A method for efficient optimization of a nozzle parameter of a downhole jet flushing tool

Abstract

The present application relates to a kind of wellbore jet flushing tool nozzle parameter high-efficiency optimization method, comprising the following steps: nozzle diameter, nozzle number, nozzle inclination are factor to formulate orthogonal experiment table, obtain annular jet flushing time data set in combination with fluid flushing simulation;The influence degree of each factor on flushing time is evaluated by range analysis, and the design input variable of subsequent prediction model is determined;Annular jet flushing time prediction model based on Kriging method is constructed and accuracy is verified;Combining improved particle swarm optimization algorithm, after multiple independent restart optimization, finally determine the global optimal process parameter combination.The present application quantifies the coupling influence of multiple factors by orthogonal experiment, determines the significant influencing factors by range analysis, realizes high-precision prediction by fitting nonlinear relationship using Kriging model, breaks through local optimum by combining improved particle swarm algorithm, determines the optimal process parameter combination of jet flushing tool, and provides theoretical basis for downhole annular cleaning tool optimization.

Description

Technical Field

[0001] This invention belongs to the field of petroleum drilling and production technology, specifically relating to an efficient optimization method for nozzle parameters of downhole jet flushing tools, applicable to parameter optimization and flushing time prediction of downhole annular jet cleaning tools. Technical Background

[0002] Downhole jet flushing is a crucial step in petroleum engineering to remove residual cement and rock from wellbores and ensure the quality of subsequent operations. Its core lies in using high-speed jets to impact and strip away the residue. The jet flushing tool, as the core equipment for achieving flushing, directly determines the cleaning efficiency, while the nozzle arrangement (including parameters such as diameter, number, and inclination angle) is a key factor affecting the jet coverage, impact force distribution, and cleaning effect. However, field experiments require setting up a realistic wellbore environment and repeatedly testing different nozzle parameter combinations, which is not only time-consuming and resource-intensive but also carries safety risks such as wellbore damage due to trial and error. Therefore, using fluid simulation technology to simulate the jet flushing process has become a key approach to efficiently explore the relationship between nozzle parameters and flushing effectiveness, replacing high-cost field experiments.

[0003] While high-fidelity fluid simulation can accurately simulate the scouring process of fluid flow on gravel particles, it suffers from significant bottlenecks such as high computational resource consumption and long processing time, making it difficult to directly support rapid evaluation and optimization of multi-parameter combinations. Regarding prediction and optimization methods, the Kriging model, as an interpolation method based on Gaussian processes, can accurately fit nonlinear relationships and provide quantification of prediction uncertainty, significantly reducing computational costs while maintaining accuracy. Therefore, this invention uses the Kriging model to construct a prediction model for annular jet flushing time. In terms of optimization algorithms, the standard particle swarm optimization (PSO) algorithm is prone to getting trapped in local optima and suffers from slow convergence speed. The improved PSO algorithm, by introducing adaptive inertia weights and dynamic learning factors, effectively balances global exploration and local exploitation capabilities, improving optimization efficiency and convergence accuracy in complex parameter spaces.

[0004] To address the bottlenecks in fluid simulation and the limitations of traditional prediction and optimization methods, this invention proposes a solution that integrates range analysis, the Kriging model, and an improved particle swarm optimization algorithm. First, a limited number of representative samples are obtained through orthogonal experiments. Then, range analysis is used to assess the influence of each factor to determine the modeling variables. Next, a high-precision Kriging prediction model is constructed. Finally, the improved particle swarm optimization algorithm is combined to achieve global parameter optimization, providing an effective approach for optimizing jet flushing tools and predicting flushing time. Summary of the Invention

[0005] In view of this, the purpose of this invention is to address the problems of difficulty in rapidly evaluating and optimizing parameter combinations in high-precision fluid simulation calculations, low fitting accuracy and large sample requirements of general surrogate models, and low optimization efficiency of standard particle swarm optimization algorithms. This invention provides a nozzle parameter optimization method for jet flushing tools that integrates orthogonal experiments, range analysis, Kriging prediction, and an improved particle swarm optimization algorithm, mainly solving the following two technical problems:

[0006] (1) To address the problems of high-precision fluid simulation computational resource consumption and long computation time, making it difficult to support rapid evaluation and optimization of multi-parameter combinations, as well as the low fitting accuracy and large sample requirements of general surrogate models in highly nonlinear problems, this invention obtains a limited number of representative samples through orthogonal experimental design, uses range analysis to evaluate the influence of each factor to determine the modeling variables, and then uses the Kriging model to construct a high-precision surrogate prediction model for annular jet flushing time. This significantly reduces computational costs while achieving accurate fitting and prediction of flushing time under the coupling effect of multiple factors, providing an efficient and reliable model foundation for subsequent rapid parameter optimization.

[0007] (2) To address the problem that the standard particle swarm optimization algorithm is prone to getting stuck in local optima and has a slow convergence speed in complex parameter spaces, this invention designs an improved particle swarm optimization algorithm. By introducing an adaptive inertia weight based on exponential decay and a dynamic adjustment mechanism for the learning factor, it maintains a strong global exploration capability in the early stage of optimization to expand the search range, strengthens the local exploration capability in the later stage of optimization to accelerate convergence, and combines a multiple independent restart strategy to retain historical optimal solutions, effectively improving the global optimization efficiency and convergence accuracy of the algorithm in the parameter space of the jet flushing tool nozzle.

[0008] To achieve the above objectives, the present invention provides the following technical solution:

[0009] 1. A method for efficiently optimizing nozzle parameters of a downhole jet flushing tool, characterized by comprising the following steps:

[0010] S1: Develop an orthogonal experimental table and obtain the annular jet flushing time dataset through fluid simulation;

[0011] S2: The influence of each factor on the flushing time was evaluated by range analysis, and it was verified that the nozzle diameter, number of nozzles and nozzle tilt angle all had a significant impact, and these were determined as all input variables for the subsequent prediction model.

[0012] S3: Construct a prediction model for annular jet flushing time based on the Kriging method;

[0013] S4: Design an improved particle swarm optimization algorithm, using the flushing time predicted by the Kriging model as the fitness function, minimizing the flushing time as the optimization objective, and using nozzle diameter, number of nozzles, and nozzle tilt angle as decision variables, setting the search range for each variable, and performing global optimization.

[0014] S5: From the optimal solutions obtained by multiple independent restarts and optimizations, select the solution with the smallest fitness function value as the global optimal nozzle parameter combination, and verify it through fluid simulation.

[0015] Furthermore, step S1 includes:

[0016] S11: The factors affecting the annular jet flushing time were selected as nozzle diameter (A), number of nozzles (B), and nozzle tilt angle (C), and the experiment was arranged using the L16(4³) orthogonal experimental table;

[0017] S12: Using ANSYS Fluent fluid simulation, simulate the time taken for the jet flushing tool to remove cement stone particles from the annulus until the cleaning rate reaches 95%, and obtain the flushing time data corresponding to each set of parameters to form a jet flushing time dataset.

[0018] Furthermore, step S2 specifically includes the following steps:

[0019] S21: Perform range analysis on the original experimental dataset to calculate the mean rinsing time for each factor at different levels. A larger range indicates a greater influence of that factor on rinsing time. The calculation formula is as follows:

[0020] ;

[0021] Among them, R A Let AA1, AA2, ..., AAm represent the average values ​​of all experimental results when factor A takes the 1st, 2nd, ..., mth levels, respectively.

[0022] S22: Sort the influence of nozzle diameter, number of nozzles, and nozzle tilt angle on jet flushing time by range from largest to smallest, and verify whether all three factors have a significant impact, so as to determine the input variables for the design of the subsequent Kriging prediction model.

[0023] Furthermore, step S3, which involves constructing annular jet flushing time prediction model based on the Kriging method, specifically includes:

[0024] S31: Using the factors identified as having a significant impact in the range analysis as input variables and rinsing time as the output variable, construct a Kriging surrogate model. The core expression is:

[0025] ;

[0026] Among them, f l (x) is a basis function, β l Let z(x) be the regression coefficient, and z(x) be a normal distribution N(0, σ). 2 A random process.

[0027] (a) The basis functions are in the form of second-order polynomials, containing constant terms, linear terms of each input variable, square terms, and pairwise interaction terms, and their expressions are as follows:

[0028] ;

[0029] (b) The relevant function is a Gaussian function, and its expression is:

[0030] ;

[0031] Where d is the dimension of the input variable, θ h The correlation parameter for the h-th input variable;

[0032] (c) Regression coefficient β l The vector is estimated using the weighted least squares method, and the calculation formula is as follows:

[0033] ;

[0034] Where F is the basis function matrix, R is the sample correlation matrix, and Y is the output response vector of the training samples.

[0035] S32: The model parameters are calibrated using the maximum likelihood estimation method, and the model accuracy is evaluated by leave-one-out cross-validation. The dataset is randomly divided into training and test sets at a ratio of 85% and 15% for model construction and validation.

[0036] Furthermore, step S4 specifically includes the following steps:

[0037] S41: Design an improved Particle Swarm Optimization (PSO) algorithm. Improvement strategies include:

[0038] (a) An exponentially decaying weight adjustment strategy is used to achieve a smooth transition from global exploration to local development. This maintains strong exploration capabilities in the early stages of optimization while focusing on refined local search in the later stages. The expression is as follows:

[0039] ;

[0040] Among them, w initial For the initial inertia weights, w final The final inertia weight is α, the attenuation coefficient is T. max The maximum number of iterations is t, and the current iteration number is t.

[0041] (b) Introducing a dynamic adjustment mechanism for learning factors, the dynamic adjustment of learning factors is achieved through a time-varying dynamic equilibrium equation, the mathematical expression of which is as follows:

[0042] ;

[0043] Among them, β and γ are adjustment parameters to ensure that c1 > c2 in the early stage of optimization to enhance exploration, and c2 > c1 in the later stage of optimization to accelerate convergence;

[0044] (c) Multiple independent restarts, each time with random initialization of the population and retention of historical best solutions, the mathematical expression of which is as follows:

[0045] ;

[0046] Using the predicted flushing time of the Kriging model as the fitness function and minimizing the flushing time as the objective, the search range of each factor was set, and the optimal solution was recorded after multiple restarts and optimizations.

[0047] Furthermore, step S5 specifically includes the following steps:

[0048] S51: From the optimal solutions obtained by multiple independent restarts and optimizations, select the solution with the smallest fitness function value (predicted flushing time) as the global optimal combination of process parameters;

[0049] S52: Substitute the obtained optimal parameter combination into the ANSYS Fluent fluid simulation model for verification, confirming that the actual flushing time matches the predicted value and meets the engineering requirements.

[0050] The beneficial effects of this invention are as follows: This invention proposes a solution that integrates range analysis, Kriging model and improved particle swarm optimization algorithm. By quantifying the influence of factors through limited samples, constructing a high-precision prediction model and achieving global parameter optimization, it provides an effective way to optimize jet flushing tools and predict flushing time, avoiding the problems of high resource consumption in high-precision fluid simulation calculations, low accuracy of flushing time prediction under multi-factor coupling and difficulty in parameter optimization.

[0051] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination, or may be learned from practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description

[0052] To make the objectives, technical solutions, and advantages of the present invention clearer, the preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings, wherein:

[0053] Figure 1This is a schematic diagram of the process of the present invention.

[0054] Figure 2 Two jet flushing tool models with different nozzle parameters are provided.

[0055] Figure 3 This is a schematic diagram of the volume mesh generation result containing the annular fluid domain.

[0056] Figure 4 A schematic diagram of the result of filling the annulus with crushed stone particles.

[0057] Figure 5 This represents the jet flushing time for each nozzle parameter combination.

[0058] Figure 6 This is a comparison chart of the actual and predicted values ​​of the Kriging prediction model on the training and test sets.

[0059] Figure 7 Convergence curves for improving the particle swarm optimization algorithm.

[0060] Figure 8 A three-dimensional model of the jet flushing tool with nozzles installed based on the global optimal solution and a particle distribution diagram at the beginning and end of the jet flushing process. Detailed Implementation

[0061] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0062] The accompanying drawings are for illustrative purposes only and are schematic diagrams, not actual pictures. They should not be construed as limiting the invention. To better illustrate the embodiments of the invention, some parts in the drawings may be omitted, enlarged, or reduced, and do not represent the actual product dimensions. It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings.

[0063] In the accompanying drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components. In the description of the present invention, it should be understood that if terms such as "upper," "lower," "left," "right," "front," and "rear" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the terms used to describe positional relationships in the drawings are only for illustrative purposes and should not be construed as limiting the present invention. For those skilled in the art, the specific meaning of the above terms can be understood according to the specific circumstances.

[0064] To address the challenges and shortcomings of existing methods in predicting annular jet flushing time, this invention provides a method for optimizing nozzle parameters of jet flushing tools based on the Kriging model. The technical solutions in the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0065] like Figure 1 As shown, this invention provides a method for optimizing nozzle parameters of a jet flushing tool based on the Kriging model, comprising the following steps:

[0066] (1) An orthogonal experimental table was developed, and the annular jet flushing time dataset was obtained through fluid simulation, as follows:

[0067] (1.1) The factors affecting the annular jet flushing time were selected as nozzle diameter (A), number of nozzles (B), and nozzle tilt angle (C). The nozzle diameter was categorized into four levels: 6, 8, 9, and 10 mm. The number of nozzles was categorized into four levels: 8, 12, 16, and 20. The nozzle tilt angle was further categorized into four levels based on the average tilt angle and the range of angle variation: small angle dominant (average tilt angle < 50°), medium angle (50° ≤ average tilt angle ≤ 70°), large angle dominant (average tilt angle > 70°), and full-range coverage (angle variation range > 40°). An orthogonal experimental setup was used to arrange the experiments. The orthogonal experimental setup for the jet flushing tool process parameters is as follows:

[0068] Table 1. Orthogonal Experiment Table of Process Parameters for Jet Flushing Tools

[0069]

[0070] (1.2) Fluid simulation calculations were performed using ANSYS Fluent, with the same model for the main body, but the nozzle diameter and arrangement were changed. Figure 2The image shows two sets of three-dimensional models of jet flushing tools with different nozzle parameters. Figure 3 The image shows the volume mesh generation results of a fluid simulation model containing an annular fluid domain, with surface meshes of a maximum size of 7 mm and a minimum size of 0.3 mm, and using the Poly-hexcore mesh generation method. Figure 4 The image shows the result of filling the annulus with gravel particles through discrete phase injection with a diameter of 5 mm. The total computation time for a single experiment on a workstation equipped with an i9-13900K processor exceeded 80 hours, demonstrating that fluid simulation calculations are computationally intensive and time-consuming.

[0071] (1.3) Record the total number of discrete phase particles before the start of the jet flushing process, select the time taken to remove the number of discrete phase particles of the annular cement stone crushed stone from the jet flushing tool to 95%, and record it as the flushing time data corresponding to each set of nozzle parameters to form the original experimental dataset containing 16 sets of samples. Figure 5 This represents the jet flushing time for each nozzle parameter combination.

[0072] (2) The influence of each factor on the rinsing time was obtained through range analysis, and the design input variables of the subsequent prediction model were determined as follows:

[0073] (2.1) Perform range analysis on the original experimental dataset to calculate the mean rinsing time for each factor at different levels. The larger the range, the greater the influence of the factor on the rinsing time. The calculation formula is as follows:

[0074] ;

[0075] Among them, R A Let AA1, AA2, ..., AAm represent the average values ​​of all experimental results when factor A takes the 1st, 2nd, ..., mth levels, respectively.

[0076] (2.2) The influence of nozzle diameter, number of nozzles, and nozzle tilt angle on jet flushing time were sorted from largest to smallest according to the range. The results of the range analysis are shown in Table 2:

[0077] Table 2 Range Analysis Results

[0078]

[0079] Table 2 shows that the range of the number of nozzles is the largest, followed by the horizontal tilt angle, while the range of the nozzle diameter is relatively small. However, the ranges of all three factors are significantly greater than zero, indicating that the nozzle diameter, number of nozzles, and nozzle tilt angle all have a non-negligible impact on the annular jet flushing time. Therefore, these three factors are all used as input variables in the design of the subsequent Kriging prediction model to ensure that the model can fully capture the flushing time variation under the coupled effects of multiple factors.

[0080] (3) Construct an annular jet flushing time prediction model based on the Kriging method, specifically including:

[0081] (3.1): Using the factors with significant influence identified in the range analysis as input variables and rinsing time as the output variable, a Kriging surrogate model is constructed. The core expression is:

[0082] ;

[0083] Among them, f l (x) is a basis function, β l Let z(x) be the regression coefficient, and z(x) be a normal distribution N(0, σ). 2 A random process.

[0084] (a) The basis functions are in second-order polynomial form, and their expressions are as follows:

[0085] ;

[0086] Where x1, x2, and x3 represent the nozzle diameter, the number of nozzles, and the horizontal nozzle tilt angle, respectively. This basis function form can effectively capture the nonlinear coupling effect between the various nozzle parameters.

[0087] (b) The relevant function is a Gaussian function, and its expression is:

[0088] ;

[0089] Where d=3 is the dimension of the input variable, θ h The correlation parameter for the h-th input variable;

[0090] (c) Regression coefficient β l The vector is estimated using the weighted least squares method, and the calculation formula is as follows:

[0091] ;

[0092] Where F is the basis function matrix, R is the sample correlation matrix, Y is the washing time vector of the training samples, and the correlation parameter θ h With the variance σ of the random process 2 Simultaneous optimization is performed using maximum likelihood estimation.

[0093] (3.2): The model parameters were calibrated using the maximum likelihood estimation method, and the accuracy was evaluated using leave-one-out cross-validation. After removing the outlier group 4, 15 groups were added to the original model group, totaling 16 groups. These groups were then divided into 16 groups at a ratio of 0.85 to construct the model. Figure 6 Table 3 shows the performance of different surrogate models on a random test set, comparing the actual and predicted values ​​of the established Kriging prediction model on the training and test sets.

[0094] Table 3. Performance of different proxy models on the random test set

[0095]

[0096] As shown in Table 3, the Kriging model has the smallest prediction error, significantly outperforming support vector regression and random forest models. This indicates that under conditions of small sample size and strong nonlinearity, the Kriging model has higher prediction accuracy and generalization ability, accurately capturing the complex mapping relationship between jet flushing time and nozzle parameters, providing a reliable surrogate model basis for subsequent optimization.

[0097] Furthermore, step S4 specifically includes the following steps:

[0098] (4) Design an improved particle swarm optimization (PSO) algorithm, specifically including:

[0099] (4.1) An exponentially decaying weight adjustment strategy is used to achieve a smooth transition from global exploration to local development. This maintains strong exploration capabilities in the early stages of optimization while focusing on refined local search in the later stages. The expression is as follows:

[0100] ;

[0101] Among them, w initial =0.9 is the initial inertia weight, w final =0.3 is the final inertia weight, α=0.006 is the attenuation coefficient, T max =100 is the maximum number of iterations;

[0102] A dynamic adjustment mechanism for the learning factor is introduced, which achieves dynamic adjustment of the learning factor through a time-varying dynamic equilibrium equation. Its mathematical expression is as follows:

[0103] ;

[0104] Among them, β and γ are adjustment parameters to ensure that c1 > c2 in the early stage of optimization to enhance exploration, and c2 > c1 in the later stage of optimization to accelerate convergence;

[0105] Multiple independent restarts, with random initialization of the population each time and retention of historical optimal solutions, can be mathematically expressed as follows:

[0106] ;

[0107] (4.2) Using the predicted flushing time of the Kriging model as the fitness function and minimizing the flushing time as the objective, the parameter search range was set as nozzle diameter 6-10mm, number 8-20, and tilt angle horizontal 1-4. The number of optimization restarts was set to 5, and the optimal solution was recorded each time. Figure 7 The convergence curves of the improved particle swarm optimization algorithm are shown in Table 4, with each restart and the corresponding optimal fitness as shown in Table 5. The correspondence between the optimal solution and the configuration is shown in Table 5.

[0108] Table 4. Number of restarts and corresponding optimal fitness

[0109]

[0110] Table 5. Correspondence between Optimal Solution and Configuration

[0111]

[0112] Furthermore, step S5 specifically includes the following steps:

[0113] (5.1) Select the solution with the best fitness (2.4200S) from the optimal solutions after 5 restarts, and determine the globally optimal combination of process parameters as shown in Table 6:

[0114] Table 6 Optimal Nozzle Parameters for Jet Flushing Tools

[0115]

[0116] (5.2) Substitute the optimal parameter combination into the fluid simulation model for verification. Figure 8 The model of the jet flushing tool with optimal nozzle parameters and the initial and final discrete phase gravel particle distribution of the jet flushing were obtained. The actual flushing time was 2.42 s, which is highly consistent with the predicted optimal fitness, confirming the accuracy and engineering feasibility of the optimization results. This embodiment demonstrates that the method of the present invention can efficiently and accurately determine the nozzle parameter combination of the jet flushing tool, maximizing the jet flushing efficiency.

[0117] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for efficiently optimizing nozzle parameters of a downhole jet flushing tool, characterized in that: Includes the following steps: S1: Develop an orthogonal experimental table and obtain the annular jet flushing time dataset through fluid simulation; S2: The influence of each factor on the flushing time was evaluated by range analysis, and it was verified that the nozzle diameter, number of nozzles and nozzle tilt angle all had a significant impact, and these were determined as all input variables for the subsequent prediction model; S3: Construct a prediction model for annular jet flushing time based on the Kriging method; S4: Design an improved particle swarm optimization algorithm, using the flushing time predicted by the Kriging model as the fitness function, minimizing the flushing time as the optimization objective, and using nozzle diameter, number of nozzles, and nozzle tilt angle as decision variables, setting the search range for each variable, and performing global optimization. S5: From the optimal solutions obtained by multiple independent restarts and optimizations, select the solution with the smallest fitness function value as the global optimal nozzle parameter combination, and verify it through fluid simulation.

2. The method for efficient optimization of nozzle parameters of downhole jet flushing tools according to claim 1, characterized in that, Step S1 specifically includes: S11: The factors affecting the annular jet flushing time were selected as nozzle diameter (A), number of nozzles (B), and nozzle tilt angle (C), and the experiment was arranged using the L16(4³) orthogonal experimental table; S12: Using ANSYS Fluent fluid simulation, simulate the time taken for the jet flushing tool to remove cement stone particles from the annulus until the cleaning rate reaches 95%, and obtain the flushing time data corresponding to each set of parameters to form a jet flushing time dataset.

3. The method for efficient optimization of nozzle parameters of downhole jet flushing tools according to claim 1, characterized in that: Step S2 specifically includes: S21: Perform range analysis on the original experimental dataset to calculate the mean rinsing time for each factor at different levels. A larger range indicates a greater influence of that factor on rinsing time. The calculation formula is as follows: ； Among them, R A For the range of factor A, A A1 A A2 , ...A Am These represent the average values ​​of all experimental results when factor A is at levels 1, 2, ..., m, respectively. S22: Sort the influence of nozzle diameter, number of nozzles, and nozzle tilt angle on jet flushing time by range from largest to smallest, verify whether the three factors have a significant impact, and determine the input variables for the design of the subsequent Kriging prediction model.

4. The method for efficient optimization of nozzle parameters of downhole jet flushing tools according to claim 1, characterized in that: Step S3, which involves constructing annular jet flushing time prediction model based on the Kriging method, specifically includes: S31: Using the factors identified as having a significant impact in the range analysis as input variables and rinsing time as the output variable, construct a Kriging surrogate model. The core expression is: ； Among them, f l (x) is a basis function, β l Let z(x) be the regression coefficient, and z(x) be a normal distribution N(0, σ). 2 A random process; (a) The basis functions are in the form of second-order polynomials, containing constant terms, linear terms of each input variable, square terms, and pairwise interaction terms, and their expressions are as follows: ； (b) The relevant function is a Gaussian function, and its expression is: ； Where d is the dimension of the input variable, θ h The correlation parameter for the h-th input variable; (c) Regression coefficient β l The vector is estimated using the weighted least squares method, and the calculation formula is as follows: ； Where F is the basis function matrix, R is the sample correlation matrix, and Y is the output response vector of the training sample; S32: The model parameters are calibrated using the maximum likelihood estimation method, and the model accuracy is evaluated by leave-one-out cross-validation. The dataset is randomly divided into training and test sets at a ratio of 85% and 15% for model construction and validation.

5. The method for efficient optimization of nozzle parameters of downhole jet flushing tools according to claim 1, characterized in that: Step S4 specifically includes: S41: Design an improved Particle Swarm Optimization (PSO) algorithm. Improvement strategies include: (a) An exponentially decaying weight adjustment strategy is used to achieve a smooth transition from global exploration to local development. This maintains strong exploration capabilities in the early stages of optimization while focusing on refined local search in the later stages. The expression is as follows: ； Among them, w initial For the initial inertia weights, w final The final inertia weight is α, the attenuation coefficient is T. max The maximum number of iterations is t, and the current iteration number is t. (b) Introducing a dynamic adjustment mechanism for learning factors, the dynamic adjustment of learning factors is achieved through a time-varying dynamic equilibrium equation, the mathematical expression of which is as follows: ； Among them, β and γ are adjustment parameters to ensure that c1 > c2 in the early stage of optimization to enhance exploration, and c2 > c1 in the later stage of optimization to accelerate convergence; (c) Multiple independent restarts, each time with random initialization of the population and retention of historical best solutions, the mathematical expression of which is as follows: ； S42: Using the predicted flushing time of the Kriging model as the fitness function and minimizing the flushing time as the objective, the search range of each factor is set, and the optimal solution is recorded after multiple restarts for optimization.

6. The method for efficiently optimizing nozzle parameters of downhole jet flushing tools according to claim 1, characterized in that: Step S5, which involves selecting and verifying the optimal solution, specifically includes: S51: From the optimal solutions obtained by multiple independent restarts and optimizations, select the solution with the smallest fitness function value (predicted flushing time) as the global optimal combination of process parameters; S52: Substitute the obtained optimal parameter combination into the ANSYS Fluent fluid simulation model for verification, confirming that the actual flushing time matches the predicted value and meets the engineering requirements.

Images