates a non-dominated sorting genetic algorithm (NSGA-II) to introduce new solutions through a hybrid strategy.

[0145] 5. Result analysis and optimization.

[0146] After the optimization process is completed, the inversion results are compared with the real data to evaluate the accuracy of the model. Figure 6 The DAS data are decomposed into independent contributions of flow along the wellbore and flow in the reservoir, and the real DAS data are compared with the calculated DAS data. Figure 7 The accuracy of the inversion results can be evaluated through error analysis. Figure 8 The comparison between the real DTS data and the DTS data calculated by the forward model is shown to ensure that the temperature curve is consistent with the actual situation and to verify the accuracy of the model.

[0147] After obtaining reliable inversion results, the Figure 9 The DAS data-driven interpretation approach shown and Figure 10 The DTS data-driven interpretation method shown is used to analyze the water injection profile and determine the distribution of fluids in the horizontal well.

[0148] In summary, the present invention introduces Auto-encoder technology to compress the high-dimensional space of flow distribution into a low-dimensional feature space, and then uses the optimization algorithm to solve it, thereby improving the efficiency and robustness of the inversion. The bidirectional long short-term memory network framework is used as the basic architecture of the Auto-encoder to ensure the spatial continuity and optimization effect of the flow distribution. During the joint inversion process, a hybrid multi-objective particle swarm optimization algorithm is adopted. By introducing the hybrid strategy of the NSGA-Ⅱ algorithm and combining it with the multi-objective optimization model, the error minimization fitting of DTS and DAS data is achieved, forming a complete set of processes to achieve efficient joint inversion of DTS and DAS data, significantly improving the diagnostic accuracy of the horizontal well water absorption profile.

[0149] An embodiment of the present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the method for joint inversion of wellbore fluid production profiles based on DTS and DAS data as described in any one of the above embodiments of the present invention is implemented.

[0150] Memory, as a computer-readable storage medium, can be used to store software programs, computer executable programs, and modules. The processor executes various functional applications and data processing of the electronic device by running the software programs, instructions, and modules stored in the memory. The memory may mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system and at least one application required for a function; the data storage area can store data created based on the use of the terminal, etc. In addition, the memory may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to the electronic device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, corporate intranet, local area network, mobile communication network, and combinations thereof.

[0151] Embodiments of the present invention further provide a computer-readable storage medium having a computer program stored thereon. When executed by a processor, the computer program implements the joint inversion method for wellbore fluid production profiles based on DTS and DAS data according to any embodiment of the present invention. The storage medium may be ROM / RAM, a magnetic disk, an optical disk, or the like.

[0152] An embodiment of the present invention further provides a computer program product, which, when executed on a computer, enables the computer to execute the joint inversion method for wellbore fluid production profile based on DTS and DAS data according to any of the above embodiments of the present invention.

[0153] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A joint inversion method for wellbore fluid production profile based on DTS and DAS data, characterized by: include: Establishing a DTS forward model and a DAS forward model respectively, wherein the DTS forward model is used to simulate the temperature and pressure distribution in the wellbore, and the DAS forward model is used to simulate the flow distribution in the wellbore; Establishing a joint error based on an error between DTS calculated data and measured data and an error between DAS calculated data and measured data, wherein the joint error is a vector form of the error between the DTS calculated data and measured data and the error between the DAS calculated data and measured data; wherein the DTS calculated data is calculated according to the DTS forward model, and the DAS calculated data is calculated according to the DAS forward model; Using the joint error minimization model as a multi-objective optimization model; The multi-objective optimization model is solved by a multi-objective optimization algorithm to obtain the optimal parameters of the wellbore fluid production profile.

2. The joint inversion method for wellbore fluid production profile based on DTS and DAS data according to claim 1, characterized in that: The DTS forward model includes a wellbore flow model and a wellbore thermal model; the wellbore flow model is expressed as: Where p is the wellbore pressure, x is the wellbore depth, and ρ w is the density of the fluid in the wellbore, f is the friction coefficient of the wellbore wall, v w is the velocity of injected water in the wellbore, θ is the wellbore inclination angle, g is the acceleration of gravity, and R is the flow coefficient; The wellbore thermal model is expressed as: Where T is the temperature of the reservoir wall, K JT is the Joule-Thomson coefficient, β is the correction coefficient, ρ is the wellbore density, C p is the specific heat capacity of the wellbore, α w is the total heat transfer coefficient of injected water from the wellbore to the well wall, and T1 is the initial temperature.

3. The joint inversion method for wellbore fluid production profile based on DTS and DAS data according to claim 1, characterized in that: Methods for establishing DAS forward models include: S3-1, establish a quantitative model of DAS signal and flow rate, expressed as: Among them, RMS f [i,m] is the DAS signal strength generated by the pipe flow at time m and wellbore section i, RMS p [i,m] is the DAS signal intensity generated along the reservoir direction at time m and wellbore section i, θ1 and θ2 are unknown coefficients, Q i is the pipe flow rate at wellbore section i, q i is the flow rate in the reservoir direction, κ i is the flow coefficient in the reservoir direction; S3-2, establish the parameter optimization objective function, expressed as: Among them, RMS true [i,m] is the DAS measured signal strength at time m and wellbore section i, Q is the production profile vector, Q trial is the given value of the production profile dimension; S3-3, based on the given production profile vector, solve the parameter optimization objective function to obtain the values ​​of the unknown coefficients θ1 and θ2, and substitute the solved θ1 and θ2 into the DAS signal and flow rate quantitative model to obtain the DAS forward model.

4. The joint inversion method for wellbore fluid production profile based on DTS and DAS data according to claim 3, characterized in that: When solving the parameter optimization objective function, the undetermined coefficients θ1 and θ2 are randomly perturbed by simulated annealing algorithm, particle swarm optimization algorithm or genetic algorithm, wherein θ1 is [e -6 ,e -1 ] range, θ2 is randomly perturbed in [e -8 ,e -4 ] until the values ​​of θ1 and θ2 satisfy the optimization objective function.

5. The joint inversion method for wellbore fluid production profile based on DTS and DAS data according to claim 1, characterized in that: The multi-objective optimization model is expressed as: minimize(F(x)=(f1(x), f2(x))); Where f1(x) is the error between the DAS calculated data and the DAS measured data at the wellbore depth x, f2(x) is the error between the DTA calculated data and the DTA measured data at the wellbore depth x, and F(x) is the joint error of f1(x) and f2(x).

6. The joint inversion method for wellbore fluid production profile based on DTS and DAS data according to claim 1, characterized in that: Solving the multi-objective optimization model by a multi-objective optimization algorithm includes: S6-1, Initialize particle swarm: Each particle represents a set of DAS and DTS fitting parameters, and randomly initialize the position x of a group of particles in the search space i (0) and velocity v i (0), where i = 1, 2, …, N, N is the number of particles; S6-2, calculate the fitness of each particle, based on the Pareto dominance criterion, using Pareto rank and crowding distance to evaluate fitness; S6-3, update the optimal position p of each particle i and the global optimal position g best : p i =x i (t),if f j (x i )<f j (p i (t-1)); g best =x i (t),if f j (x i )<f j (g best (t-1)); Among them, p i represents the updated position of the i-th particle, x i (t) represents the position of the i-th particle in the t-th iteration, f j (x i ) indicates that the i-th particle is at its current position x i The j-th objective function value at j (p i (t-1)) represents the position of the i-th particle at its individual optimal position p i (t-1), that is, the j-th objective function value at the best position in the t-1-th iteration, g best represents the global optimal position of the particle swarm, f j (g best (t-1)) represents the global optimal position g best (t-1), that is, the j-th objective function value at the global optimal position in the t-1-th iteration; S6-4, update the particle's velocity and position: v i (t+1)=ωv i (t)+c1r1(p i (t)-x i (t))+c2r2(p i (t)-x i (t)); x i (t+1)=x i (t)+v i (t+1); Among them, v i (t+1) is the velocity of the i-th particle in the t+1-th iteration, v i (t) is the velocity of the i-th particle in the t-th iteration, ω is the inertia weight, c1 and c2 are learning factors, r1 and r2 are random numbers in the interval [0,1], p i (t) is the optimal position of the i-th particle in the t-th iteration, x i (t+1) is the position of the i-th particle in the t+1-th iteration, x i (t) is the position of the i-th particle in the t-th iteration; S6-5, combined with the NSGA-Ⅱ algorithm, introduces new solutions through crossover and mutation operations and updates the particle swarm; S6-6, using the DAS forward model and the DTS forward model to evaluate the fitness of the particles in the updated particle swarm, and selecting the non-dominated solution set to update the Pareto front; S6-7, repeat steps S6-2 to S6-6 until the maximum number of iterations is reached, and output the final Pareto front as the non-dominated solution set of DAS and DTS fitting parameters.

7. The joint inversion method for wellbore fluid production profile based on DTS and DAS data according to claim 6, characterized in that: Before solving, the flow parameters in the fitting parameters are subjected to dimensionality reduction processing through the autoencoder to obtain low-dimensional flow features. When solving, the low-dimensional flow features are used for iterative calculation, and the feature dimension of the low-dimensional flow features is less than or equal to 64.

8. The joint inversion method for wellbore fluid production profile based on DTS and DAS data according to claim 7, characterized in that: The autoencoder uses a bidirectional long short-term memory network as its basic framework, including an encoder and a decoder. Flow samples with different production rates distributed along the wellbore space are generated using Perlin noise as a training set. The encoder and decoder are jointly trained, and the trained encoder is used to perform dimensionality reduction processing on flow parameters.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method for joint inversion of wellbore fluid production profile based on DTS and DAS data as described in any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for joint inversion of wellbore fluid production profile based on DTS and DAS data as described in any one of claims 1 to 8 is implemented.

Images