A high-head residual pressure pipeline water turbine runner design method and system

Abstract

A design method and system for a high-head, residual pressure pipeline turbine runner, belonging to the field of runner design technology, addresses the problem that traditional axial-flow turbine runners are mainly suitable for lower heads, and under high head conditions, the cavitation performance of the runner cannot meet the requirements, causing unit vibration and seriously threatening the safe and stable operation of the unit. The steps include: constructing an initial runner model and determining the parameter setting range; selecting several geometric control points as optimization objects, determining the initial values, upper limits, and lower limits of each control point, with hydraulic efficiency and minimum blade pressure as optimization objectives; performing parameter sensitivity analysis on several geometric control points to screen key optimization control points; establishing a database using optimal Latin hypercube sampling, fitting a response surface model, fitting a surrogate model based on cross-validation, and constructing functional relationships; and using an evolutionary algorithm for intelligent optimization to obtain the optimal design scheme. This invention is applicable to the design scenario of high-head, residual pressure pipeline turbine runners.

Description

Technical Field

[0001] This invention belongs to the field of turbine runner design technology, specifically relating to a design method for a high-head residual pressure pipeline turbine runner. Background Technology

[0002] In recent years, my country's electricity demand has increased significantly. Under the advocacy of a green economy, traditional thermal power has been restricted, while the market size of the hydropower industry has been boosted. Large amounts of redundant pressure energy exist in pressurized pipelines such as cooling water pipelines in large chemical plants and urban water supply pipelines. This energy is typically depressurized using pressure reducing valves, but the lost energy cannot be reused, resulting in waste. Hydropower turbines can recover and utilize this residual pressure energy in the pipelines, making it a clean energy source. This approach is energy-saving, environmentally friendly, and increases revenue, while also aligning with national carbon emission reduction requirements and sustainable development strategies.

[0003] Based on their working principles, water turbines are mainly divided into two categories: impulse turbines and reaction turbines. The selection of the turbine type depends on the residual pressure and flow rate of the circulating water system. The selection principles are similar to those for small hydropower stations, but considering that the residual pressure equipment often replaces the pressure reducing valve, the turbine outlet must maintain sufficient pressure for the water delivery pipeline and to maintain the normal operating pressure of the nozzles. This means that the turbine outlet needs a higher positive pressure; therefore, the reaction turbine becomes a more suitable choice. Traditional axial-flow turbine runners are mainly suitable for lower heads. Under high head conditions, the cavitation performance of the runner cannot meet the requirements, causing vibration in the unit and seriously threatening its safe and stable operation. Therefore, it is necessary to design a turbine runner that can operate safely, stably, and efficiently in high-head residual pressure pipelines. Summary of the Invention

[0004] This invention provides a design method and system for a high-head residual pressure pipeline turbine runner, which aims to solve the problem that the runners of traditional axial flow turbines are mainly suitable for lower heads, and under high head conditions, the cavitation performance of the runners cannot meet the requirements, resulting in unit vibration and seriously threatening the safe and stable operation of the unit.

[0005] Firstly, the purpose of this invention is to provide a design method for a high-head residual pressure pipeline turbine runner, comprising the following steps: S1: Construct the initial model of the impeller and determine the flow field parameters and the range of impeller parameter settings; S2: Select several geometric control points as optimization objects, determine the initial value, upper limit and lower limit of the geometric control points of each control point, and take hydraulic efficiency and blade minimum pressure as optimization objectives; S3: Optimal Latin hypercube sampling is used to perform parameter sensitivity analysis on several geometric control points to screen out key optimized control points; S4: Based on the key optimization control points, a database is established using optimal Latin hypercube sampling, a response surface model is fitted, a surrogate model is fitted based on cross-validation, and a functional relationship is constructed. S5: Employs an evolutionary algorithm for intelligent optimization to obtain the optimal design scheme.

[0006] Furthermore, a preferred embodiment is provided: the flow field parameters include flow rate, and the impeller parameters include impeller model diameter and rotational speed.

[0007] Furthermore, a preferred embodiment is provided: in S2, several geometric control points include: upper crown profile control point, lower ring profile control point, blade leading edge and trailing edge related control points, surface streamline related control points, and blade shape control point at the upper crown.

[0008] Furthermore, a preferred solution is provided: in S2, The minimum pressure on the blade is the minimum pressure at any point on the blade. The formula for calculating water conservancy efficiency is expressed as follows: , in, Represents axial torque, unit: , Represents angular velocity, unit: , The density of water, in units of , This is the acceleration due to gravity, with units of 1. , For traffic, The unit is Water head, unit: .

[0009] Furthermore, a preferred solution is provided: In step S3, the parameter sensitivity analysis includes: First, all control points and their range of variation are set. Then, 50 points are automatically sampled within the range using hypercube sampling to generate a new wheel model. The new model is then meshed, and the mesh is solved in CFX. Based on the CFX solution results, the impact of the changes in each parameter on the target is analyzed, and the change values ​​are given.

[0010] Furthermore, a preferred solution is provided: In step S4, fitting the surrogate model based on cross-validation includes the following steps: The model is fitted using 70%–85% of the samples in the database, and the model is tested using the remaining samples in the database. The prediction error is calculated, and the operation is repeated until all samples are tested once and only once.

[0011] Furthermore, a preferred solution is provided: the evolutionary algorithm includes the following steps: S51: An initial population is randomly generated based on Latin hypercube sampling, and all individuals in the population come from the key optimization control points. S52: The fitness of each individual is calculated using the Pareto dominance method. A solution is considered non-dominated when both the minimum blade pressure and efficiency are higher than their initial values. S53: Perform crossover operations on individuals in the population using a simulated binary crossover method, with the crossover probability set to 0.8 and the crossover distribution parameter set to 20; S54: Perform mutation operations on individuals in the population with a mutation probability of 8%; S55: The offspring population competes with the parent population in a tournament-style ordering process, with 5 individuals participating in each round, ultimately forming a new parent breeding population containing 24 individuals. S56: Determine whether the new parent breeding population meets the optimization objective. If it does, the optimization ends and the optimal rotating scheme is obtained. If it does not meet the objective, return to S52 and repeat the iteration until the maximum number of iterations is reached.

[0012] Secondly, the purpose of this invention is to propose a high-head, residual pressure pipeline turbine runner design system, which is implemented based on a high-head, residual pressure pipeline turbine runner design method as described in any one or more of the above-mentioned schemes. The system includes: Runner modeling module: used to build the initial model of the runner, determine the flow field parameters and the range of runner parameter settings; Optimization control module: used to select several geometric control points as optimization objects, determine the initial value, upper limit and lower limit of the geometric control points of each control point, and take hydraulic efficiency and blade minimum pressure as optimization objectives; Sensitivity Analysis Module: Used to perform parameter sensitivity analysis on several geometric control points using optimal Latin hypercube sampling, and to screen out key optimized control points; Model fitting module: Based on the key optimization control points, it establishes a database using optimal Latin hypercube sampling, fits the response surface model, fits the surrogate model based on cross-validation, and constructs the functional relationship; Intelligent optimization module: Used to perform intelligent optimization using evolutionary algorithms to obtain the optimal design solution.

[0013] Thirdly, the present invention aims to provide a computer device, the computer device including a memory and a processor, the memory storing a computer program, and when the processor runs the computer program stored in the memory, the processor executes a high-head residual pressure pipeline turbine runner design method according to any one or more of the above-described schemes.

[0014] Fourthly, the present invention aims to provide a computer-readable storage medium for storing a computer program that executes the high-head residual pressure pipeline turbine runner design method described in any one or more of the above-described schemes.

[0015] Compared with the prior art, the advantages of the present invention are: The method proposed in this invention, through precise screening and optimization of the runner's geometric parameters, combined with sensitivity analysis of optimal Latin hypercube sampling, identifies key influencing parameters. Then, through surrogate model construction and improved genetic algorithm optimization, the minimum pressure of the blades is comprehensively improved. Especially in key areas prone to cavitation, such as the initial relative chord length, the pressure improvement effect is significant. This fundamentally reduces the cavitation risk under high head conditions, solves the core defects of traditional through-flow runners in high head environments, such as insufficient cavitation performance and frequent unit vibration, and ensures the long-term safe and stable operation of the equipment.

[0016] Secondly, in terms of hydraulic efficiency and operating range, the intelligently optimized impeller, compared with the manual adjustment scheme, not only significantly improves efficiency at the design flow rate, but also achieves a breakthrough in efficiency in operating conditions where the flow rate deviates significantly from the design flow rate. This significantly expands the impeller's efficient operating range, adapts to the actual operating conditions of frequent flow fluctuations in high-head residual pressure pipelines, improves the overall efficiency of residual pressure energy recovery, and creates more considerable energy-saving benefits for users.

[0017] This invention is applicable to the design of turbine runners in high-head, residual pressure pipeline systems. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating a high-head residual pressure pipeline turbine runner design method according to a specific embodiment of the present invention; Figure 2 This is a schematic diagram of a three-dimensional model of the optimized water turbine and runner according to a specific embodiment of the present invention; Figure 3 This is a schematic diagram of the cross-section of the turbine runner according to a specific embodiment of the present invention; Figure 4 This is a schematic diagram of the streamlined profile of the turbine runner blades according to a specific embodiment of the present invention; Figure 5This is a schematic diagram showing the distribution of the upper crown profile, lower ring profile, and key geometric control points of the blades of the turbine runner according to a specific embodiment of the present invention. Figure 6 This is a schematic diagram showing the distribution of control points corresponding to the streamlines on the surface of the turbine runner blades according to a specific embodiment of the present invention; Figure 7 This is a schematic diagram showing the distribution of control points corresponding to blade thickness according to a specific embodiment of the present invention; Figure 8 This is a statistical chart showing the degree of influence of the parameters at each control point on the hydraulic efficiency of the turbine, as described in a specific embodiment of the present invention. Figure 9 This is a statistical chart showing the influence of the control point parameters described in the specific embodiments of the present invention on the minimum pressure (cavitation performance) of the turbine runner blades. Figure 10 This is a schematic diagram comparing the prediction coefficients (CoP) of different proxy models for hydraulic efficiency according to a specific embodiment of the present invention. Figure 11 This is a schematic diagram comparing the prediction coefficients (CoP) of different proxy models regarding the minimum blade pressure in a specific embodiment of the present invention; Figure 12 This is a comparison chart of the error analysis between the fitting and prediction results of the hydraulic efficiency of the turbine described in a specific embodiment of the present invention and the actual simulation results. Figure 13 This is a comparison chart of the error analysis between the fitting and prediction results of the minimum pressure of the turbine blades described in a specific embodiment of the present invention and the actual simulation results. Figure 14 This is a schematic diagram of the response surface of hydraulic efficiency and key parameters based on the anisotropic Kriging model according to a specific embodiment of the present invention. Figure 15 This is a schematic diagram of the response surface of the minimum pressure and key parameters of the turbine runner blade based on the genetic aggregation function, as described in a specific embodiment of the present invention. Figure 16 This is a flowchart illustrating the intelligent optimization process of turbine runner parameters based on an improved genetic algorithm, as described in a specific embodiment of the present invention. Figure 17 This is a schematic diagram of the Pareto front solution set obtained during the optimization of turbine runner parameters according to a specific embodiment of the present invention; Figure 18 This is a comparison chart of the proportion of key geometric parameter ranges between the manual adjustment scheme and the intelligent optimization scheme described in a specific embodiment of the present invention; Figure 19 This is a comparison diagram of the pressure distribution along the relative chord length of the flow surface 0.1 of the turbine runner blades in a specific embodiment of the present invention; Figure 20 A comparison diagram of the pressure distribution along the relative chord length of the turbine runner blade flow surface 0.5 of the present invention, as described in a specific embodiment of the present invention; Figure 21 A comparison diagram of the pressure distribution along the relative chord length of the flow surface 0.9 of the turbine runner blades in a specific embodiment of the present invention; Figure 22 This is a statistical chart comparing the hydraulic efficiency of the manual adjustment scheme and the intelligent optimization scheme under different calculated flow rates, as described in a specific embodiment of the present invention. Detailed Implementation

[0020] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application can also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of this application with unnecessary detail.

[0021] The technical solutions in the embodiments of this application 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 application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] Many specific details are set forth in the following description in order to provide a full understanding of this application. However, this application may also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the spirit of this application. Therefore, this application is not limited to the specific embodiments disclosed below.

[0023] Implementation Method 1 This embodiment proposes a design method for a high-head, residual pressure pipeline turbine runner. It utilizes solver software to obtain flow field parameters at each design point, establishes a database, performs sensitivity analysis to identify significantly influential geometric parameters, resamples to establish a new database, fits a surrogate model to construct functional relationships, and uses a genetic algorithm to optimize the runner design. The flowchart of the method is shown below. Figure 1 As shown, it includes the following steps: S1: Construct an initial model of the impeller, determine the flow field parameters and the range of impeller parameter settings, wherein the flow field parameters include the flow rate and the impeller parameters include the impeller model diameter.

[0024] S2: Select several geometric control points as optimization objects, including: upper crown profile control points, lower ring profile control points, blade leading and trailing edge related control points, blade streamline related control points, and blade thickness control points; determine the initial value, upper limit, and lower limit of the geometric control points of each control point, with hydraulic efficiency and blade minimum pressure as optimization objectives, where the blade minimum pressure is the minimum pressure at any point on the blade; The formula for calculating water conservancy efficiency is expressed as follows: , in, Represents axial torque, unit: , Represents angular velocity, unit: , The density of water, in units of , This is the acceleration due to gravity, with units of 1. , For traffic, The unit is Water head, unit: .

[0025] S3: Optimal Latin hypercube sampling is used to perform parameter sensitivity analysis on several geometric control points to screen out key optimized control points; Specifically, the sensitivity analysis includes: first, setting all control points and their range of variation; then, automatically sampling 50 points within the range using hypercube sampling to generate a new wheel model; and finally, meshing the new model to generate a mesh, solving the problem in CFX, and analyzing the impact of changes in each parameter on the target based on the CFX solution results, and providing the change values.

[0026] The key control points are those that have a significant impact on both the target and a single target, and are expressed as follows: , The percentage of the impact on efficiency. To represent the percentage effect on minimum pressure, select... The maximum 12 parameter control points.

[0027] S4: Based on the key optimization control points, a database is established using optimal Latin hypercube sampling, a response surface model is fitted, a surrogate model is fitted based on cross-validation, and a functional relationship is constructed. Specifically, the fitting of the surrogate model is implemented based on the Optislang platform. The steps include: first, setting the selected key control points and their range of variation; Optislang automatically sampling several points within the range through hypercube sampling; and then transmitting the parameter data back to the model software Cfturbo to automatically generate a new rotary model. This data is then passed to the mesh generation software turbogrid to automatically generate the mesh. The model is solved in CFX. Optislang establishes a response surface model based on the CFX solution results. The Optislang platform provides various models such as: Polynomial Model, Moving Least Squares (MLS), Ordinary Kriging with isotropic and anisotropic kernel functions, Genetic Aggregation Response Surface (GARS), Support Vector Regression (SVR), Deep Feed Forward Network (DFFN), and Deep Infinite Mixture Gaussian Process (DFFN). The software (DIM-GP) automatically validates all models and provides the response surface model with the highest Cop as the final response model.

[0028] S5: Employs an evolutionary algorithm for intelligent optimization to obtain the optimal design scheme.

[0029] Implementation Method 2 This embodiment is a further illustrative example of the design method for a high-head residual pressure pipeline turbine runner described in Embodiment 1.

[0030] (1) Parametric construction of the wheel model Reservoir water supply is one of the main scenarios where high water head occurs in residual pressure pipelines. In this embodiment, the impeller is proposed to be used in the pipeline with a diameter of 1m and a rated speed of 600r / min. The daily water supply flow of the reservoir is about 600,000 tons, and the water head is above 100m.

[0031] A preliminary model of the impeller is constructed, with a diameter of 0.4 m. The proposed calculation speed is 430 r / min, and the calculated flow rate is 300 kg / s. Considering that the structural strength of the blades of the through-flow impeller is insufficient to support its operation under high head, although it is applied to a through-flow channel, the impeller blades will be designed as those of a mixed-flow impeller. The three-dimensional model of the impeller is as follows. Figure 2As shown, it is assembled onto a flow-through water intake mechanism. After the fluid enters the flow area, it first passes through the fixed guide vanes that provide circulation and support, and then through the movable guide vanes that control the flow rate, before reaching the impeller flow area. The impeller model has a shaft structure that extends backward from the discharge cone.

[0032] For the initial runner model, manual adjustments are made to the blade inlet and outlet angles, wrap angles, and blade profiles. This allows for the approximate determination of the runner's geometric parameter setting range, verifies the rationality of the design direction, and serves as the basis for subsequent intelligent optimization.

[0033] The preliminary optimized scheme includes the rotor meridian plane, blade rib lines, and blade shape as follows: Figure 3 , Figure 4 As shown.

[0034] (2) Determination of optimization parameters The parameter adjustment of the runner meridional plane mainly involves adjusting the upper crown curve profile, lower ring curve profile, and blade profile, while the parameter adjustment of the runner blade mainly involves the blade streamline and the profile curve of the airfoil. These curves are all based on the Bezier curve.

[0035] In this implementation method, 34 geometric control points were initially selected as optimization targets, and the final optimization objectives were hydraulic efficiency and the minimum blade pressure characterizing cavitation performance.

[0036] The upper and lower limits of each parameter are shown in Table 1, and the corresponding control points for each parameter are as follows: Figure 5 , Figure 6 , Figure 7 As shown, the runner meridional plane diagram is on the runner axial plane, the blade streamlines are based on conformal transformation, and the blade profile curve is based on the blade profile at the upper crown of the runner.

[0037] Table 1. Control points, initial values, and upper and lower limits for each parameter.

[0038] (3) Parameter sensitivity analysis More parameters can make the fitted model more accurate, but the required sample library is also larger. Considering the limitations of computing resources, in order to make the fit as reasonable as possible, it is necessary to initially screen out the geometric parameters with greater influence from the 35 geometric parameters. Therefore, sensitivity analysis is performed on these parameters.

[0039] The sampling method used was optimal Latin hypercube sampling, calculating 50 sample points. The effects of each parameter on efficiency and minimum pressure are as follows: Figure 8 , Figure 9 As shown. Taking all factors into consideration, 12 parameters were finally selected as optimization control points, namely Xs1, Ys2, Ls, TS, L1t, T5t, L5c, T1t, T1c, T5c, Xi1, and Yi1.

[0040] The parameter sensitivity analysis specifically includes the following steps: The parameter sensitivity analysis is implemented on the Optislang platform. First, all control points and their variation ranges are set. Optislang automatically samples 50 points within the range through hypercube sampling and transmits the parameter data back to the model software Cfturbo to automatically generate a new rotary model. It is then passed to the mesh generation software turbogrid to automatically generate the mesh and solves the problem in CFX. Based on the solution results of CFX, Optislang analyzes the impact of the changes in each parameter on the target and provides the change values.

[0041] Furthermore, the key optimization control points for screening are those that simultaneously have a significant impact on the objective or have a very large impact on a certain objective, expressed by the expression... , The percentage of the impact on efficiency. To represent the percentage effect on minimum pressure, select... The maximum 12 parameter control points.

[0042] (4) Proxy model construction Based on a large number of sample point parameters and simulation results, the surrogate model uses mathematical modeling methods to construct an intuitive functional relationship graph, providing a frontier solution set for subsequent optimization calculations. The number of training sample points needs to meet the theoretical lower limit of the surrogate model; even if the prediction results are only approximate, their accuracy is usually sufficient to meet the practical needs of most engineering designs.

[0043] Based on 12 final optimized control points (Xs1, Ys2, Ls, TS, L1t, T5t, L5c, T1t, T1c, T5c, Xi1, and Yi1), the optimal Latin hypercube sampling method was used to calculate a database of 182 sample points and fit the response surface model.

[0044] The optimization process is based on the Optislang platform. Several surrogate models are used to establish the relationship between control points and output variables. Sample points are used to test the accuracy of each model. The model evaluation coefficients use the prediction coefficient CoP proposed in Optislang; a higher prediction coefficient indicates higher model accuracy. When CoP is 1, the predictive model can be considered completely equivalent to the true model. Model building and error analysis are based on cross-validation. In this process, for the existing database, the majority of samples are used for fitting, while a small subset of samples is used to test the model. The prediction error for this small subset of samples is calculated, and this process continues until all sample points have been tested once and only once.

[0045] Furthermore, the fitting of the surrogate model is implemented based on the Optislang platform. The steps include: first, setting 12 control points after screening and their variation range; Optislang automatically samples 182 points within the range through hypercube sampling; and then transmitting the parameter data back to the model software Cfturbo to automatically generate a new rotary model. The data is then transmitted to the mesh generation software turbogrid to automatically generate a mesh. The model is solved in CFX. Based on the solution results of CFX, Optislang establishes a response surface model. The Optislang platform provides a variety of models, including: Polynomial Model, Moving Least Squares (MLS), Ordinary Kriging with isotropic and anisotropic kernel functions, Genetic Aggregation Response Surface (GARS), Support Vector Regression (SVR), Deep Feed Forward Network (DFFN), and Deep Infinite Mixture Gaussian Process (DIM-GP). The validation results of each model, such as Figure 10 —As shown in Figure 15.

[0046] (5) Intelligent algorithm optimization OptiSLang offers evolutionary algorithms, particle swarm optimization algorithms, stochastic design improvement algorithms, and covariance matrix adaptive algorithms. The algorithm generates an initial population based on Latin hypercube sampling, with individuals derived from previous sensitivity analyses. From this initial population, the required number of optimal design schemes are selected to breed the next generation.

[0047] This implementation employs an evolutionary algorithm, which mimics the natural biological evolutionary process of adaptation, selection, and mutation. OptiSLang uses a flexible combination of genetic algorithms (GA) and evolutionary strategies (ES), striking a balance between the global search of genetic algorithms and the local search-focused approach of evolutionary strategies. Evolutionary algorithms can be used for discrete and continuous design variables, as well as single-objective and multi-objective optimization tasks.

[0048] The initial population size is set to 12, with 24 individuals per generation. A maximum of 416 iterations are defined. Pareto dominance is used to evaluate individual fitness. To ensure the inheritance of desirable traits, each generation's offspring and their parents participate in a tournament-style selection process, with 5 individuals participating in each round. This results in a new breeding population of 24 individuals. The algorithm process is as follows: Figure 16 As shown. The crossover method uses simulated binary crossover with a crossover probability of 0.8 and a specified crossover distribution parameter of 20. Smaller values ​​allow for solutions that differ significantly from their parents, while larger values ​​restrict the generation of offspring to those near their parents. A solution is considered non-dominated when both the blade's minimum pressure and efficiency are higher than their initial values, and the mutation probability is set to 8%. The calculated Pareto front is shown below. Figure 17 As shown.

[0049] The manually adjusted rotor is used as Scheme A, and sample point 3296 on the Pareto front is selected as Scheme B for comparison, representing the intelligently optimized rotor. The distribution of parameters in different schemes within the parameter range is as follows. Figure 18 As shown.

[0050] After optimization, the blade pressure is increased, and the risk of cavitation is reduced. The pressure along the streamline of the blade is as follows: Figure 19 As shown in Figure 22, flow surfaces of 0.1, 0.5, and 0.9 were selected. After optimization, the blade pressure was improved in all aspects, especially the minimum pressure was significantly improved at the beginning of the relative chord length direction.

[0051] The two schemes were tested under different flow rates. The efficiency of both schemes was improved after intelligent optimization, especially in areas where the flow rate deviated significantly from the design flow rate. This means that the optimization greatly expanded the operating range of the rotor.

[0052] Implementation Method 3 This embodiment proposes a high-head, residual pressure pipeline turbine runner design system. The system is based on a high-head, residual pressure pipeline turbine runner design method as described in either Embodiment 1 or Embodiment 2. The system includes: Runner modeling module: used to build the initial model of the runner, determine the flow field parameters and the range of runner parameter settings; Optimization control module: used to select several geometric control points as optimization objects, determine the initial value, upper limit and lower limit of the geometric control points of each control point, and take hydraulic efficiency and blade minimum pressure as optimization objectives; Sensitivity Analysis Module: Used to perform parameter sensitivity analysis on several geometric control points using optimal Latin hypercube sampling, and to screen out key optimized control points; Model fitting module: Based on the key optimization control points, it establishes a database using optimal Latin hypercube sampling, fits the response surface model, fits the surrogate model based on cross-validation, and constructs the functional relationship; Intelligent optimization module: Used to perform intelligent optimization using evolutionary algorithms to obtain the optimal design solution.

[0053] It is understood that the present invention has been described through some embodiments, and those skilled in the art will recognize that various changes or equivalent substitutions can be made to these features and embodiments without departing from the spirit and scope of the invention. Furthermore, under the teachings of the present invention, these features and embodiments can be modified to adapt to specific situations and materials without departing from the spirit and scope of the invention. Therefore, the present invention is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the protection scope of the present invention.

Claims

1. A design method for a high-head residual pressure pipeline turbine runner, characterized in that, Includes the following steps: S1: Construct the initial model of the impeller and determine the flow field parameters and the range of impeller parameter settings; S2: Select several geometric control points as optimization objects, determine the initial value, upper limit and lower limit of the geometric control points of each control point, and take hydraulic efficiency and blade minimum pressure as optimization objectives; S3: Optimal Latin hypercube sampling is used to perform parameter sensitivity analysis on several geometric control points to screen out key optimized control points; S4: Based on the key optimization control points, a database is established using optimal Latin hypercube sampling, a response surface model is fitted, a surrogate model is fitted based on cross-validation, and a functional relationship is constructed. S5: Employs an evolutionary algorithm for intelligent optimization to obtain the optimal design scheme.

2. The design method for a high-head residual pressure pipeline turbine runner according to claim 1, characterized in that, The flow field parameters include flow rate, and the impeller parameters include impeller model diameter and rotational speed.

3. The design method for a high-head residual pressure pipeline turbine runner according to claim 1, characterized in that, In S2, several geometric control points include: upper crown contour control point, lower ring contour control point, blade leading and trailing edge related control points, surface streamline related control points, and blade shape control point at the upper crown.

4. The design method for a high-head residual pressure pipeline turbine runner according to claim 1, characterized in that, In S2, The minimum pressure on the blade is the minimum pressure at any point on the blade. The formula for calculating water conservancy efficiency is expressed as follows: ， in, Represents axial torque, unit: , Represents angular velocity, unit: , The density of water, in units of , This is the acceleration due to gravity, in units of 1. , For traffic, The unit is Water head, unit: .

5. The design method for a high-head residual pressure pipeline turbine runner according to claim 1, characterized in that, In S3, the parameter sensitivity analysis includes: First, all control points and their range of variation are set. Then, 50 points are automatically sampled within the range using hypercube sampling to generate a new wheel model. The new model is then meshed, and the mesh is solved in CFX. Based on the CFX solution results, the impact of the changes in each parameter on the target is analyzed, and the change values ​​are given.

6. The design method for a high-head residual pressure pipeline turbine runner according to claim 1, characterized in that, In step S4, fitting the surrogate model based on cross-validation includes the following steps: The model is fitted using 70%–85% of the samples in the database, and the model is tested using the remaining samples in the database. The prediction error is calculated, and the operation is repeated until all samples are tested once and only once.

7. The design method for a high-head residual pressure pipeline turbine runner according to claim 1, characterized in that, The evolutionary algorithm includes the following steps: S51: An initial population is randomly generated based on Latin hypercube sampling, and all individuals in the population come from the key optimization control points. S52: The fitness of each individual is calculated using the Pareto dominance method. A solution is considered non-dominated when both the minimum blade pressure and efficiency are higher than their initial values. S53: Perform crossover operations on individuals in the population using a simulated binary crossover method, with the crossover probability set to 0.8 and the crossover distribution parameter set to 20; S54: Perform mutation operations on individuals in the population with a mutation probability of 8%; S55: The offspring population competes with the parent population in a tournament-style ordering process, with 5 individuals participating in each round, ultimately forming a new parent breeding population containing 24 individuals. S56: Determine whether the new parent breeding population meets the optimization objective. If it does, the optimization ends and the optimal rotating scheme is obtained. If it does not meet the objective, return to S52 and repeat the iteration until the maximum number of iterations is reached.

8. A design system for a high-head residual pressure pipeline turbine runner, characterized in that, The system is implemented based on a high-head residual pressure pipeline turbine runner design method as described in any one of claims 1-7, and the system includes: Runner modeling module: used to build the initial model of the runner, determine the flow field parameters and the range of runner parameter settings; Optimization control module: used to select several geometric control points as optimization objects, determine the initial value, upper limit and lower limit of the geometric control points of each control point, and take hydraulic efficiency and blade minimum pressure as optimization objectives; Sensitivity Analysis Module: Used to perform parameter sensitivity analysis on several geometric control points using optimal Latin hypercube sampling, and to screen out key optimized control points; Model fitting module: Based on the key optimization control points, it establishes a database using optimal Latin hypercube sampling, fits the response surface model, fits the surrogate model based on cross-validation, and constructs the functional relationship; Intelligent optimization module: Used to perform intelligent optimization using evolutionary algorithms to obtain the optimal design solution.

9. A computer device, characterized in that, The computer device includes a memory and a processor. The memory stores a computer program. When the processor runs the computer program stored in the memory, the processor executes a high-head residual pressure pipeline turbine runner design method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store a computer program that executes a high-head residual pressure pipeline turbine runner design method according to any one of claims 1-7.

Images