Data-driven hydraulic parameter real-time optimization method and system

Abstract

The invention relates to the technical field of parameter optimization, in particular to a data-driven hydraulic parameter real-time optimization method and system, and the method comprises the steps: obtaining drilling multi-source heterogeneous data, and carrying out the cleaning and multi-modal space-time alignment fusion, and generating a standardized multiphase flow parameter; inputting a pre-trained multiphase flow transient model through a data-model combined driving mechanism, dynamically correcting fluid density distribution and phase change parameters, and outputting a transient simulation result matched with a drilling working condition in real time; and determining hydraulic parameters to be optimized based on a simulation result, and generating a real-time optimization scheme of optimal displacement, throttling pressure and well killing parameters by combining wellbore pressure constraint and a multi-objective optimization function and adopting an adaptive particle swarm optimization algorithm for iterative calculation. According to the method, multi-source data space-time alignment and model dynamic correction are achieved, transient simulation precision is improved through a combined driving mechanism, and an optimal hydraulic parameter scheme meeting wellbore safety constraints can be generated by combining an adaptive optimization algorithm.

Description

Technical Field

[0001] The present invention relates to the technical field of parameter optimization, and in particular to a data-driven real-time optimization method and system for hydraulic parameters. Background Art

[0002] Existing deepwater drilling hydraulic parameter optimization methods have problems with inconsistent spatiotemporal benchmarks and low data cleaning efficiency when fusing multi-source heterogeneous data, making it difficult to align measurement while drilling data with downhole sensor data in real time. Traditional multiphase flow models are limited by static parameter settings and cannot dynamically correct fluid density distribution and phase change rate according to drilling conditions, resulting in large deviations in the prediction of wellbore pressure fluctuations. In addition, existing optimization algorithms do not combine wellbore pressure constraints with multi-objective collaborative mechanisms. The use of a single-objective gradient descent method causes the optimization results of parameters such as displacement and throttling pressure to deviate from the actual safety threshold, thereby increasing the incidence of well killing accidents. In addition, the decision response time is long and cannot meet the real-time control needs of deepwater drilling.

[0003] Therefore, it is urgent to provide a technical solution to solve the above problems. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention provides a data-driven real-time optimization method and system for hydraulic parameters.

[0005] In a first aspect, the present invention provides a data-driven real-time optimization method for hydraulic parameters, the technical solution of which is as follows:

[0006] Acquire multi-source heterogeneous data during the drilling process and convert the multi-source heterogeneous data into standardized multiphase flow parameters with a unified time base through data cleaning and multi-modal spatiotemporal alignment fusion algorithm;

[0007] Using a data-model joint driving mechanism, the standardized multiphase flow parameters are input into a pre-trained multiphase flow transient simulation model, and the fluid density distribution and phase change parameters in the multiphase flow transient simulation model are updated through a dynamic parameter correction algorithm to generate transient simulation results that match the current drilling conditions;

[0008] Based on the transient simulation results, the hydraulic parameters to be optimized are determined, and according to the wellbore pressure constraints and the multi-objective optimization function, the hydraulic parameters are iteratively calculated using an adaptive particle swarm optimization algorithm to generate a real-time optimization solution including the optimal displacement, throttling pressure, and well killing parameters.

[0009] The beneficial effects of the data-driven real-time optimization method of hydraulic parameters of the present invention are as follows:

[0010] The method of the present invention realizes the spatiotemporal alignment of multi-source data and dynamic correction of the model, improves the transient simulation accuracy through a joint driving mechanism, and can generate the optimal hydraulic parameter scheme that meets the wellbore safety constraints by combining an adaptive optimization algorithm.

[0011] Based on the above solution, the data-driven real-time optimization method of hydraulic parameters of the present invention can be further improved as follows.

[0012] In an optional manner, the step of converting the multi-source heterogeneous data into standardized multiphase flow parameters with a unified time base through data cleaning and multimodal spatiotemporal alignment fusion algorithm includes:

[0013] Eliminate outliers and fill in missing values ​​for measurement while drilling data, downhole sensor data, and geomechanics data in the multi-source heterogeneous data to extract valid data segments;

[0014] According to the drilling fluid circulation period and formation depth information, a dynamic time warping algorithm is used to align the timestamps of the valid data segments to eliminate the transmission delay difference of the multi-source heterogeneous data;

[0015] Based on the wellbore spatial coordinate system transformation model, downhole temperature, pressure, and flow rate data of different sampling frequencies are mapped to a unified spatial grid to generate a spatiotemporally continuous multimodal data matrix.

[0016] The multimodal data matrix is ​​subjected to dimensionality reduction processing by a multi-scale feature fusion algorithm, and the standardized multiphase flow parameters including wellbore pressure gradient, fluid flow rate, gas content and pressure fluctuation data are output.

[0017] In the above optional method, the dynamic time warping algorithm is used to eliminate the delay difference in multi-source data transmission, and the spatial grid mapping is combined to achieve accurate spatiotemporal alignment of downhole temperature, pressure, and flow rate data. The multi-scale feature fusion is used to reduce the dimension of the original data to core parameters including pressure gradient, flow rate, and gas content, thereby improving the efficiency of the standardized processing of multiphase flow parameters and providing high-confidence input for transient simulation.

[0018] In an optional manner, the steps of inputting the standardized multiphase flow parameters into a pre-trained multiphase flow transient simulation model using a data-model joint driving mechanism, and updating the fluid density distribution and phase change parameters in the multiphase flow transient simulation model using a dynamic parameter correction algorithm to generate a transient simulation result matching the current drilling conditions include:

[0019] Inputting the standardized multiphase flow parameters into a fluid density distribution correction algorithm, and generating a corrected target fluid density distribution based on the correlation between the wellbore pressure gradient and the fluid flow rate;

[0020] Based on the gas fraction, updating the phase change parameters in the multiphase flow transient simulation model by using a Bayesian inference algorithm to obtain target phase change parameters;

[0021] Using the target fluid density distribution and the target phase change parameters, combined with wellbore geometric constraints, the multiphase flow transient simulation model is run to generate transient pressure propagation simulation results;

[0022] The transient pressure propagation simulation results are compared with the pressure fluctuation data in the standardized multiphase flow parameters. If the deviation exceeds a preset deviation threshold, the iterative update of the model weight coefficient is triggered until the transient simulation results matching the current drilling conditions are generated.

[0023] In the above optional method, the Bayesian inference algorithm is used to correct the phase change parameters in real time, and the dynamic update of the multiphase flow model parameters is realized in combination with the fluid density distribution correction algorithm, so that the deviation between the transient pressure propagation simulation results and the measured pressure fluctuations is reduced, thereby effectively supporting the real-time and accurate mapping of the downhole working conditions.

[0024] In an optional manner, the steps of determining hydraulic parameters to be optimized based on the transient simulation results, and iteratively calculating the hydraulic parameters using an adaptive particle swarm optimization algorithm according to wellbore pressure constraints and a multi-objective optimization function to generate a real-time optimization solution including optimal displacement, throttling pressure, and well killing parameters include:

[0025] Extracting the wellbore pressure gradient change rate, fluid velocity non-uniformity coefficient, and gas content risk threshold from the transient simulation results, and determining the range of displacement, throttling pressure, and killing fluid density parameters to be optimized;

[0026] Constructing the multi-objective optimization function including maximizing mechanical drilling rate, minimizing well killing cost, and minimizing wellbore pressure fluctuation; wherein the weight coefficient of each objective is dynamically adjusted according to the gas content risk threshold;

[0027] Initializing the particle swarm population, using the displacement, throttling pressure and killing fluid density parameters as particle position vectors, and setting particle velocity boundaries based on the current wellbore pressure constraint;

[0028] During the iteration process, the inertia weight and learning factor are adaptively adjusted according to the difference between the global optimal solution and the individual historical optimal solution of the particle, and the particle positions that deviate from the wellbore pressure constraint and the particle velocity boundary are updated first;

[0029] When the weighted variance of the multi-objective optimization function after a preset number of consecutive iterations is less than a preset variance threshold, the solution with the highest safety weight in the Pareto front is output as the real-time optimization solution.

[0030] In the above optional method, the multi-objective weights are dynamically adjusted through the gas content risk threshold, and the adaptive particle swarm algorithm is combined to quickly converge to the Pareto optimal solution under the wellbore pressure constraint, shortening the optimization decision time for displacement, throttling pressure, and well-killing fluid density, while also reducing the incidence of well-killing accidents.

[0031] In a second aspect, the present invention provides a data-driven real-time optimization system for hydraulic parameters, the technical solution of which is as follows:

[0032] The data-driven hydraulic parameter real-time optimization system includes: an initialization module, a parameter updating module and a solution generation module;

[0033] The initialization module is used to: acquire multi-source heterogeneous data during the drilling process, and convert the multi-source heterogeneous data into standardized multiphase flow parameters with a unified time reference through data cleaning and multi-modal spatiotemporal alignment fusion algorithm;

[0034] The parameter updating module is configured to input the standardized multiphase flow parameters into a pre-trained multiphase flow transient simulation model using a data-model joint driving mechanism, and update the fluid density distribution and phase change parameters in the multiphase flow transient simulation model using a dynamic parameter correction algorithm to generate a transient simulation result that matches the current drilling condition;

[0035] The solution generation module is used to: determine the hydraulic parameters to be optimized based on the transient simulation results, and iteratively calculate the hydraulic parameters using an adaptive particle swarm optimization algorithm according to the wellbore pressure constraints and the multi-objective optimization function to generate a real-time optimization solution including the optimal displacement, throttling pressure and well killing parameters.

[0036] The beneficial effects of the data-driven real-time optimization system for hydraulic parameters of the present invention are as follows:

[0037] The system of the present invention realizes the spatiotemporal alignment of multi-source data and dynamic correction of models, improves the transient simulation accuracy through a joint driving mechanism, and can generate the optimal hydraulic parameter scheme that meets the wellbore safety constraints by combining an adaptive optimization algorithm.

[0038] Based on the above solution, the data-driven hydraulic parameter real-time optimization system of the present invention can also be improved as follows.

[0039] In an optional manner, the initialization module is specifically configured to:

[0040] Eliminate outliers and fill in missing values ​​for measurement while drilling data, downhole sensor data, and geomechanics data in the multi-source heterogeneous data to extract valid data segments;

[0041] According to the drilling fluid circulation period and formation depth information, a dynamic time warping algorithm is used to align the timestamps of the valid data segments to eliminate the transmission delay difference of the multi-source heterogeneous data;

[0042] Based on the wellbore spatial coordinate system transformation model, downhole temperature, pressure, and flow rate data of different sampling frequencies are mapped to a unified spatial grid to generate a spatiotemporally continuous multimodal data matrix.

[0043] The multimodal data matrix is ​​subjected to dimensionality reduction processing by a multi-scale feature fusion algorithm, and the standardized multiphase flow parameters including wellbore pressure gradient, fluid flow rate, gas content and pressure fluctuation data are output.

[0044] In the above optional method, the dynamic time warping algorithm is used to eliminate the delay difference in multi-source data transmission, and the spatial grid mapping is combined to achieve accurate spatiotemporal alignment of downhole temperature, pressure, and flow rate data. The multi-scale feature fusion is used to reduce the dimension of the original data to core parameters including pressure gradient, flow rate, and gas content, thereby improving the efficiency of the standardized processing of multiphase flow parameters and providing high-confidence input for transient simulation.

[0045] In an optional manner, the parameter updating module is specifically configured to:

[0046] Inputting the standardized multiphase flow parameters into a fluid density distribution correction algorithm, and generating a corrected target fluid density distribution based on the correlation between the wellbore pressure gradient and the fluid flow rate;

[0047] Based on the gas fraction, updating the phase change parameters in the multiphase flow transient simulation model by using a Bayesian inference algorithm to obtain target phase change parameters;

[0048] Using the target fluid density distribution and the target phase change parameters, combined with wellbore geometric constraints, the multiphase flow transient simulation model is run to generate transient pressure propagation simulation results;

[0049] The transient pressure propagation simulation results are compared with the pressure fluctuation data in the standardized multiphase flow parameters. If the deviation exceeds a preset deviation threshold, the iterative update of the model weight coefficient is triggered until the transient simulation results matching the current drilling conditions are generated.

[0050] In the above optional method, the Bayesian inference algorithm is used to correct the phase change parameters in real time, and the dynamic update of the multiphase flow model parameters is realized in combination with the fluid density distribution correction algorithm, so that the deviation between the transient pressure propagation simulation results and the measured pressure fluctuations is reduced, thereby effectively supporting the real-time and accurate mapping of the downhole working conditions.

[0051] In an optional manner, the solution generation module is specifically configured to:

[0052] Extracting the wellbore pressure gradient change rate, fluid velocity non-uniformity coefficient, and gas content risk threshold from the transient simulation results, and determining the range of displacement, throttling pressure, and killing fluid density parameters to be optimized;

[0053] Constructing the multi-objective optimization function including maximizing mechanical drilling rate, minimizing well killing cost, and minimizing wellbore pressure fluctuation; wherein the weight coefficient of each objective is dynamically adjusted according to the gas content risk threshold;

[0054] Initializing the particle swarm population, using the displacement, throttling pressure and killing fluid density parameters as particle position vectors, and setting particle velocity boundaries based on the current wellbore pressure constraint;

[0055] During the iteration process, the inertia weight and learning factor are adaptively adjusted according to the difference between the global optimal solution and the individual historical optimal solution of the particle, and the particle positions that deviate from the wellbore pressure constraint and the particle velocity boundary are updated first;

[0056] When the weighted variance of the multi-objective optimization function after a preset number of consecutive iterations is less than a preset variance threshold, the solution with the highest safety weight in the Pareto front is output as the real-time optimization solution.

[0057] In the above optional method, the multi-objective weights are dynamically adjusted through the gas content risk threshold, and the adaptive particle swarm algorithm is combined to quickly converge to the Pareto optimal solution under the wellbore pressure constraint, shortening the optimization decision time for displacement, throttling pressure, and well-killing fluid density, while also reducing the incidence of well-killing accidents.

[0058] In a third aspect, the technical solution of an electronic device of the present invention is as follows:

[0059] The invention comprises a memory, a processor and a program stored in the memory and running on the processor. When the processor executes the program, the steps of the data-driven hydraulic parameter real-time optimization method of the present invention are realized.

[0060] In a fourth aspect, the present invention provides a computer-readable storage medium having the following technical solution:

[0061] The computer-readable storage medium stores instructions, and when the computer-readable storage medium reads the instructions, the computer-readable storage medium executes the steps of the data-driven hydraulic parameter real-time optimization method of the present invention.

[0062] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are specifically listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] The accompanying drawings are only used to illustrate the embodiments and are not to be considered as limiting the present invention. In addition, the same reference symbols are used to represent the same components throughout the drawings. In the drawings:

[0064] Figure 1 Schematic diagram of a flow chart of an embodiment of a data-driven hydraulic parameter real-time optimization method of the present invention;

[0065] Figure 2 Schematic diagram of the structure of an embodiment of a data-driven hydraulic parameter real-time optimization system of the present invention;

[0066] Figure 3 The figure is a schematic structural diagram of an embodiment of an electronic device of the present invention. DETAILED DESCRIPTION

[0067] The exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0068] Figure 1 FIG. 1 shows a flow chart of an embodiment of a data-driven hydraulic parameter real-time optimization method provided by the present invention, as shown in FIG. Figure 1 As shown, the following steps are included:

[0069] S1: Acquire multi-source heterogeneous data during the drilling process and convert them into standardized multiphase flow parameters with a unified time base through data cleaning and multi-modal spatiotemporal alignment fusion algorithm. In S1:

[0070] Multi-source heterogeneous data refers to a collection of raw data collected during the drilling process from various devices (such as measurement while drilling instruments, downhole sensors, and geomechanical systems) in different formats, sampling frequencies, and time and space references. This data provides diverse information input on drilling conditions, encompassing both real-time dynamic parameters (such as pressure and temperature) and static geological parameters (such as formation depth), providing the foundation for subsequent fusion analysis.

[0071] Data cleaning is a preprocessing process that controls the quality of raw data by removing outliers (e.g., filtering outliers) and filling missing values ​​(e.g., interpolation). Data cleaning eliminates noise, extracts valid data segments, and ensures the integrity and reliability of input data.

[0072] The multimodal spatiotemporal alignment and fusion algorithm is a composite data processing method that combines a dynamic time warping algorithm (to eliminate transmission delays) with a wellbore spatial coordinate system transformation model (to unify spatial grid mapping). Its purpose is to unify data with different time bases and spatial resolutions into a standardized spatiotemporal framework, resolving spatiotemporal inconsistencies in multi-source data.

[0073] A unified time base refers to a synchronized time series established using timestamp alignment technology, based on the drilling fluid circulation cycle. This unified time base eliminates timing misalignments caused by transmission delays in multi-source data, ensuring consistency in the data's time dimension.

[0074] Standardized multiphase flow parameters are a standardized dataset generated through dimensionality reduction, containing key parameters such as wellbore pressure gradient, fluid velocity, and gas fraction. These parameters provide structured input data for subsequent multiphase flow models, supporting the real-time calculation and optimization of hydraulic parameters.

[0075] Specifically, the multi-source heterogeneous data generated during the drilling process are first acquired through measurement while drilling, downhole sensors, and geomechanical systems. Subsequently, the data are cleaned by removing outliers and filling missing values ​​to extract valid data segments. A dynamic time warping algorithm is then used to align the timestamps of the multi-source data according to the formation depth and drilling fluid circulation period to eliminate transmission delay differences. At the same time, the temperature, pressure, and flow rate data with different sampling frequencies are mapped to a unified spatial grid through the wellbore spatial coordinate system conversion model to form a spatiotemporally continuous multimodal data matrix. Finally, a multi-scale feature fusion algorithm is used to reduce the dimensionality of the matrix and output standardized multiphase flow parameters including wellbore pressure gradient, fluid flow rate, gas content, and pressure fluctuation data, providing a consistent input benchmark for subsequent model calculations.

[0076] S2. Using the data-model joint driving mechanism, the standardized multiphase flow parameters are input into the pre-trained multiphase flow transient simulation model. The fluid density distribution and phase change parameters in the multiphase flow transient simulation model are updated through the dynamic parameter correction algorithm to generate transient simulation results that match the current drilling conditions. In S2:

[0077] The data-model joint drive mechanism is a technical framework that enables real-time data and physical models to work together. It implements a two-way feedback loop between data input and model prediction through dynamic parameter correction. Its purpose is to integrate the dynamic characteristics of real-time data with prior knowledge of physical models, improving the adaptability of simulation results to actual working conditions.

[0078] The pretrained multiphase transient flow simulation model is a numerical model pre-trained based on historical drilling data and multiphase flow theory. It simulates the dynamic processes of fluid pressure, flow rate, and phase changes within the wellbore. The multiphase transient flow simulation model provides an initial simulation benchmark, rapidly responds to real-time data input, and generates preliminary transient pressure propagation predictions.

[0079] The dynamic parameter correction algorithm is a real-time parameter update method that combines Bayesian inference with fluid dynamics constraints to correct the density distribution and phase change parameters in the model. The dynamic parameter correction algorithm eliminates deviations between model predictions and measured data, ensuring that simulation results are physically consistent with current drilling conditions.

[0080] Fluid density distribution describes the spatial variation of fluid density at different locations within a wellbore. It is affected by temperature, pressure, and phase changes. Fluid density distribution directly impacts the accuracy of wellbore pressure gradient calculations and is a core input variable for transient simulations.

[0081] Phase change parameters are physical quantities that characterize the gas-liquid two-phase conversion rate and the dynamic characteristics of the phase interface in multiphase flow, such as the rate of change of gas void fraction. Phase change parameters determine the multiphase flow model's ability to simulate complex conditions such as gas invasion and well kick.

[0082] Transient simulation results are parameter-corrected, time-series data on wellbore pressure propagation, including key indicators such as pressure fluctuation amplitude and propagation velocity. These results provide a basis for dynamic wellbore pressure field distribution for hydraulic parameter optimization.

[0083] Specifically, the standardized multiphase flow parameters are first input into the pre-trained multiphase flow transient simulation model through the data-model joint driving mechanism. The fluid density distribution correction algorithm is combined with the correlation between the wellbore pressure gradient and the flow velocity to generate a corrected density distribution. At the same time, the phase change parameters in the model are updated through the Bayesian inference algorithm based on the gas content data. Subsequently, the corrected density distribution and phase change parameters are substituted into the model, and the transient simulation is run in combination with the wellbore geometric structure constraints to generate the pressure propagation time series results. The deviation of the simulation results is compared with the measured pressure fluctuation data. If the deviation exceeds the threshold, the iterative update of the model weight coefficient is triggered until the transient simulation result that matches the dynamic pressure characteristics of the current drilling conditions is output.

[0084] S3. Based on the transient simulation results, the hydraulic parameters to be optimized are determined. Based on the wellbore pressure constraints and the multi-objective optimization function, the adaptive particle swarm optimization algorithm is used to iteratively calculate the hydraulic parameters. A real-time optimization solution is generated that includes the optimal displacement, throttling pressure, and well killing parameters. In S3:

[0085] Wellbore pressure constraints are defined as upper and lower pressure limits set during drilling to ensure wellbore stability. These include dynamic thresholds such as formation fracture pressure and pore pressure. Wellbore pressure constraints are used to limit the optimization range of hydraulic parameters and prevent downhole risks such as kicks and lost circulation.

[0086] The multi-objective optimization function is a composite objective function that combines maximizing ROP, minimizing kill costs, and minimizing wellbore pressure fluctuations. It dynamically balances these conflicts through weighted coefficients. This function quantifies engineering requirements into a mathematical optimization problem, driving the search for optimal parameters.

[0087] The adaptive particle swarm optimization algorithm is an intelligent optimization algorithm that dynamically adjusts the inertia weight and learning factor based on the difference between the global optimal solution and individual historical solutions during the particle swarm iteration process. The adaptive particle swarm optimization algorithm is used to efficiently search for non-inferior solutions under complex constraints, avoiding local optima.

[0088] The particle position vector is a vector space composed of displacement, throttling pressure, and killing fluid density, representing candidate solutions for the optimization parameters. The particle position vector can map hydraulic parameters into mathematical variables that can be iteratively optimized.

[0089] The Pareto front is the set of all non-inferior solutions (those that cannot improve one objective without compromising others) in multi-objective optimization. It is used to provide compromise solutions with the highest safety weights, supporting engineering decision-making.

[0090] Specifically, the wellbore pressure gradient change rate, fluid velocity non-uniformity coefficient and gas content risk threshold are first extracted from the transient simulation results to determine the optimizable parameter ranges of displacement, throttling pressure and well-killing fluid density. Subsequently, a multi-objective optimization function is constructed, which includes minimizing the mechanical drilling rate, well-killing cost and pressure fluctuation, and the objective weight coefficient is dynamically adjusted according to the gas content risk. The particle swarm is initialized, and the parameters to be optimized are used as the particle position vector. The particle velocity boundary is set based on the wellbore pressure constraint. During the iterative process, the inertia weight and learning factor are adaptively adjusted to prioritize the update of the particle positions that deviate from both the pressure constraint and the velocity boundary. The iteration is terminated when the weighted variance of the multi-objective optimization function for a preset number of consecutive times is less than the threshold. Finally, the solution with the highest safety weight is selected from the Pareto front as the real-time optimization scheme for the optimal displacement, throttling pressure and well-killing parameters.

[0091] The technical solution of this embodiment realizes the spatiotemporal alignment of multi-source data and dynamic correction of the model, improves the transient simulation accuracy through a joint driving mechanism, and can generate the optimal hydraulic parameter solution that meets the wellbore safety constraints by combining an adaptive optimization algorithm.

[0092] In an optional manner, the steps of converting multi-source heterogeneous data into standardized multiphase flow parameters with a unified time base through data cleaning and multimodal spatiotemporal alignment fusion algorithm include:

[0093] Eliminate outliers and fill missing values ​​in measurement while drilling data, downhole sensor data, and geomechanics data from multi-source heterogeneous data to extract valid data segments;

[0094] Based on the drilling fluid circulation period and formation depth information, a dynamic time warping algorithm is used to align the timestamps of valid data segments to eliminate the transmission delay difference of multi-source heterogeneous data;

[0095] Based on the wellbore spatial coordinate system transformation model, downhole temperature, pressure, and flow rate data of different sampling frequencies are mapped to a unified spatial grid to generate a spatiotemporally continuous multimodal data matrix.

[0096] The multimodal data matrix is ​​reduced in dimensionality using a multi-scale feature fusion algorithm, and standardized multiphase flow parameters including wellbore pressure gradient, fluid velocity, gas fraction, and pressure fluctuation data are output.

[0097] In this embodiment, after obtaining the measurement while drilling data, downhole sensor data and geomechanical data, the outlier removal operation is first performed: based on the dynamic change characteristics of the wellbore pressure gradient, an adaptive sliding window method is used to establish an anomaly detection mechanism. Specifically, the window length is 1.2 times the drilling fluid circulation period of the current well depth, and the standard deviation of the pressure gradient data in the window is calculated in real time. When a data point deviates from the mean by more than 2.8 times the standard deviation, it is determined to be an outlier.

[0098] To fill missing values, a Kriging interpolation algorithm was used based on the spatiotemporal correlation between formation depth and drilling fluid velocity. Spatial weights were determined by an exponential decay function of the well depth coordinate difference, while temporal weights were tied to the correlation of the drilling fluid return time series. The final output was a valid data segment containing complete spatiotemporal attributes.

[0099] The valid data segments are input into the dynamic time warping algorithm for timestamp alignment. The algorithm uses the drilling fluid circulation period as the baseline constraint and divides the data alignment interval according to the formation depth stratification results. The time series of the measurement while drilling data and the downhole sensor data are nonlinearly matched. The maximum time offset is limited to 15% of the drilling fluid return time corresponding to the current well depth. The algorithm then generates aligned data with a unified time base by minimizing the cumulative path error.

[0100] A spatial grid mapping model is established based on the wellbore cylindrical coordinate system: the wellbore radial direction is gridded with an accuracy of 5% of the casing diameter, and the vertical grid boundaries coincide with the lithologic interfaces of the formation. A cubic spline interpolation algorithm weighted by the formation thermal conductivity is used to complete spatial mapping for downhole temperature, pressure, and velocity data at different sampling frequencies. The interpolation weight of high-frequency data increases with increasing formation thermal conductivity, while the interpolation weight of low-frequency data is dynamically adjusted based on formation permeability. The specific formula is as follows:

[0101]

[0102] Among them, d r is the radial coordinate; d z is the vertical well depth coordinate; B k (d r , d z ) is the cubic B-spline basis function; β k is the interpolation coefficient, which is iteratively solved according to the data gradient of adjacent grid points and the formation thermal conductivity matrix to generate a multimodal data matrix with consistent temporal and spatial resolution.

[0103] Multi-scale feature fusion is implemented on the multimodal data matrix: first, wavelet packet decomposition is used to extract the low-frequency trend component, medium-frequency periodic component and high-frequency transient component of the pressure fluctuation, and the energy proportion of each frequency band is calculated as the characteristic index.

[0104] At the same time, kernel principal component analysis is used to perform nonlinear dimensionality reduction on the temperature-pressure-flow rate coupling matrix, and the principal components with cumulative variance contribution exceeding 90% are retained.

[0105] Finally, the above features are dynamically weighted and fused according to the formation fracture pressure threshold, where the low-frequency pressure fluctuation feature weight is positively correlated with the formation compressive strength of the current well depth, and the core principal component weight is adaptively adjusted with the fluid flow rate change rate, outputting a standardized multiphase flow parameter set including wellbore pressure gradient, fluid flow rate, gas content and pressure fluctuation spectrum characteristics. Specifically, the wavelet energy feature vector and the core principal component vector are finally spliced ​​with the formation fracture pressure weighted feature, and the weight coefficient w is based on the formation fracture pressure threshold P at the current well depth. f According to W=0.6·(P f / P max )+0.4. Among them, P max The maximum historical fracture pressure value of the current well section is output, and the standardized multiphase flow parameters including wellbore pressure gradient, fluid flow rate, gas content and pressure fluctuation spectrum characteristics are output.

[0106] In an optional approach, a data-model joint driving mechanism is used to input standardized multiphase flow parameters into a pre-trained multiphase flow transient simulation model, and a dynamic parameter correction algorithm is used to update the fluid density distribution and phase change parameters in the multiphase flow transient simulation model to generate transient simulation results that match the current drilling conditions. The steps include:

[0107] The standardized multiphase flow parameters are input into the fluid density distribution correction algorithm, and the corrected target fluid density distribution is generated based on the correlation between the wellbore pressure gradient and the fluid flow rate;

[0108] Based on the gas fraction, the phase change parameters in the multiphase flow transient simulation model are updated using the Bayesian inference algorithm to obtain the target phase change parameters.

[0109] Using the target fluid density distribution and target phase change parameters, combined with wellbore geometry constraints, a multiphase flow transient simulation model is run to generate transient pressure propagation simulation results.

[0110] The transient pressure propagation simulation results are compared with the pressure fluctuation data in the standardized multiphase flow parameters. If the deviation exceeds the preset deviation threshold, the iterative update of the model weight coefficient is triggered until a transient simulation result matching the current drilling conditions is generated.

[0111] In this embodiment, the wellbore pressure gradient and fluid velocity data in the standardized multiphase flow parameters are input into the fluid density distribution correction algorithm. First, a pressure gradient-velocity correlation matrix is ​​established, in which the density correction value Δρ at each grid point is calculated using the following formula:

[0112]

[0113] Wherein, α is the formation permeability correlation coefficient, and the value range of α (e.g., 0.02-0.15) can be determined based on the formation lithologic characteristics of the current well section; is the vertical pressure gradient; v is the flow velocity; ε is a minimum constant to prevent division by zero.

[0114] The initial fluid density distribution is updated grid by grid based on the density correction. When the mean square error of the density distribution between two adjacent iterations is less than 0.1 kg / m 3 Output target fluid density distribution;

[0115] Based on the target fluid density distribution, the gas fraction data is input into the Bayesian inference algorithm, and the phase change parameters θ in the multiphase flow transient simulation model are updated through Markov chain Monte Carlo sampling, including the gas-liquid interface tension correction coefficient and the bubble burst frequency parameter. The prior distribution is set to the normal distribution of the formation pressure N(μ(d),σ 2(d)), μ(d) is the historical formation pressure mean at the current well depth, and σ(d) is the standard deviation function that increases with well depth;

[0116] Using the updated target phase change parameter θ', combined with the casing diameter change curve and wellbore inclination data within the wellbore geometry constraints, the multiphase flow transient simulation model is run to generate transient pressure propagation simulation results that include pressure wave propagation velocity and attenuation characteristics.

[0117] The transient pressure propagation simulation results are compared with the pressure fluctuation data in the standardized multiphase flow parameters by spectrum analysis, and the energy distribution difference in the 1-50 Hz frequency band is calculated using the following formula:

[0118] δ=∫|S (sim) (f)-S (real) (f)|df;

[0119] Among them, S (sim) (f) represents the energy density of the simulated pressure fluctuation spectrum obtained by fast Fourier transform of the transient pressure propagation simulation results, in MPa 2 / Hz, its physical meaning is the pressure fluctuation energy distribution within unit bandwidth at frequency f; S (real) (f) represents the energy density of the spectrum of the measured pressure fluctuation data in the standardized multiphase flow parameters, which is calculated by windowing the raw data collected by the downhole high-frequency pressure sensor. δ is the cumulative energy difference between the simulated and measured spectra in the 1-50 Hz characteristic frequency band. This frequency band is selected based on the physical propagation characteristics of gas-liquid two-phase flow pressure fluctuations in deepwater drilling and effectively eliminates high-frequency noise interference above 50 Hz.

[0120] When δ exceeds the preset deviation threshold of 0.15, the update of the model weight coefficient matrix W is triggered:

[0121] W (new) =W (old) +β·(X T E);

[0122] Where β is the learning rate, X is the input feature matrix, which is constructed by the wellbore pressure gradient, fluid velocity and gas fraction data in the standardized multiphase flow parameters according to the grid position after time and space alignment. The matrix element x (ij) represents the normalized value of the jth characteristic parameter of the i-th grid; E is the simulation error vector, which is generated by decomposing the δ value into local error components according to the grid position. The vector dimension is consistent with the number of wellbore space grids m; W (old) W represents the weight coefficient matrix of the multiphase flow transient simulation model in the current iteration cycle, and the matrix dimension is m×n (m is the number of wellbore space grids, and n is the dimension of multiphase flow characteristic parameters); (new)It represents the target weight coefficient matrix after being updated by the dynamic parameter correction algorithm, and its update amount is determined by the linear combination of the input features and the error.

[0123] By iteratively updating until δ≤0.15, a transient simulation result matching the current drilling conditions is generated.

[0124] In an optional method, based on the transient simulation results, the hydraulic parameters to be optimized are determined, and according to the wellbore pressure constraints and the multi-objective optimization function, the hydraulic parameters are iteratively calculated using an adaptive particle swarm optimization algorithm to generate a real-time optimization solution including the optimal displacement, throttling pressure, and well killing parameters, including the following steps:

[0125] Extract the wellbore pressure gradient change rate, fluid velocity non-uniformity coefficient, and gas fraction risk threshold from the transient simulation results to determine the range of displacement, throttling pressure, and killing fluid density parameters to be optimized;

[0126] A multi-objective optimization function is constructed, including maximizing the mechanical drilling rate, minimizing the well killing cost, and minimizing the wellbore pressure fluctuation. The weight coefficient of each objective is dynamically adjusted according to the gas content risk threshold.

[0127] Initialize the particle swarm population, use the displacement, throttling pressure and killing fluid density parameters as particle position vectors, and set the particle velocity boundary based on the current wellbore pressure constraint;

[0128] During the iteration process, the inertia weight and learning factor are adaptively adjusted according to the difference between the global optimal solution and the individual historical optimal solution of the particle, and the particle positions that deviate from both the wellbore pressure constraint and the particle velocity boundary are updated first.

[0129] When the weighted variance of the multi-objective optimization function after a preset number of consecutive iterations is less than the preset variance threshold, the solution with the highest safety weight in the Pareto front is output as the real-time optimization solution.

[0130] In this embodiment, the wellbore pressure gradient change rate is extracted from the transient simulation results. in, is the change in vertical pressure gradient within the preset time window, and Δt is the length of the corresponding time window.

[0131] Calculate the fluid velocity non-uniformity coefficient η = σ(v) / μ(v); where σ(v) is the standard deviation of the flow velocity of each grid in the wellbore axis, and μ(v) is the average flow velocity.

[0132] According to the gas void fraction risk threshold λ, when the real-time gas void fraction φ ≥ 0.15, λ = φ / (1-φ), otherwise λ = 0.

[0133] Determine the displacement Q∈[Q based on the values ​​of γ, η, and λmin ,Q max ]、Throttling pressure P choke ∈[P wellhead +2MPa,P fracture -3MPa] and killing fluid density ρ mud ∈[1.2g / cm 3 ,ρ pore +0.3g / cm 3 ] parameter range, where Q min To maintain the minimum displacement for wellbore cleaning, Q max Calculated from the current pump power, P fracture is the formation fracture pressure, ρ pore is the formation pore fluid density.

[0134] Construct a multi-objective optimization function F = w1(1 / ROP) + w2C cost +w3σ(P), where ROP=K(Q / (D h 2 μ mud )) n is the mechanical drilling speed model, D h is the wellbore diameter, μ mud is the viscosity of drilling fluid, K and n are the formation drillability coefficients; C cost =αQt+βΔρ mud is the well killing cost function, α is the energy consumption coefficient per unit displacement, β is the killing fluid density adjustment cost coefficient; σ(P) is the standard deviation of wellbore pressure fluctuation; weight coefficients w1 = 1 - e^(-5λ), w2 = 0.3λ, and w3 = 1 - w1 - w2 are used to achieve dynamic coupling of gas void fraction risk to the objective function.

[0135] When initializing the particle swarm, the position vector X of each particle i =(Q i ,P choke_i ,ρ mud_i ), the velocity boundary is set to v max =0.2(Q max -Q min ); and subject to the constraint condition P choke ≥P wellhead +2MPa and ρ mud ≤ρ fracture -0.1g / cm 3 Among them, Q i Indicates the displacement parameter corresponding to the i-th particle, in cubic meters per second (m 3 / s), its physical meaning is the drilling fluid circulation rate, and its value range is determined by the pump power and the wellbore cleaning requirements; P choke_irepresents the throttling pressure parameter corresponding to the i-th particle, in megapascals (MPa), representing the control pressure of the wellhead throttle valve, and its lower limit is constrained by the wellhead pressure safety margin; ρ mud_i The density parameter of the well-killing fluid corresponding to the i-th particle is expressed in grams per cubic centimeter (g / cm 3 ), the upper limit of this parameter is directly related to the formation fracture pressure; Q max is the maximum allowable displacement, which is calculated based on the rated power of the drilling pump and the pressure bearing capacity of the pipeline; Q min is the minimum required displacement, determined based on the experimental data of wellbore diameter and cuttings transport efficiency; P wellhead is the wellhead pressure measured in real time; ρfracture is the formation fracture pressure corresponding to the equivalent density.

[0136] During the iteration process, the global optimal solution G is calculated. best and the individual historical optimal solution P best The difference is as follows:

[0137] θ = ||G_best - P_best|| / D;

[0138] Where D is the diameter of the solution space. When θ>0.5, the inertia weight ω decreases linearly from 0.9 to 0.4, the learning factor c1 decreases from 2.0 to 1.0, and c2 increases from 2.0 to 3.0.

[0139] Prioritize the choice of satisfying both |Q i -Q (opt) |>0.3(Q max -Q min ) and ρ mud_i >ρ pore +0.2g / cm 3 The position of the particles is updated.

[0140] When Σ(w i Var(F i ))<0.05, extract the Pareto frontier that satisfies the safety weight S=0.7(1-λ)+0.3(P fracture -P choke ) / P fracture The largest solution is used as the real-time optimization solution.

[0141] Figure 2 FIG. 2 shows a structural diagram of an embodiment of a data-driven hydraulic parameter real-time optimization system 200 provided by the present invention. Figure 2 As shown, the system 200 includes: an initialization module 210, a parameter updating module 220 and a solution generating module 230;

[0142] The initialization module 210 is used to: acquire multi-source heterogeneous data during the drilling process, and convert the multi-source heterogeneous data into standardized multiphase flow parameters with a unified time base through data cleaning and multi-modal spatiotemporal alignment fusion algorithm;

[0143] The parameter updating module 220 is used to: utilize a data-model joint driving mechanism to input standardized multiphase flow parameters into a pre-trained multiphase flow transient simulation model, and update the fluid density distribution and phase change parameters in the multiphase flow transient simulation model through a dynamic parameter correction algorithm to generate a transient simulation result that matches the current drilling conditions;

[0144] The solution generation module 230 is used to: determine the hydraulic parameters to be optimized based on the transient simulation results, and use the adaptive particle swarm optimization algorithm to iteratively calculate the hydraulic parameters according to the wellbore pressure constraints and the multi-objective optimization function to generate a real-time optimization solution including the optimal displacement, throttling pressure and well killing parameters.

[0145] In an optional manner, the initialization module 210 is specifically configured to:

[0146] Eliminate outliers and fill missing values ​​in measurement while drilling data, downhole sensor data, and geomechanics data from multi-source heterogeneous data to extract valid data segments;

[0147] Based on the drilling fluid circulation period and formation depth information, a dynamic time warping algorithm is used to align the timestamps of valid data segments to eliminate the transmission delay difference of multi-source heterogeneous data;

[0148] Based on the wellbore spatial coordinate system transformation model, downhole temperature, pressure, and flow rate data of different sampling frequencies are mapped to a unified spatial grid to generate a spatiotemporally continuous multimodal data matrix.

[0149] The multimodal data matrix is ​​reduced in dimensionality using a multi-scale feature fusion algorithm, and standardized multiphase flow parameters including wellbore pressure gradient, fluid velocity, gas fraction, and pressure fluctuation data are output.

[0150] In an optional manner, the parameter updating module 220 is specifically configured to:

[0151] The standardized multiphase flow parameters are input into the fluid density distribution correction algorithm, and the corrected target fluid density distribution is generated based on the correlation between the wellbore pressure gradient and the fluid flow rate;

[0152] Based on the gas fraction, the phase change parameters in the multiphase flow transient simulation model are updated using the Bayesian inference algorithm to obtain the target phase change parameters.

[0153] Using the target fluid density distribution and target phase change parameters, combined with wellbore geometry constraints, a multiphase flow transient simulation model is run to generate transient pressure propagation simulation results.

[0154] The transient pressure propagation simulation results are compared with the pressure fluctuation data in the standardized multiphase flow parameters. If the deviation exceeds the preset deviation threshold, the iterative update of the model weight coefficient is triggered until a transient simulation result matching the current drilling conditions is generated.

[0155] In an optional manner, the solution generation module 230 is specifically configured to:

[0156] Extract the wellbore pressure gradient change rate, fluid velocity non-uniformity coefficient, and gas fraction risk threshold from the transient simulation results to determine the range of displacement, throttling pressure, and killing fluid density parameters to be optimized;

[0157] A multi-objective optimization function is constructed, including maximizing the mechanical drilling rate, minimizing the well killing cost, and minimizing the wellbore pressure fluctuation. The weight coefficient of each objective is dynamically adjusted according to the gas content risk threshold.

[0158] Initialize the particle swarm population, use the displacement, throttling pressure and killing fluid density parameters as particle position vectors, and set the particle velocity boundary based on the current wellbore pressure constraint;

[0159] During the iteration process, the inertia weight and learning factor are adaptively adjusted according to the difference between the global optimal solution and the individual historical optimal solution of the particle, and the particle positions that deviate from both the wellbore pressure constraint and the particle velocity boundary are updated first.

[0160] When the weighted variance of the multi-objective optimization function after a preset number of consecutive iterations is less than the preset variance threshold, the solution with the highest safety weight in the Pareto front is output as the real-time optimization solution.

[0161] The technical solution of this embodiment realizes the spatiotemporal alignment of multi-source data and dynamic correction of the model, improves the transient simulation accuracy through a joint driving mechanism, and can generate the optimal hydraulic parameter solution that meets the wellbore safety constraints by combining an adaptive optimization algorithm.

[0162] The above-mentioned steps for implementing corresponding functions of each parameter and each module in the data-driven hydraulic parameter real-time optimization system 200 of this embodiment can refer to the various parameters and steps in the embodiment of the data-driven hydraulic parameter real-time optimization method above, and will not be repeated here.

[0163] like Figure 3 As shown, an electronic device 300 according to an embodiment of the present invention includes a processor 320 coupled to a memory 310. The memory 310 stores at least one computer program 330. The at least one computer program 330 is loaded and executed by the processor 320 to enable the electronic device 300 to implement any of the above-mentioned data-driven real-time optimization methods for hydraulic parameters, specifically:

[0164] The electronic device 300 may vary significantly due to different configurations or performances, and may include one or more processors 320 (Central Processing Units, CPUs) and one or more memories 310, wherein the one or more memories 310 store at least one computer program 330, which is loaded and executed by the one or more processors 320 to enable the electronic device 300 to implement any of the data-driven real-time optimization methods for hydraulic parameters provided in the above embodiments. Of course, the electronic device 300 may also have components such as a wired or wireless network interface, a keyboard, and an input / output interface for input and output. The electronic device 300 may also include other components for implementing device functions, which will not be described in detail here.

[0165] A computer-readable storage medium according to an embodiment of the present invention stores at least one computer program, which is loaded and executed by a processor to enable a computer to implement any of the above-mentioned data-driven real-time optimization methods for hydraulic parameters.

[0166] Alternatively, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc (CD-ROM), a magnetic tape, a floppy disk, an optical data storage device, or the like.

[0167] In an exemplary embodiment, a computer program product or computer program is also provided. The computer program product or computer program includes computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform any of the aforementioned data-driven real-time hydraulic parameter optimization methods.

[0168] It should be noted that the terms "first," "second," and the like in the specification and claims of this application are used to distinguish similar objects and to define a specific order or precedence. Where appropriate, the order used for similar objects may be interchanged, such that the embodiments of the present application described herein can be implemented in an order other than the order shown or described.

[0169] Those skilled in the art will appreciate that the present invention may be implemented as a system, method, or computer program product. Therefore, the present disclosure may be implemented in the following forms: entirely in hardware, entirely in software (including firmware, resident software, microcode, etc.), or in a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." Furthermore, in some embodiments, the present invention may be implemented in the form of a computer program product embodied in one or more computer-readable media containing computer-readable program code.

[0170] Any combination of one or more computer-readable media can be used. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device or device.

[0171] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A data-driven real-time optimization method for hydraulic parameters, characterized in that: include: Acquire multi-source heterogeneous data during the drilling process and convert the multi-source heterogeneous data into standardized multiphase flow parameters with a unified time base through data cleaning and multi-modal spatiotemporal alignment fusion algorithm; Using a data-model joint driving mechanism, the standardized multiphase flow parameters are input into a pre-trained multiphase flow transient simulation model, and the fluid density distribution and phase change parameters in the multiphase flow transient simulation model are updated through a dynamic parameter correction algorithm to generate transient simulation results that match the current drilling conditions; Based on the transient simulation results, the hydraulic parameters to be optimized are determined, and according to the wellbore pressure constraints and the multi-objective optimization function, the hydraulic parameters are iteratively calculated using an adaptive particle swarm optimization algorithm to generate a real-time optimization solution including the optimal displacement, throttling pressure, and well killing parameters.

2. The data-driven real-time optimization method for hydraulic parameters according to claim 1, characterized in that: The step of converting the multi-source heterogeneous data into standardized multiphase flow parameters with a unified time base through data cleaning and multimodal spatiotemporal alignment fusion algorithm includes: Eliminate outliers and fill in missing values ​​for measurement while drilling data, downhole sensor data, and geomechanics data in the multi-source heterogeneous data to extract valid data segments; According to the drilling fluid circulation period and formation depth information, a dynamic time warping algorithm is used to align the timestamps of the valid data segments to eliminate the transmission delay difference of the multi-source heterogeneous data; Based on the wellbore spatial coordinate system transformation model, downhole temperature, pressure, and flow rate data of different sampling frequencies are mapped to a unified spatial grid to generate a spatiotemporally continuous multimodal data matrix. The multimodal data matrix is ​​subjected to dimensionality reduction processing by a multi-scale feature fusion algorithm, and the standardized multiphase flow parameters including wellbore pressure gradient, fluid flow rate, gas content and pressure fluctuation data are output.

3. The data-driven real-time optimization method for hydraulic parameters according to claim 2, characterized in that: The steps of inputting the standardized multiphase flow parameters into a pre-trained multiphase flow transient simulation model by using a data-model joint driving mechanism, and updating the fluid density distribution and phase change parameters in the multiphase flow transient simulation model by using a dynamic parameter correction algorithm to generate a transient simulation result matching the current drilling conditions include: Inputting the standardized multiphase flow parameters into a fluid density distribution correction algorithm, and generating a corrected target fluid density distribution based on the correlation between the wellbore pressure gradient and the fluid flow rate; Based on the gas fraction, updating the phase change parameters in the multiphase flow transient simulation model by using a Bayesian inference algorithm to obtain target phase change parameters; Using the target fluid density distribution and the target phase change parameters, combined with wellbore geometric constraints, the multiphase flow transient simulation model is run to generate transient pressure propagation simulation results; The transient pressure propagation simulation results are compared with the pressure fluctuation data in the standardized multiphase flow parameters. If the deviation exceeds a preset deviation threshold, the iterative update of the model weight coefficient is triggered until the transient simulation results matching the current drilling conditions are generated.

4. The data-driven real-time optimization method for hydraulic parameters according to claim 3, characterized in that: The steps of determining hydraulic parameters to be optimized based on the transient simulation results, and iteratively calculating the hydraulic parameters using an adaptive particle swarm optimization algorithm according to wellbore pressure constraints and a multi-objective optimization function to generate a real-time optimization solution including optimal displacement, throttling pressure, and well killing parameters include: Extracting the wellbore pressure gradient change rate, fluid velocity non-uniformity coefficient, and gas content risk threshold from the transient simulation results, and determining the range of displacement, throttling pressure, and killing fluid density parameters to be optimized; Constructing the multi-objective optimization function including maximizing mechanical drilling rate, minimizing well killing cost, and minimizing wellbore pressure fluctuation; wherein the weight coefficient of each objective is dynamically adjusted according to the gas content risk threshold; Initializing the particle swarm population, using the displacement, throttling pressure and killing fluid density parameters as particle position vectors, and setting particle velocity boundaries based on the current wellbore pressure constraint; During the iteration process, the inertia weight and learning factor are adaptively adjusted according to the difference between the global optimal solution and the individual historical optimal solution of the particle, and the particle positions that deviate from the wellbore pressure constraint and the particle velocity boundary are updated first; When the weighted variance of the multi-objective optimization function after a preset number of consecutive iterations is less than a preset variance threshold, the solution with the highest safety weight in the Pareto front is output as the real-time optimization solution.

5. A data-driven real-time optimization system for hydraulic parameters, characterized in that: include: Initialization module, parameter update module and solution generation module; The initialization module is used to: acquire multi-source heterogeneous data during the drilling process, and convert the multi-source heterogeneous data into standardized multiphase flow parameters with a unified time reference through data cleaning and multi-modal spatiotemporal alignment fusion algorithm; The parameter updating module is configured to input the standardized multiphase flow parameters into a pre-trained multiphase flow transient simulation model using a data-model joint driving mechanism, and update the fluid density distribution and phase change parameters in the multiphase flow transient simulation model using a dynamic parameter correction algorithm to generate a transient simulation result that matches the current drilling condition; The solution generation module is used to: determine the hydraulic parameters to be optimized based on the transient simulation results, and iteratively calculate the hydraulic parameters using an adaptive particle swarm optimization algorithm according to the wellbore pressure constraints and the multi-objective optimization function to generate a real-time optimization solution including the optimal displacement, throttling pressure and well killing parameters.

6. The data-driven hydraulic parameter real-time optimization system according to claim 5, characterized in that: The initialization module is specifically used for: Eliminate outliers and fill in missing values ​​for measurement while drilling data, downhole sensor data, and geomechanics data in the multi-source heterogeneous data to extract valid data segments; According to the drilling fluid circulation period and formation depth information, a dynamic time warping algorithm is used to align the timestamps of the valid data segments to eliminate the transmission delay difference of the multi-source heterogeneous data; Based on the wellbore spatial coordinate system transformation model, downhole temperature, pressure, and flow rate data of different sampling frequencies are mapped to a unified spatial grid to generate a spatiotemporally continuous multimodal data matrix. The multimodal data matrix is ​​subjected to dimensionality reduction processing by a multi-scale feature fusion algorithm, and the standardized multiphase flow parameters including wellbore pressure gradient, fluid flow rate, gas content and pressure fluctuation data are output.

7. The data-driven hydraulic parameter real-time optimization system according to claim 6, characterized in that: The parameter updating module is specifically used for: Inputting the standardized multiphase flow parameters into a fluid density distribution correction algorithm, and generating a corrected target fluid density distribution based on the correlation between the wellbore pressure gradient and the fluid flow rate; Based on the gas fraction, updating the phase change parameters in the multiphase flow transient simulation model by using a Bayesian inference algorithm to obtain target phase change parameters; Using the target fluid density distribution and the target phase change parameters, combined with wellbore geometric constraints, the multiphase flow transient simulation model is run to generate transient pressure propagation simulation results; The transient pressure propagation simulation results are compared with the pressure fluctuation data in the standardized multiphase flow parameters. If the deviation exceeds a preset deviation threshold, the iterative update of the model weight coefficient is triggered until the transient simulation results matching the current drilling conditions are generated.

8. The data-driven hydraulic parameter real-time optimization system according to claim 7, characterized in that: The solution generation module is specifically used for: Extracting the wellbore pressure gradient change rate, fluid velocity non-uniformity coefficient, and gas content risk threshold from the transient simulation results, and determining the range of displacement, throttling pressure, and killing fluid density parameters to be optimized; Constructing the multi-objective optimization function including maximizing mechanical drilling rate, minimizing well killing cost, and minimizing wellbore pressure fluctuation; wherein the weight coefficient of each objective is dynamically adjusted according to the gas content risk threshold; Initializing the particle swarm population, using the displacement, throttling pressure and killing fluid density parameters as particle position vectors, and setting particle velocity boundaries based on the current wellbore pressure constraint; During the iteration process, the inertia weight and learning factor are adaptively adjusted according to the difference between the global optimal solution and the individual historical optimal solution of the particle, and the particle positions that deviate from the wellbore pressure constraint and the particle velocity boundary are updated first; When the weighted variance of the multi-objective optimization function after a preset number of consecutive iterations is less than a preset variance threshold, the solution with the highest safety weight in the Pareto front is output as the real-time optimization solution.

9. An electronic device, characterized in that: The electronic device includes a processor, which is coupled to a memory. The memory stores at least one computer program, and the at least one computer program is loaded and executed by the processor so that the electronic device implements the data-driven real-time optimization method for hydraulic parameters as described in any one of claims 1 to 4.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores at least one computer program, which is loaded and executed by a processor so that the computer-readable storage medium implements the data-driven real-time optimization method for hydraulic parameters according to any one of claims 1 to 4.

Images