A method for collaborative optimization of multi-parameter coupling experiment and intelligent model of vertical axis tidal current energy turbine

Abstract

The present application relates to a kind of vertical axis tidal current energy turbine multi-parameter coupling experiment and intelligent model collaborative optimization method, the method includes: based on orthogonal experimental design and response surface method to construct the multi-level interaction working condition of blade key geometric parameters, obtain power and thrust coefficient performance curve, introduce global optimization of genetic algorithm and obtain optimal geometric configuration;Using variable pitch mechanism, determine the optimal variable pitch strategy considering power improvement and load fluctuation suppression;Introduce turbulence intensity correction coefficient, represent the performance degradation law of turbine under complex inflow, construct robust prediction model;Phase-locked PIV flow field measurement is executed under different working conditions, establish the quantitative correlation model of flow field characteristics and instantaneous load fluctuation, and feedback optimal parameter combination back to experimental design.Relative to the prior art, the present application has disclosed the multi-parameter nonlinear coupling mechanism of turbine energy capture efficiency and operation stability, provides physical basis for its optimization design and the like advantages.

Description

Technical Field

[0001] This invention relates to the field of vertical axis tidal current turbines, and in particular to a method for the collaborative optimization of multi-parameter coupling experiments and intelligent models for vertical axis tidal current turbines. Background Technology

[0002] Existing research both domestically and internationally has mainly focused on the following three directions, resulting in a certain level of technological accumulation, but also exhibiting significant limitations: Single-factor optimization of airfoil and geometric parameters: This study investigates the effect of single parameters such as airfoil section, aspect ratio, and cable-wheel ratio on the turbine power coefficient C using CFD (Computational Fluid Dynamics) numerical simulations and flume experiments. P To understand the influence patterns, a relatively comprehensive single-factor performance database was established.

[0003] Pitch strategy research: Some studies have explored the improvement of VATT performance by active or passive pitch control, and verified the effectiveness of pitch strategies in power enhancement and load suppression.

[0004] Wake and Turbulence Influence: In recent years, there has been an increasing number of studies on the wake characteristics of VATT arrays and the impact of incoming flow turbulence on performance, revealing systematic differences between real marine environments and uniform incoming flow conditions in laboratories.

[0005] However, existing technologies still have some problems: Existing research has consistently treated geometric parameter optimization and pitch strategy research as separate processes, never integrating them into a unified, collaborative optimization framework. The fundamental reason for this is that the dynamic stall characteristics of blades differ significantly under different geometric configurations (combinations of airfoil, aspect ratio, and cable-wheel ratio). An optimal pitch strategy derived for one airfoil may be ineffective or even harmful for another. Therefore, conclusions drawn from studying pitch strategies independently of geometric parameters have serious limitations and cannot achieve global optimization.

[0006] Single-factor optimization neglects the nonlinear coupling effect of multiple parameters. Existing research uses the method of controlling variables to change individual geometric parameters one by one and simply superimposes the single-factor optimal values ​​of each parameter. However, there is a strong nonlinear interaction effect among airfoil section, aspect ratio, and cable-wheel ratio. The single-factor optimization method cannot fundamentally capture this coupling effect, resulting in the obtained parameter combination being only a local optimum. Under multiple operating conditions, the system performance is unstable and there are irreparable system errors.

[0007] The lack of robustness studies under turbulent conditions is a significant problem. Almost all existing studies have been conducted under low-turbulence, uniform inflow conditions, while real ocean currents exhibit significant variations in turbulence intensity (TI) and turbulence scale. Current optimization schemes lack quantitative models of performance degradation under turbulent inflow conditions, leading to substantial performance declines in actual marine environments and significant systematic biases between engineering predictions and measurements. This is because turbulent inflows significantly affect blade dynamic stall characteristics and wake structure, and exhibit complex interactions with geometric parameters and pitch control strategies, which have never been systematically addressed in existing research.

[0008] The microscopic mechanism of the flow field is unclear, and optimization lacks a physical basis. Existing research mainly focuses on optimization based on macroscopic performance indicators. It lacks high-resolution quantitative flow field measurements of dynamic stall vortex evolution and wake vortex dissipation structure under multi-parameter coupling conditions. It cannot reveal from a physical essence why a certain combination of parameters performs better, making the optimization process more dependent on empirical trial and error rather than mechanistic laws, and lacking the ability to predict and extrapolate under new operating conditions. Summary of the Invention

[0009] The purpose of this invention is to propose a multi-parameter coupling experiment and intelligent model collaborative optimization method for vertical axis tidal power turbines, which solves the problem in the prior art that the turbine efficiency is lower than the theoretical value in actual multi-condition marine environments due to the complex nonlinear coupling between the turbine's geometric parameters and operating parameters. This method aims to improve the energy harvesting efficiency and operational stability of VATT in complex marine environments and provide a standardized optimization method that can be directly applied to engineering design.

[0010] The objective of this invention can be achieved through the following technical solutions: A method for co-optimization of multi-parameter coupling experiments and intelligent models for vertical axis tidal current turbines, comprising the following steps: Geometric parameter optimization involves constructing an experimental system model and defining various levels and interaction conditions for the geometric parameters, and obtaining the instantaneous power coefficient C. P and thrust coefficient C T The performance curves are used to establish a multidimensional response surface model of the fluctuation of the maximum power coefficient and the minimum thrust coefficient to the geometric parameters. The global optimization is carried out with the dual objectives of maximizing the instantaneous power coefficient and minimizing the fluctuation of the thrust coefficient to determine the optimal combination of geometric parameters and obtain the optimal geometric configuration. Optimize operating parameters by setting up active and passive pitch mechanisms, installing them in the optimal geometric configuration obtained in the previous step, obtaining and adopting different pitch strategies, and selecting and executing the optimal pitch strategy, i.e., the optimal operating parameters. Robustness correction and model optimization were performed by measuring the instantaneous power coefficient C under optimal geometry and optimal pitch strategy using an experimental system model. P and thrust coefficient C TThe attenuation characteristics were investigated, and a quantitative prediction model of the impact of turbulent inflow on performance was established. A robustness correction factor was obtained, and the experimental system model was optimized based on the robustness correction factor. Intelligent model construction involves high spatiotemporal resolution flow field measurements under different flow field conditions using multi-parameter combinations of optimal geometric configuration and optimal pitch strategy to obtain PIV data. Dynamic mode decomposition (DMD) or intrinsic orthogonal decomposition (POD) is used to analyze the PIV data. The analyzed flow field characteristics are then used to construct an intelligent correlation model with turbine load fluctuations through deep learning clustering algorithms.

[0011] The experiment and model were optimized in tandem. Based on the multi-parameter coupled experimental data and flow field analysis of the turbine under different flow field conditions, the intelligent correlation model was trained and subjected to physical constraints to obtain a quantitative correlation model with accurate prediction. Subsequently, the optimal combination of turbine geometry and operating parameters can be predicted through this model, and this parameter combination can be fed back into the experimental design in the first step. Finally, the results of the multi-parameter coupled experiment and intelligent model were optimized in tandem.

[0012] Furthermore, the experimental system model includes a modular VATT scaled-down experimental model, a dynamic load measurement system, and a reflux water tank experimental platform.

[0013] Furthermore, the geometric parameters include the blade airfoil section, aspect ratio AR, and cable-wheel ratio SC.

[0014] Furthermore, obtain the instantaneous power coefficient C. P and thrust coefficient C T The specific steps for creating the performance curve are as follows: For each set of geometric parameters, a systematic scan is performed on the entire tip speed ratio (TSR) range to obtain the instantaneous power coefficient C. P and thrust coefficient C T The performance curve.

[0015] Furthermore, the active pitch mechanism achieves active control by adjusting the pitch angle β in real time according to the blade azimuth angle θ through an electric drive unit and a control system.

[0016] Furthermore, the passive pitch mechanism achieves adaptive passive pitch by utilizing the self-balancing relationship between the elastic restoring torque generated by the elastic hinge and the hydrodynamic torque on the blade.

[0017] Furthermore, pitch control strategies include fixed pitch, active pitch, and passive adaptive pitch.

[0018] Furthermore, the instantaneous power coefficient C under optimal geometric configuration and optimal pitch strategy was measured using an experimental system model. P and thrust coefficient C T The specific steps for attenuation characteristics are as follows: By installing a passive turbulence grid in the return flume, controllable incoming turbulence with different turbulence intensities TI and integral scales Λ was generated. The suppression effect of the micro-flow field on dynamic stall and wake dissipation under optimal geometry and optimal pitch control strategy was measured. This study elucidates the impact of high-intensity and high-scale incoming flows on the turbine's unsteady attached flow and wake structure, as well as its effect on the turbine's instantaneous hydrodynamic power coefficient C. P and thrust coefficient C T Robust attenuation effect.

[0019] Furthermore, during the process of measuring the flow field with high spatiotemporal resolution for the optimal geometric configuration and optimal pitch strategy under typical operating conditions, phase-locked particle image velocimetry (PIV) technology is adopted. The measurement area covers the boundary layer around the blade and the near-wake region. The PIV system is triggered synchronously by the turbine angle and encoder signal phase to stably acquire flow field images as PIV data at a preset azimuth angle.

[0020] Furthermore, the specific steps for analyzing PIV data and obtaining the intelligent association model by combining Dynamic Mode Decomposition (DMD) or Intrinsic Orthogonal Decomposition (POD) with deep learning clustering algorithms are as follows: Dynamic mode decomposition (DMD) or intrinsic orthogonal decomposition (POD) is used to extract the dominant dynamic modes of the flow field from PIV data, identify the modal characteristics of the flow field, and extract the correlation coefficients of the dominant modes. The aforementioned modal coefficients are input into a pre-trained convolutional autoencoder clustering network. A deep learning clustering algorithm is used to classify the PIV data according to the typical stages of dynamic stall vortex evolution. The clustering network divides the flow state into K classes in an unsupervised manner, obtaining the classification results. Using the flow field modal characteristics of each class as input and the synchronously measured instantaneous thrust coefficient as output, the flow field characteristics within each class are compared with C... T Perform least squares linear regression to establish C for each category. T The prediction sub-model is called by the cluster category index to establish the mapping relationship between flow field characteristics and thrust coefficient, and the intelligent association model is obtained. The K value is determined according to the principle of minimizing the fluctuation of thrust coefficient change.

[0021] Key geometric parameters of the blades were screened using orthogonal experimental design (OED), and then the experimental system model and the various levels and interactive conditions of the geometric parameters were constructed using central composite design (CCD) under the response surface methodology (RSM) to obtain the instantaneous power coefficient C. P and thrust coefficient C TThe performance curves were analyzed, and a multidimensional response surface model of the fluctuations in maximum power coefficient and thrust coefficient with respect to geometric parameters was established. A genetic algorithm (GA) was introduced to maximize the power coefficient C. P Minimize the thrust coefficient C T The fluctuation is used as a dual objective for global optimization to determine the optimal combination of geometric parameters and obtain the optimal geometric configuration. The experimental system model includes a modular VATT scaled-down experimental model, a dynamic load measurement system, and a return water tank experimental platform. The geometric parameters include the blade airfoil section, aspect ratio AR, and cable-wheel ratio SC. An active pitch control mechanism and a passive pitch control mechanism are configured and installed in an optimal geometry. Different pitch control strategies are selected and executed. The active pitch control mechanism achieves active control by adjusting the pitch angle β in real time according to the blade azimuth angle θ through an electric drive unit and a control system. The passive pitch control mechanism performs adaptive passive pitch control by utilizing the self-balancing relationship between the elastic restoring torque generated by the elastic hinge and the hydrodynamic torque on the blade. Pitch control strategies include fixed pitch, active pitch control, and passive adaptive pitch control. By setting a passive turbulence grid in the return water tank, controllable incoming turbulence with different turbulence intensities TI and integral scales Λ is generated. The suppression effect of optimal geometric configuration and optimal pitch strategy on vorticity field and turbulence fluctuation intensity is measured. A predictive model of the impact of turbulent incoming flow on performance is established, and a robustness correction factor is obtained. The experimental model is optimized based on the robustness correction factor. Phase-locked particle image velocimetry (PIV) technology is employed to perform high spatiotemporal resolution flow field measurements under typical operating conditions for optimal geometric configurations and optimal pitch control strategies. The measurement area covers the boundary layer around the blades and the near-wake region. The PIV system achieves phase synchronization triggering via turbine angle encoder signals. Dynamic mode decomposition (DMD) or intrinsic orthogonal decomposition (POD) is used to extract the dominant dynamic modes of the flow field and identify flow field modal characteristics. Deep learning clustering algorithms are used to classify the PIV data according to the typical stages of dynamic stall vortex evolution. Using the classified flow field modal characteristics as input and the synchronously measured instantaneous thrust coefficient as output, an intelligent correlation model is established.

[0022] The specific steps for constructing a collaborative optimization mechanism between experimental design and intelligent model are as follows: The intelligent correlation model is trained using turbine experimental data obtained under different flow field conditions and different parameter couplings. Meanwhile, the flow field analysis is used to impose physical constraints on the intelligent correlation model, and the constraints are repeatedly trained to obtain a quantitative correlation model with physical basis. Under other operating conditions, the optimal combination of turbine geometry and operating parameters is predicted by a quantitative correlation model, and the parameter combination is fed back into the experimental design in the first step, finally obtaining the results of multi-parameter coupled experiment and intelligent model collaborative optimization.

[0023] Compared with the prior art, the present invention has the following beneficial effects: This invention constructs a collaborative optimization method for multi-parameter coupled experiments and intelligent models of vertical axis tidal power turbines. It forms a complete set of core technical features from four dimensions: underlying technical architecture, optimization method system, experimental testing means, and intelligent analysis mechanism. This invention innovatively builds an integrated collaborative optimization framework for the operating parameters of geometric parameters such as airfoil section, aspect ratio, and cable-wheel ratio, and pitch angle pitch strategy. It designs a multi-parameter coupled response surface model using CCD, Response Surface Method (RSM), and Genetic Algorithm (GA). It incorporates the differentiated hydrodynamic characteristics of blade dynamic stall under different geometric configurations into the pre-constraints of pitch strategy design. Instead of deriving pitch strategy in isolation, it matches a dedicated optimal pitch strategy for each set of optimal geometric parameters. This completely solves the limitation of traditional single-airfoil pitch strategy being ineffective or even counterproductive when applied to other configurations, achieving global optimization in parameter space rather than local single-point optimization.

[0024] Secondly, this invention does not adopt the traditional single-factor optimization mode of control variables. Instead, it uses a three-parameter coupled experimental design to fully capture the strong nonlinear interactive coupling effect between airfoil, aspect ratio, and cable-wheel ratio. Through multi-objective Pareto optimization, it takes into account both maximizing the power coefficient and minimizing the thrust coefficient fluctuation, avoiding the systematic error caused by the simple superposition of single-factor optimal values. This allows the optimized combination of geometric parameters to be adapted to multiple blade tip speed ratios and multiple operating conditions, greatly improving the adaptability and operational stability of the turbine.

[0025] Meanwhile, this invention introduces a passive turbulence grid in the return flume to generate a controllable gradient turbulent inflow. The system tests the performance degradation law of the optimal geometry and pitch combination under different turbulence intensities and turbulence integral scales, and establishes a quantitative robust prediction model for turbulent inflow-hydrodynamic performance. At the same time, it uses unsteady Reynolds-Averaged Navier-Stokes (URANS) or Large Eddy Simulation (LES) numerical simulations for verification analysis, clarifying the interaction mechanism between turbulence, geometric parameters, and pitch strategy. This makes up for the shortcomings of existing research which is limited to low-turbulence uniform inflow, effectively bridging the performance prediction deviation between ideal laboratory conditions and the complex tidal environment of the real ocean, and enabling the optimized scheme to have practical adaptability under engineering sea conditions.

[0026] Finally, this invention introduces phase-locked particle image velocimetry (PIV) high-resolution flow field measurement technology, combined with DMD dynamic mode decomposition, POD intrinsic orthogonal decomposition, and deep learning clustering intelligent analysis methods. It quantitatively analyzes the microscopic flow field structure of blade dynamic stall vortex evolution, leading-edge vortex shedding, and wake vortex dissipation under multi-parameter coupled conditions. It establishes a quantitative correlation model from flow field modal characteristics to macroscopic thrust load fluctuations, changing the existing mode of optimization that relies solely on empirical trial and error based on macroscopic indicators such as power coefficient and thrust coefficient. It elucidates the intrinsic mechanism of geometric parameters and pitch control strategies in suppressing dynamic stall and reducing load pulsation from the perspective of fluid physics, and provides an interpretable, reusable, and cross-condition predictive extrapolation physical basis for optimized design. It fundamentally overcomes the inherent defects of unclear flow field microscopic mechanisms and lack of theoretical support for optimization. Attached Figure Description

[0027] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0028] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0029] Table 1 Abbreviations This invention proposes a method for the collaborative optimization of multi-parameter coupling experiments and intelligent models for vertical axis tidal power turbines. The method includes the following steps: S1. Construct the experimental system model and geometric parameters at various levels and in interactive conditions, and obtain the instantaneous power coefficient C. P and thrust coefficient C T The performance curve is used to perform global optimization with the dual objectives of maximizing the instantaneous power coefficient and minimizing the thrust coefficient variation fluctuation, and the optimal combination of geometric parameters is determined to obtain the optimal geometric configuration; S2. Set up an active pitch mechanism and a passive pitch mechanism, install the active pitch mechanism and the passive pitch mechanism on the optimal geometric configuration, obtain and adopt different pitch strategies, select and execute the optimal pitch strategy, that is, the optimal operating parameters. S3. Measure the instantaneous power coefficient C under optimal geometric configuration and optimal pitch strategy using an experimental system model. P and thrust coefficient C T The attenuation characteristics were investigated, and a quantitative prediction model of the impact of turbulent inflow on performance was established. A robustness correction factor was obtained, and the experimental system model was optimized based on the robustness correction factor. S4. Perform high spatiotemporal resolution flow field measurements on the optimal geometric configuration and optimal pitch strategy under different flow field conditions to obtain PIV data. Analyze the PIV data using dynamic mode decomposition (DMD) or intrinsic orthogonal decomposition (POD). Construct an intelligent correlation model between the analyzed flow field characteristics and turbine load fluctuations using a deep learning clustering algorithm. S5. Based on the multi-parameter coupling experimental data and flow field analysis of the turbine under different flow field conditions, the intelligent correlation model is trained and physically constrained to obtain a quantitative correlation model. The optimal combination of turbine geometry and operating parameters is predicted through the quantitative correlation model, and this parameter combination is fed back to the optimized experimental system model in S1 and S2 for verification. Finally, the collaborative optimization results of multi-parameter coupling experiment and intelligent model are obtained. The results include the optimal combination of turbine geometry and operating parameters, and the results are used for the design of actual vertical axis tidal power turbines.

[0030] The flowchart of the method is as follows: Figure 1 As shown.

[0031] The technical solution of this invention consists of three functional modules, representing a three-stage progressive technical route and design system: macroscopic optimization testing, microscopic mechanism revelation, and robustness verification and modeling. The three modules are interconnected, with the output of the previous module serving as the input for the next. Module 1: Geometric Parameter Coupling Optimization Module. Based on multi-parameter coupling experiments, a response surface model between geometric parameters and performance indicators is established, and the optimal combination of geometric parameters is determined through multi-objective optimization.

[0032] Module 2: Pitch Strategy Cooperative Control Module. Based on the optimal geometric configuration determined in Module 1, a novel pitch mechanism and control strategy are designed and verified. Simultaneously, the impact of turbulent inflow on performance is systematically studied, and a robust attenuation model is established.

[0033] Module 3: Flow Field Visualization and Mechanism Analysis. This module employs high-precision flow field measurement technology and intelligent data analysis methods to quantitatively reveal the flow field mechanism of the optimal solution and construct a correlation model between the flow field structure and macroscopic loads.

[0034] The specific implementation plan is as follows: Module 1: Geometric Parameter Coupling Optimization This module breaks through the limitations of traditional single-factor optimization. It first uses OED to screen key geometric parameters of the blade, and then uses CCD in RSM to plan the multi-level interactive experimental matrix of these key geometric parameters. It systematically studies the impact of three core geometric parameters—airfoil section, aspect ratio (AR), and cable-wheel ratio (SC)—on the power coefficient (C). P and thrust coefficient C TThe coupling effect is investigated. A multidimensional performance response surface is established using the Response Surface Method (RSM). A multi-objective global optimization is performed using the Genetic Algorithm (GA) to generate the Pareto optimal front and determine the optimal combination of geometric parameters that balances the highest energy capture efficiency with the lowest load fluctuation.

[0035] The experimental system consists of the following parts: Modular VATT scaled-down experimental model: made using high-precision CNC machining and 3D printing technology, the blade cross-sectional shape, aspect ratio and cable-reel ratio are all replaceable / adjustable modular designs, and the manufacturing precision ensures the accurate controllability of each parameter.

[0036] Dynamic load measurement system: Includes a high-frequency dynamic torque sensor and a thrust sensor, capable of acquiring the instantaneous power coefficient C as it varies with azimuth angle in real time. P (θ) and thrust coefficient C T (θ).

[0037] Reflux tank experimental platform: provides stable and controllable uniform inflow conditions, meeting the velocity range and cross-sectional size requirements of scaled-down model experiments.

[0038] Experimental design and optimization process: The experimental matrix is ​​planned according to CCD, covering all levels and interactive conditions of the three geometric parameters; For each geometric configuration, a systematic scan was performed across the entire tip speed ratio (TSR) range to obtain C. P and C T The complete performance curve; C was established using RSM fitting. P and C T A multidimensional response surface model of fluctuation quantity to three geometric parameters; To maximize C P With minimizing C T The fluctuation is a dual objective; global Pareto optimization is performed using GA to determine the optimal combination of geometric parameters. min F(x) = [f1(x), f2(x)] T = [-C P (x),σ CT (x)] T x=[x1,x2,x3,x4,x5,x6,x7] T =[c,λ,β,N,H / D,δ,t / c] T Where σ CT For thrust fluctuation, c is the chord length, λ is the tip speed ratio, N is the number of blades, H / D is the aspect ratio, δ is the blade camber angle, t / c is the relative thickness, and f1(x) represents the first objective, which is to maximize C. Pf2(x) represents the second objective, which is to minimize C. T Fluctuation, F(x) represents a dual objective, and x1, x2, x3, x4, x5, x6, x7 represent c, λ, β, N, H / D, δ, t / c respectively.

[0039] Experimental verification and uncertainty analysis were conducted on the optimal combination of geometric parameters to confirm the effectiveness and reliability of the optimization scheme.

[0040] Module 2: Pitch Strategy Cooperative Control Novel Pitch Mechanism Design. This invention designs two types of novel pitch mechanisms, both mounted on the optimal geometric configuration determined in Module 1: Active pitch mechanism: Through electric drive unit and control system, it realizes active control of pitch angle β to adjust in real time with blade azimuth angle θ. It can execute any preset pitch strategy β(θ) and the tracking accuracy meets the experimental requirements.

[0041] Passive pitch mechanism: It utilizes the self-balancing relationship between the elastic restoring torque generated by the elastic hinge and the hydrodynamic torque on the blade to achieve adaptive passive pitch. It has a simple structure and does not require an active control system.

[0042] The specific design and integration methods of the two types of pitch mechanisms are as follows: The active pitch mechanism uses a servo motor to drive the pitch shaft rotation via a synchronous belt or gear transmission. The control system outputs the corresponding pitch angle β command based on the real-time blade azimuth angle θ signal fed back by the encoder, achieving closed-loop control with high tracking accuracy. The sealing scheme uses a combination of a rotating dynamic seal and a static seal, and signal transmission is achieved through a waterproof slip ring. The passive pitch mechanism adjusts the adaptive pitch law by changing the elastic restoring torque characteristics through adjusting the spring stiffness coefficient and preload torque of the elastic hinge. The structure is integrated with the blade bearing housing via a flange, requiring no electrical connection, resulting in a simple and reliable structure.

[0043] The experiment content of the pitch strategy experiment includes system testing of various pitch strategies for the turbine C. P and C T The effects of fixed pitch, active pitch, and passive adaptive pitch are considered to determine the optimal pitch strategy.

[0044] The study on turbulence effects and robustness involved installing a passive turbulence grid in the return flume to generate controllable incoming turbulence with different turbulence intensities TI and integral scales Λ. The system then measured the C0 of the optimal turbine design under various turbulence conditions. P and C T We will investigate the attenuation characteristics and establish a quantitative prediction model for the impact of turbulent inflow on performance, providing a robust correction factor for engineering applications.

[0045] CFD numerical simulation collaborative verification uses unsteady numerical simulation methods (URANS or LES) to establish a numerical model corresponding to the experiment, and verifies it with experimental macroscopic performance data. The verified model is then used to perform refined numerical analysis on the optimal pitch scheme and turbulent attenuation flow field, which is mutually verified with the experiment and guides the optimization of the experimental scheme in reverse.

[0046] Module 3: Flow Field Visualization and Intelligent Mechanism Analysis Phase-locked PIV flow field measurement: Phase-locked particle image velocimetry (PIV) technology was employed to perform high spatiotemporal resolution flow field measurements on the optimal geometric and optimal pitch models under typical operating conditions. The measurement area covered the boundary layer around the blades and the near-wake region. The PIV system achieved precise phase synchronization triggering via turbine angle encoder signals, ensuring stable acquisition of flow field images at a preset azimuth angle. The instantaneous velocity field was calculated using a cross-correlation algorithm, thereby obtaining the vorticity field and turbulence intensity distribution.

[0047] Intelligent data analysis: This invention innovatively combines Dynamic Mode Decomposition (DMD) or Intrinsic Orthogonal Decomposition (POD) with deep learning clustering algorithms for application to VATT high-dimensional PIV data: DMD or POD mode decomposition: extract the dominant dynamic modes of the flow field and identify the dominant frequency characteristics and spatial structure of the dynamic stall vortex development, shedding and dissipation process; Deep learning clustering: A convolutional autoencoder is used to reduce the dimensionality of the high-dimensional flow field, and a clustering algorithm is combined to automatically classify the typical stages of dynamic stall vortex evolution; Flow field-load intelligent correlation model: instantaneous C-values ​​measured synchronously with flow field modal characteristics as input. T (θ) is the output. An intelligent correlation model is established to accurately locate and quantify the source of instantaneous load fluctuations in the turbine, providing a data-driven basis for subsequent active load suppression, and feeding back the optimal combination parameters into the experimental design.

[0048] Compared with the closest existing technical solution, the beneficial effects and economic benefits of this invention are as follows: (1) Beneficial effects: To achieve synergistic optimization of geometric parameters and pitch strategy and overcome the limitations of local optima: This invention establishes a complete multi-parameter coupled response surface through OED+RSM and uses a genetic algorithm to perform Pareto multi-objective optimization in the global parameter space. It can identify and utilize the positive interaction effect between geometric parameters. The determined optimal solution is superior to the simple superposition of any single-factor optimization solution in terms of power coefficient and load stability, achieving true synergistic global optimization.

[0049] The robust prediction model for turbulence fills a gap in engineering applications: This invention is the first to systematically establish a quantitative prediction model for the performance degradation of VATT under different turbulence intensities (TI) and integral scales (Λ). This model can be directly used as an engineering design correction factor, transforming the optimal laboratory scheme into actual performance prediction under real marine conditions, effectively bridging the long-standing systematic deviation between laboratory results and engineering measurements in existing technologies.

[0050] Intelligent flow field analysis provides a physical basis for optimized design: This invention, through an intelligent analysis system of PIV+DMD / POD + deep learning, reveals for the first time the intrinsic physical mechanism of geometric optimization and pitch control strategy on dynamic stall vortex delay and wake vortex dissipation suppression under multi-parameter coupling conditions. It establishes a quantitative causal chain from the microstructure of the flow field to macroscopic performance indicators, providing a solid physical theoretical basis for subsequent precise engineering design of turbines.

[0051] Achieving Collaborative Optimization of Multi-Parameter Coupled Experiments and Intelligent Models: This invention first uses VATT multi-parameter coupled experimental design to obtain corresponding experimental data, which is then used as data training for the intelligent prediction model. Simultaneously, flow field analysis is performed using DMD and POD to give the intelligent prediction model practical physical meaning and basis. Then, the trained intelligent model with physical constraints is used to predict the optimal geometric parameters and pitch strategy, which is fed back into the experimental design, providing a more efficient and accurate design scheme for turbine performance optimization in reality.

[0052] Dual protection of system and method, high engineering transformation value: This invention provides both hardware system (modular turbine model and pitch mechanism) and method (collaborative optimization process and strategy), forming dual coverage of product rights and method rights. The optimized solution can be directly used in the engineering design of a new generation of high-efficiency vertical axis tidal power turbines, and has direct industrial application value.

[0053] (2) Economic benefits If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A method for co-optimization of multi-parameter coupling experiments and intelligent models for vertical axis tidal current turbines, characterized in that, The method includes the following steps: S1. Construct the experimental system model and geometric parameters at various levels and in interactive conditions, and obtain the instantaneous power coefficient C. P and thrust coefficient C T The performance curve is used to perform global optimization with the dual objectives of maximizing the instantaneous power coefficient and minimizing the thrust coefficient variation fluctuation, and the optimal combination of geometric parameters is determined to obtain the optimal geometric configuration; S2. Set up an active pitch mechanism and a passive pitch mechanism, install the active pitch mechanism and the passive pitch mechanism on the optimal geometric configuration, obtain and adopt different pitch strategies, select and execute the optimal pitch strategy, that is, the optimal operating parameters. S3. Measure the instantaneous power coefficient C under optimal geometric configuration and optimal pitch strategy using an experimental system model. P and thrust coefficient C T The attenuation characteristics were investigated, and a quantitative prediction model of the impact of turbulent inflow on performance was established. A robustness correction factor was obtained, and the experimental system model was optimized based on the robustness correction factor. S4. Perform high spatiotemporal resolution flow field measurements on the optimal geometric configuration and optimal pitch strategy under different flow field conditions to obtain PIV data. Analyze the PIV data using dynamic mode decomposition (DMD) or intrinsic orthogonal decomposition (POD). Construct an intelligent correlation model between the analyzed flow field characteristics and turbine load fluctuations using a deep learning clustering algorithm. S5. Based on the multi-parameter coupling experimental data and flow field analysis of the turbine under different flow field conditions, the intelligent correlation model is trained and physically constrained to obtain a quantitative correlation model. The optimal combination of turbine geometry and operating parameters is predicted through the quantitative correlation model, and this parameter combination is fed back to the optimized experimental system model in S1 and S2 for verification. Finally, the collaborative optimization results of multi-parameter coupling experiment and intelligent model are obtained. The results include the optimal combination of turbine geometry and operating parameters, and the results are used for the design of actual vertical axis tidal power turbines.

2. The method for multi-parameter coupling experiment and intelligent model collaborative optimization of a vertical axis tidal power turbine according to claim 1, characterized in that, The experimental system model includes a modular VATT scaled-down experimental model, a dynamic load measurement system, and a reflux water tank experimental platform.

3. The method for multi-parameter coupling experiment and intelligent model collaborative optimization of a vertical axis tidal current turbine according to claim 1, characterized in that, The geometric parameters include the blade airfoil section, aspect ratio AR, and cable-wheel ratio SC.

4. The method for multi-parameter coupling experiment and intelligent model collaborative optimization of a vertical axis tidal power turbine according to claim 3, characterized in that, Obtain the instantaneous power coefficient C P and thrust coefficient C T The specific steps for creating the performance curve are as follows: For each set of geometric parameters, a systematic scan is performed on the entire tip speed ratio (TSR) range to obtain the instantaneous power coefficient C. P and thrust coefficient C T The performance curve.

5. The method for multi-parameter coupling experiment and intelligent model collaborative optimization of a vertical axis tidal power turbine according to claim 1, characterized in that, The active pitch mechanism achieves active control by adjusting the pitch angle β in real time according to the blade azimuth angle θ through an electric drive unit and a control system.

6. The method for multi-parameter coupling experiment and intelligent model collaborative optimization of a vertical axis tidal power turbine according to claim 1, characterized in that, The passive pitch mechanism achieves adaptive passive pitch by utilizing the self-balancing relationship between the elastic restoring torque generated by the elastic hinge and the hydrodynamic torque on the blade.

7. The method for multi-parameter coupling experiment and intelligent model collaborative optimization of a vertical axis tidal power turbine according to claim 1, characterized in that, Pitch strategies include fixed pitch, active pitch, and passive adaptive pitch.

8. The method for multi-parameter coupling experiment and intelligent model collaborative optimization of a vertical axis tidal current turbine according to claim 1, characterized in that, The instantaneous power coefficient C under optimal geometry and optimal pitch strategy was measured using an experimental system model. P and thrust coefficient C T The specific steps for attenuation characteristics are as follows: By installing a passive turbulence grid in the return flume, controllable incoming turbulence with different turbulence intensities TI and integral scales Λ was generated. The suppression effect of the micro-flow field on dynamic stall and wake dissipation under optimal geometry and optimal pitch control strategy was measured. This study elucidates the impact of high-intensity and high-scale incoming flows on the turbine's unsteady attached flow and wake structure, as well as its effect on the turbine's instantaneous hydrodynamic power coefficient C. P and thrust coefficient C T Robust attenuation effect.

9. The method for multi-parameter coupling experiment and intelligent model collaborative optimization of a vertical axis tidal power turbine according to claim 1, characterized in that, In the process of measuring the flow field with high spatiotemporal resolution for the optimal geometric configuration and optimal pitch strategy under typical operating conditions, phase-locked particle image velocimetry (PIV) technology is adopted. The measurement area covers the boundary layer around the blade and the near-wake region. The PIV system is triggered synchronously by turbine rotation angle and encoder signal phase to stably acquire flow field images as PIV data at a preset azimuth angle.

10. The method for multi-parameter coupling experiment and intelligent model collaborative optimization of a vertical axis tidal current turbine according to claim 1, characterized in that, The specific steps for analyzing PIV data and obtaining an intelligent association model by combining Dynamic Mode Decomposition (DMD) or Intrinsic Orthogonal Decomposition (POD) with deep learning clustering algorithms are as follows: Dynamic mode decomposition (DMD) or intrinsic orthogonal decomposition (POD) is used to extract the dominant dynamic modes of the flow field from PIV data and identify the modal characteristics of the flow field. Deep learning clustering algorithm was used to classify PIV data according to the typical stages of dynamic stall vortex evolution, and the classification results were obtained. A smart correlation model is established by taking the classified flow field modal characteristics as input and the synchronously measured instantaneous thrust coefficient as output.

Images