Calibration method and device for axial flow turbine through-flow calculation model

Abstract

This disclosure relates to a method and apparatus for calibrating a flow calculation model for an axial-flow turbine. The method includes: acquiring performance data of an axial-flow turbine in a gas turbine; based on the performance data under design conditions, using an intelligent optimization algorithm to correct the empirical model in the flow calculation, obtaining a corrected performance calculation model for the axial-flow turbine at the design point; based on performance data under non-design conditions, correcting the corrected performance calculation model at the design point, obtaining a corrected performance prediction model for the axial-flow turbine at the non-design point; and using the corrected performance calculation model at the design point and the corrected performance prediction model at the non-design point to predict the performance of the axial-flow turbine under new operating conditions, obtaining the performance prediction result. This scheme improves the generalization ability and prediction accuracy of the flow calculation model.

Description

Technical Field

[0001] This disclosure relates to the field of gas turbine technology, and in particular to a method and apparatus for calibrating a flow calculation model for an axial turbine. Background Technology

[0002] In related technologies, the turbine is the core component in gas turbines and aero-engines that realizes heat-work conversion, and its aerodynamic design level directly affects the overall performance of the engine. Two-dimensional flow path design, as a key aspect of turbine aerodynamic design, typically uses the streamline curvature method to quickly predict the velocity and pressure distribution and overall performance within the meridional channel. However, the accuracy of flow path calculations heavily relies on empirical models such as loss models, lag angle models, and cold gas mixing models. These models are mostly based on statistical analysis of early experimental data, and their universality and accuracy are limited. As heavy-duty gas turbines develop towards higher expansion ratios and larger flow rates, the internal flow of the turbine exhibits stronger three-dimensionality, unsteadiness, and multi-physics coupling characteristics. Traditional empirical models struggle to accurately characterize complex flow phenomena, leading to increased performance prediction deviations at design points and under varying operating conditions, thus limiting design accuracy and efficiency. Summary of the Invention

[0003] To overcome the problems existing in related technologies, this disclosure provides a method and apparatus for calibrating an axial turbine flow calculation model.

[0004] According to a first aspect of the present disclosure, a method for calibrating an axial turbine flow path calculation model is provided, comprising:

[0005] Obtain performance data of the axial flow turbine in the gas turbine, including design condition data and non-design condition data; Based on the performance data under the aforementioned design conditions, an intelligent optimization algorithm is used to correct the empirical model in the flow calculation, resulting in a corrected model for the design point performance calculation of the axial turbine. Based on the performance data under the non-design conditions, the performance calculation correction model at the design point is corrected to obtain the non-design point performance prediction correction model for the axial flow turbine. The performance of the axial flow turbine under the new operating conditions is predicted using the design point performance calculation correction model and the non-design point performance prediction correction model, and the performance prediction results are obtained.

[0006] According to a second aspect of the present disclosure, a calibration apparatus for an axial turbine flow calculation model is provided, comprising: The acquisition unit is used to acquire performance data of the axial flow turbine in the gas turbine, including design condition data and non-design condition data. The first correction unit is used to correct the empirical model in the flow calculation based on the performance data under the design conditions using an intelligent optimization algorithm, so as to obtain the corrected model for the design point performance calculation of the axial turbine. The second correction unit is used to correct the design point performance calculation correction model based on the performance data under the non-design conditions, so as to obtain the non-design point performance prediction correction model of the axial flow turbine. The prediction unit is used to predict the performance of the axial flow turbine under new operating conditions using the design point performance calculation correction model and the non-design point performance prediction correction model, and obtain the performance prediction results.

[0007] According to a third aspect of the present disclosure, an electronic device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method as described in any one of the first aspects.

[0008] According to a fourth aspect of the present disclosure, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the method as described in any one of the first aspects.

[0009] According to a fifth aspect of the present disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements the method as described in any one of the first aspects.

[0010] The technical solutions provided by the embodiments of this disclosure can include the following beneficial effects: by acquiring axial turbine performance data covering both design and non-design conditions, a comprehensive and high-precision data foundation is provided for model calibration; based on the design condition data, an intelligent optimization algorithm is used to correct the empirical model in the flow calculation, effectively reducing the systematic deviation of the performance prediction at the design point and significantly improving the calculation accuracy under the design condition; further, based on the non-design condition data, the design point correction model is corrected a second time, enabling the model to accurately capture the complex flow characteristics under varying conditions and ensuring its prediction reliability under non-design conditions; finally, the calibrated design point and non-design point correction models are used to predict the performance of new conditions, achieving accurate prediction across the entire operating range from the design point to the non-design condition, significantly improving the generalization ability and prediction accuracy of the flow calculation model, thereby reducing the number of design iterations and improving the efficiency of turbine aerodynamic design.

[0011] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0012] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0013] Figure 1This is a flowchart illustrating a calibration method for an axial turbine flow calculation model according to an exemplary embodiment.

[0014] Figure 2 This is a flowchart illustrating the calculation process of the streamline curvature method as shown in an embodiment of this disclosure.

[0015] Figure 3 This is a flowchart of the ANN-GWO intelligent optimization algorithm shown in an embodiment of this disclosure.

[0016] Figure 4 This is a flowchart illustrating non-design point performance prediction correction according to an exemplary embodiment.

[0017] Figure 5 This is a decision flowchart illustrating the design point performance prediction correction model for correcting the moving blades and stationary blades in an embodiment of this disclosure.

[0018] Figure 6 This is a block diagram illustrating a calibration device for an axial turbine flow calculation model according to an exemplary embodiment.

[0019] Figure 7 This is a block diagram illustrating an apparatus for a method of calibrating a flow calculation model for an axial turbine, according to an exemplary embodiment. Detailed Implementation

[0020] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the invention as detailed in the appended claims.

[0021] The terminology used in this disclosure is for the purpose of describing particular embodiments only and is not intended to be limiting of the present disclosure. The singular forms “a” and “the” as used in this disclosure and the appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise.

[0022] It should be understood that although the terms first, second, third, etc., may be used to describe various information in embodiments of this disclosure, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first information may also be referred to as second information without departing from the scope of embodiments of this disclosure, and similarly, second information may also be referred to as first information. Depending on the context, the words “if” and “suppose” as used herein may be interpreted as “when”, “when”, or “in response to a determination”.

[0023] Furthermore, various forms of processes shown in the embodiments of this disclosure can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and no limitation is imposed herein.

[0024] It should be noted that the collection, storage, use, processing, transmission, provision, and disclosure of user personal information involved in the technical solution disclosed herein all comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0025] Figure 1 This is a flowchart illustrating a calibration method for an axial turbine flow path calculation model according to an exemplary embodiment, such as... Figure 1 As shown, it should be noted that the axial turbine flow path calculation model calibration method of this disclosure is applied in an axial turbine flow path calculation model calibration device. For example... Figure 1 As shown, the method may include the following steps: Step 101: Obtain the performance data of the axial flow turbine in the gas turbine.

[0026] The performance data includes design condition data and non-design condition data.

[0027] In one embodiment, high-precision performance data of axial-flow turbines can be collected. For existing mature models, operational data can be collected from publicly available literature; for newly designed models, performance data needs to be obtained through experiments or three-dimensional computational fluid dynamics (CFD) simulation analysis. The acquired performance data is then organized and analyzed, and reliable and valid data are selected to form a high-precision database for subsequent model correction.

[0028] It is understandable that the above performance data includes both overall performance parameters under design conditions (such as efficiency, flow rate, expansion ratio, etc.) and detailed distribution data under non-design conditions (such as total pressure, total temperature, and distribution of airflow angle along the blade height).

[0029] Step 102: Based on the performance data under the design conditions, the empirical model in the flow calculation is corrected using an intelligent optimization algorithm to obtain the corrected model for the design point performance calculation of the axial turbine.

[0030] In this embodiment, design operating condition data (such as lumped performance parameters like efficiency, flow rate, and expansion ratio) can be used as a calibration target. An intelligent optimization algorithm automatically adjusts the internal parameters of the empirical models (including loss models and lag angle models) relied upon in the flow calculation, thereby ensuring that the corrected flow calculation results match the high-precision design operating condition data. This process transforms the original fixed empirical model based on general statistical data into a corrected model customized for a specific aircraft model, significantly improving the accuracy of design point performance prediction and providing a high-precision initial model foundation for subsequent non-design point corrections.

[0031] Furthermore, traditional flow calculations rely on empirical models (such as loss models and lag angle models), which are mostly based on statistical data from early experiments. These models have fixed parameters and limited universality, making it difficult to accurately reflect the true flow characteristics of new turbine models under complex operating conditions such as high expansion ratios and large flow rates. This step, however, uses high-precision performance data from the turbine's own design operating conditions as the calibration target. An intelligent optimization algorithm iteratively optimizes the undetermined correction coefficients in the empirical model, enabling the corrected model to accurately capture the aerodynamic and thermodynamic characteristics of the turbine at the design point, thereby effectively eliminating the systematic biases introduced by traditional empirical models.

[0032] Understandably, this design point correction model not only directly improves the prediction accuracy under design conditions, but more importantly, it provides a high-precision initial model foundation for subsequent non-design point corrections. Since the design point model has been precisely calibrated, its internal parameters (such as correction coefficients) possess good physical rationality, allowing for rapid convergence during non-design point corrections with only local fine-tuning. This avoids the computational overhead and uncertainty associated with calibration from scratch, laying a solid foundation for accurate predictions across the entire operating range.

[0033] In some embodiments of this disclosure, step 102 may specifically include the following sub-steps: Step a1: Establish an integrated model that includes a streamline curvature method flow calculation model and an intelligent optimization algorithm.

[0034] As an example, the flow calculation method employs the streamline curvature method, solving the radial equilibrium equation and invoking an empirical model to calculate the gas and heat parameters at each calculation station. An intelligent optimization algorithm is used for subsequent parameter optimization. Integrating the two forms an automated optimization platform, providing a foundation for subsequent corrections.

[0035] For example, such as Figure 2As shown, the calculation process of the streamline curvature method is as follows: Starting from the input operating point and reading geometric data, initial streamlines are first generated according to the principle of equal flow area, and then the streamline inclination angle, calculation station inclination angle, and streamline curvature are calculated. Based on this, the meridional velocity of the intermediate streamlines is estimated, and then the spanwise meridional velocity distribution is calculated by solving the radial equilibrium equation; the total flow rate is calculated based on the calculated meridional velocity distribution, and it is determined whether the total flow rate residual converges: if it does not converge, the meridional velocity is relaxed and the spanwise distribution is recalculated; if it converges, the next step is to call the empirical model to calculate the lag angle and aerodynamic blocking factor, and call the loss model to calculate the loss coefficient. The flow rate of each streamline is then calculated, and the new streamline position is obtained through interpolation. The convergence of the streamline position and meridional velocity residuals is then checked. If convergence is not achieved, the streamline position is relaxed, and the streamline angle and curvature are recalculated. If convergence is achieved, the distribution of losses along the spanwise and flow directions is further calculated, and the convergence of the efficiency residuals is checked. If convergence is not achieved, the efficiency is relaxed, and the calculation is recalculated. If convergence is achieved, the performance parameters are finally output. The entire process uses a three-layer nested iterative structure (flow rate iteration, streamline position iteration, and efficiency iteration) to ensure the convergence and accuracy of the calculation results.

[0036] Step a2: Replace the empirical model used in the streamline curvature method flow calculation model with a parameterized empirical model that includes undetermined correction coefficients.

[0037] It should be noted that traditional empirical models (such as loss models and lag angle models) have fixed forms and limited prediction accuracy. This step introduces undetermined correction coefficients to parameterize the original model, giving it adjustable degrees of freedom.

[0038] Specifically, in the mathematical expression of the basic empirical model, independent adjustable correction coefficients are set for different sources of physical loss, constructing a correction formula containing undetermined coefficients. These correction coefficients will be determined in subsequent optimization processes to adapt the model to specific machine models.

[0039] In some embodiments of this disclosure, the parameterized empirical model including the undetermined correction coefficients can specifically take the following form: Step a21, for the loss model, the parameterized total loss expression is:

[0040] in, The overall undetermined correction factor is... Based on the leaf shape loss of the basic model, for Undetermined correction coefficients Based on the secondary flow loss of the fundamental model, for Undetermined correction coefficients Shock wave loss in the basic model for Undetermined correction coefficients Based on the trailing edge loss of the basic model, for Undetermined correction coefficients Based on the tip leakage loss of the basic model, for Undetermined correction coefficients Losses due to cold air mixing for The undetermined correction coefficient.

[0041] It should be noted that the cold gas mixing loss is used to characterize the additional losses generated during the mixing of cooling gas with the mainstream. For heavy-duty gas turbines, the cold gas mixing effect has a significant impact on performance, so this scheme incorporates it into the parameterized model.

[0042] By introducing the aforementioned correction coefficients, the contribution of each loss can be flexibly adjusted, thereby improving the model's adaptability to specific aircraft models. The technical significance of this step lies in transforming fixed empirical formulas into optimizable parametric models, providing design variables for subsequent intelligent optimization.

[0043] In some embodiments, the lag angle model is expressed by the following formula:

[0044] in, For correction factor, Lagging angle of the basic model This is the lag angle of the corrected model.

[0045] Step a3: Using the performance data under the design conditions as the optimization target and the undetermined correction coefficient as the optimization variable, an intelligent optimization algorithm is used for iterative optimization.

[0046] High-precision performance data (such as efficiency and flow rate) under design conditions are used as target values. The optimization algorithm continuously adjusts the correction coefficients to make the calculation results of the flow model approximate the target data. In some embodiments of this disclosure, the intelligent optimization algorithm specifically employs an intelligent optimization algorithm combining an artificial neural network (ANN) and the grey wolf optimization algorithm (GWO).

[0047] In some embodiments of this application, step a3 may further include the following sub-steps: Step a31: Construct an artificial neural network as a proxy model for the flow calculation model to evaluate the calculation error under different combinations of correction coefficients.

[0048] Artificial neural networks (ANNs) are trained to simulate the nonlinear mapping relationship between correction coefficients and computational errors, serving as a rapid evaluation tool in subsequent optimization processes.

[0049] As an example of a possible implementation, an artificial neural network can be pre-built as a surrogate model for the flow computation model, specifically through the following training process: Several sets of correction coefficient combinations are sampled from the solution space of correction coefficients. Each set of correction coefficients is input into the streamline curvature method flow calculation model for calculation, and the corresponding performance prediction results are obtained. The error between the prediction result and the high-precision performance data under the design conditions is calculated, thus constructing a training sample set with the correction coefficient combination as input and the calculation error as output. The network structure of the artificial neural network is designed, including an input layer (the number of nodes corresponds to the dimension of the correction coefficients), several hidden layers (used to learn nonlinear mapping relationships), and an output layer (with 1 node, outputting the prediction error value). Appropriate activation functions (such as ReLU, Sigmoid, etc.) and loss functions (such as mean squared error, MSE) are selected. The training sample set is divided into a training set and a validation set. The backpropagation algorithm and gradient descent optimizer are used to iteratively update the network weights and biases, enabling the network to accurately fit the nonlinear mapping relationship between the correction coefficients and the calculation error. When the prediction accuracy on the validation set meets the requirements or the loss function no longer decreases, training is stopped, resulting in a trained artificial neural network surrogate model. This surrogate model can quickly evaluate the calculation error under any combination of correction coefficients in the subsequent optimization process, replacing the time-consuming real flow calculation, thereby significantly improving optimization efficiency.

[0050] In some embodiments, the radial equilibrium equations solved in the streamline curvature method are as follows:

[0051] in, Meridian speed, n For quasi-orthogonal, I For enthalpy transfer, S For entropy, For absolute tangential velocity, Angular velocity, r For radius, The angle between the meridian streamline and the quasi-orthogonal line. The angle between the meridional streamline and the axis. Let be the radius of curvature of the meridional streamline. m It is the meridian direction.

[0052] Step a311: Taking the performance data under the design conditions as the target, the Grey Wolf optimization algorithm is used to perform global optimization in the solution space of the correction coefficient to obtain the target optimal correction coefficient; wherein, the fitness evaluation is completed by the artificial neural network surrogate model.

[0053] Understandably, the Grey Wolf Optimization (GWO) algorithm simulates the social hierarchy and hunting behavior of grey wolves, searching for the optimal solution in the solution space. In each iteration, a pre-trained ANN is used to quickly calculate the fitness value (i.e., the computational error) under the current correction coefficient, thereby guiding the wolf pack to move towards a better region. This avoids directly calling the time-consuming flow calculation model, significantly improving optimization efficiency.

[0054] As an example, the process of optimizing the correction coefficient using the Grey Wolf optimization algorithm is as follows: First, a group of gray wolves is randomly initialized in the solution space of the correction coefficients, with each wolf's position representing a set of correction coefficient vectors to be optimized. In each iteration, the correction coefficient vectors corresponding to all current gray wolf individuals are input into a trained artificial neural network surrogate model, which quickly evaluates the computational error under each set of correction coefficients as the fitness value. The gray wolf population is then sorted according to the fitness values, and the three gray wolves with the best fitness are denoted as follows: These represent the optimal, second-best, and third-best solutions in the current population, respectively; the remaining gray wolf individuals... According to The relative positions of the wolves are updated according to the hunting mechanism of the gray wolf optimization algorithm. That is, by simulating the behavior of gray wolves surrounding, chasing and attacking prey, the entire population moves towards the optimal solution area. After all gray wolves have completed the position update, the next iteration begins, and the fitness value is recalculated and updated. Repeat the above process until the preset termination condition is met (such as reaching the maximum number of iterations or the fitness value no longer improving), and finally output the result. The wolf's position is used as the optimal correction coefficient found by the gray wolf optimization algorithm. This algorithm has strong global search capabilities and is not easily trapped in local optima. Combined with the fast evaluation of the artificial neural network surrogate model, it can efficiently approximate the global optimum in the huge solution space of correction coefficients.

[0055] Step a312: Substitute the target correction coefficient into the streamline curvature method flow calculation model for verification, and update the surrogate model based on the verification results.

[0056] Specifically, the optimal correction coefficient determined by the Grey Wolf optimization algorithm in the current iteration is input into the streamline curvature method flow calculation model to obtain the corresponding performance prediction result; the true error between the performance prediction result and the performance data under the design conditions is calculated. If the true error is less than the preset error value, the iteration is terminated, and the design point correction model is determined based on the current correction coefficient; if the true error is greater than or equal to the preset error value, the current correction coefficient and the corresponding performance prediction result are used as new sample points to expand the training dataset; the artificial neural network surrogate model is retrained using the expanded training dataset; the Grey Wolf optimization algorithm is returned to perform the next round of optimization until the true error is less than the preset error value, resulting in a completed updated surrogate model.

[0057] This verification and update mechanism ensures the continuous improvement of the accuracy of the surrogate model while guaranteeing the reliability of the final correction coefficients. Through repeated iterations, a set of correction coefficients is finally obtained that highly matches the flow model calculation results with high-precision design point data.

[0058] Step a4: Determine the correction coefficients obtained after iterative convergence as the correction model for design point performance calculation.

[0059] At this point, the empirical model in the flow calculation model has been calibrated and can accurately predict turbine performance under design conditions. This provides a good initial model for subsequent non-design point corrections.

[0060] As an example, such as Figure 3 As shown in the figure, this diagram illustrates the complete implementation process of the ANN-GWO intelligent optimization algorithm, which combines an artificial neural network (ANN) with the Grey Wolf Optimization (GWO) algorithm to achieve efficient optimization. The algorithm process is divided into two main stages: First, the construction and training of the artificial neural network are carried out, including data preparation (collecting input and output data and dividing them into training and validation sets), neural network construction (designing the network structure, selecting activation and loss functions), neural network training (adjusting network weights and biases using the training set), and network evaluation (evaluating network accuracy using the validation set). After the artificial neural network training is completed, the Grey Wolf Optimization algorithm optimization stage begins, including parameter initialization (setting population size, maximum number of iterations, etc.), initializing the wolf leader (GWO), and so on. ) and two wolves ( The system first determines the position of the wolf pack and initializes the position of the entire wolf pack. Then, it enters an iterative optimization loop: First, it checks if the termination condition is met (such as reaching the maximum number of iterations or convergence accuracy). If met, it outputs the optimal wolf position and its corresponding objective function value and ends the process. If not met, it calculates the objective function value of each individual wolf in the current wolf pack (quickly evaluated by a trained artificial neural network surrogate model), and then... The algorithm updates the positions of all gray wolves in the entire wolf pack and updates the corresponding objective function values. After completion, it returns to continue judging the termination condition, repeating this process until the condition is met. The entire process uses an artificial neural network to replace the time-consuming real model for fitness evaluation, combined with the global search capability of the gray wolf optimization algorithm, which significantly improves optimization efficiency while ensuring optimization accuracy.

[0061] In one embodiment, such as Figure 5 The flowchart for performance prediction and correction at non-design points is shown. As illustrated, this process begins with initial random sampling of the flow path. First, several sets of correction coefficient samples are generated, and the flow calculation model is called to obtain the corresponding performance results, thereby determining the initial Pareto optimal solution set. Then, the flow optimization objective and constraints are dimensionally decomposed, and artificial neural networks are constructed as surrogate models based on the decomposed objective and constraints for subsequent rapid evaluation of sample fitness. Upon entering the iteration loop, the positions of individuals in the population are updated first. Then, a feasible sample set satisfying the constraints is generated through constraint operations. The non-dominated samples within this set are then identified using the trained artificial neural network, and the Pareto optimal solution set is updated. After updating the Pareto set, the solutions in the Pareto set are checked against the constraints to ensure they meet all requirements. Next, it is determined whether the current number of evaluations exceeds the maximum set number. If not, the process returns to continue with position updates and subsequent steps; if it exceeds the maximum set number, the iteration terminates, and the optimal flow optimization point (i.e., the optimal correction coefficient) is output. The entire process replaces time-consuming real flow calculations with artificial neural network surrogate models. Combined with a multi-objective optimization framework, it can efficiently search the Pareto front while satisfying constraints, and finally obtain a corrected coefficient solution that takes into account multiple optimization objectives.

[0062] Step 103: Based on the performance data under non-design conditions, the performance calculation correction model at the design point is corrected to obtain the non-design point performance prediction correction model of the axial flow turbine.

[0063] In this embodiment of the disclosure, the design point performance calculation correction model obtained in step 102 is used as the initial model, and the high-precision performance data under non-design conditions is used to make a secondary correction to obtain a performance prediction correction model applicable to non-design points.

[0064] It should be noted that, unlike design point correction which uses lumped performance parameters as the target, non-design point correction requires more refined radial distribution data as the matching target. This is because the radial non-uniformity of the flow inside the turbine is significantly enhanced under varying operating conditions, and the overall performance parameters alone cannot accurately characterize the details of the flow field.

[0065] Specifically, for the moving blades, the correction coefficients corresponding to the loss model in the design point correction model need to be adjusted to match the distribution data of the outlet total pressure along the blade height under non-design conditions; simultaneously, the correction coefficients corresponding to the lag angle model are adjusted to match the distribution data of the outlet total temperature along the blade height. For the stationary blades, the correction coefficients of the loss model are adjusted to match the radial distribution of the outlet total pressure, and the correction coefficients of the lag angle model are adjusted to match the radial distribution of the outlet airflow angle. This correction process adopts a hierarchical iterative solution strategy: first, the correction coefficients are quickly adjusted in the simplified solver of the first configuration (this solver does not perform streamline relocation and boundary layer blockage coefficient updates), and then the adjusted correction coefficients are passed to the full solver of the second configuration for streamline relocation and boundary layer blockage update verification. If the convergence condition is not met, the process returns to the first configuration to continue adjustment until the calculation results match the target distribution data. Through this step, the design point correction model is further calibrated, enabling it to have high-precision prediction capabilities even under non-design conditions, thus laying the foundation for performance prediction under subsequent new operating conditions.

[0066] In some embodiments of this disclosure, step 103 may specifically include the following sub-steps: Step b1: Using the design point performance calculation correction model as the initial model, input the geometric parameters and boundary conditions for non-design working conditions.

[0067] Using the design point correction model obtained in step 102 as the starting point, import the geometric parameters (such as blade installation angle, flow channel size, etc.) and boundary conditions (such as inlet total temperature, total pressure, speed, etc.) of non-design conditions to prepare for non-design point calibration.

[0068] Step b2: For the moving blades of the axial turbine, adjust the correction coefficients corresponding to the loss model in the design point performance calculation correction model so that the calculation results match the distribution data of the total outlet pressure along the blade height under non-design conditions; adjust the correction coefficients corresponding to the lag angle model in the design point performance calculation correction model so that the calculation results match the distribution data of the total outlet temperature along the blade height under non-design conditions.

[0069] Step b3: For the stator blades of the axial turbine, adjust the correction coefficients corresponding to the loss model in the design point performance calculation correction model so that the calculation results match the distribution data of the total outlet pressure along the blade height under non-design conditions; adjust the correction coefficients corresponding to the lag angle model in the design point performance calculation correction model so that the calculation results match the distribution data of the outlet airflow angle along the blade height under non-design conditions.

[0070] It should be noted that the above distribution data was obtained from performance data under non-design conditions. Non-design condition performance data can include detailed radial distribution information, providing a richer set of targets for accurate calibration.

[0071] In some embodiments of this disclosure, the correction coefficients are adjusted to match the corresponding distribution data. A hierarchical iterative solution strategy can be adopted, which specifically includes the following sub-steps: Step b4: The correction coefficients are adjusted using the first configured flow solver. The first configured flow solver does not perform streamline relocation or boundary layer blockage coefficient updates.

[0072] The solver in the first configuration is designed for a simplified analysis model. It ignores streamline relocation and boundary layer blockage updates, focusing only on model correction calculations. Therefore, it is fast and suitable for quickly adjusting correction coefficients in the inner loop to match the target distribution data.

[0073] In this embodiment of the disclosure, during the non-design point correction process, the correction coefficient needs to be frequently adjusted to make the calculation results match the target distribution data. If the complete flow solver (including time-consuming streamline relocation and boundary layer blockage update) is called every time the adjustment is made, the computational cost will be extremely high.

[0074] Therefore, this step employs the simplified solver of the first configuration. Its core function is to quickly evaluate the impact of correction coefficient adjustments on the calculation results, focusing on solving the radial equilibrium equations and core calculations related to model corrections, while freezing streamline positions and boundary layer clogging coefficients. This simplification significantly improves calculation speed, allowing for rapid experimentation with different combinations of correction coefficients in inner iterations, quickly narrowing the search range.

[0075] Understandably, since the design point correction model has been precisely calibrated, its initial streamline positions and boundary layer blockage coefficients are already quite reasonable. Therefore, using a simplified solver for rapid adjustment in the initial stage of non-design point correction is both reasonable and efficient. Once the first-configuration solver finds a set of correction coefficients that match the target distribution data well, it is then passed to the second-configuration full solver for fine-tuning, thereby significantly improving the overall correction efficiency while ensuring accuracy.

[0076] Step b5: The adjusted correction coefficients are passed to the flow solver in the second configuration, which then performs streamline relocation and boundary layer blockage coefficient updates.

[0077] The second solver configuration is a complete analysis model that invokes the full streamline curvature method calculation process, including streamline relocation and boundary layer blockage coefficient updates, to obtain more accurate flow field results. This model is used to verify the effectiveness of the correction coefficients.

[0078] In this embodiment, after the simplified solver of the first configuration quickly finds a set of correction coefficients that can better match the target distribution data, it is necessary to verify the effectiveness of this set of coefficients under the complete physical model. This is because the simplified solver freezes the streamline positions and boundary layer clogging coefficients, and its calculation results may have certain approximation errors. After receiving this set of correction coefficients, the second configuration solver first recalculates the streamline positions based on the current flow field state, that is, it repositions the meridional streamlines according to the updated velocity distribution to match the mesh with the actual flow path; at the same time, it updates the boundary layer clogging coefficient to reflect the impact of boundary layer development on the effective flow area of ​​the flow channel.

[0079] Based on this, the second configuration solver calls the complete streamline curvature method calculation process (including solving the radial equilibrium equation, calculating loss and lag angle, checking flow conservation, etc.) to obtain refined flow field results that take into account streamline curvature and boundary layer effects.

[0080] Understandably, the approximate feasible solution quickly obtained by the simplified solver is placed under the complete physical model for verification and correction, ensuring that the finally determined correction coefficients not only match the target distribution data but also conform to the physical laws of real flow. If the calculation results of the second configuration solver meet the preset convergence conditions, it indicates that the current correction coefficients have sufficient accuracy; if not, the calculation results are fed back to the first configuration solver for further adjustment, forming a closed-loop optimization mechanism of inner and outer layer iterations.

[0081] Step b6: If the calculation result of the second configured flow solver does not meet the corresponding convergence condition, the calculation result is fed back to the first configured flow solver to continue adjusting the correction coefficient until the convergence condition is met.

[0082] As an example, the convergence criteria include multiple dimensions: First, the residual of streamline position change must be less than a preset threshold, meaning the deviation between the streamline position calculated in this iteration and the result of the previous iteration is sufficiently small, indicating that the flow field mesh has reached a stable state. Second, the update amount of the boundary layer blockage coefficient must be less than a preset threshold, reflecting that the impact of boundary layer development on the flow channel capacity has tended to stabilize. Third, the deviation between the calculation result and the target distribution data (such as the distribution of total pressure / total temperature at the moving blade outlet along the blade height, and the distribution of total pressure / airflow angle at the stationary blade outlet along the blade height) must meet the preset accuracy requirements, ensuring that the corrected model can accurately match the performance data under non-design conditions. When the above conditions are met simultaneously, the calculation result of the second configuration solver is considered to have converged, and the correction coefficient adjustment process ends. If any condition is not met, the current calculation result is fed back to the simplified solver of the first configuration as the initial basis for a new round of adjustment, and iteration continues until all convergence conditions are met. The setting of these convergence conditions ensures both the physical consistency of the flow field calculation and the matching accuracy of the corrected model to the target distribution data.

[0083] By iterating through inner and outer layers, both correction accuracy and computational efficiency are ensured. The number of outer loops (second configuration) is usually small, for example, it can be controlled to within a limited number in engineering practice to avoid excessive time consumption. Finally, when the calculation results of the second configuration match the target distribution data, the non-design point correction model is obtained.

[0084] In some embodiments, the number of outer loop iterations is controlled to be less than 5.

[0085] Step b7: The adjusted model is determined as the non-design point performance prediction correction model for axial turbines.

[0086] This completes the model calibration under non-design conditions, enabling the flow calculation model to accurately predict turbine performance under varying conditions.

[0087] As an example, such as Figure 4 The diagram illustrates the decision-making process for correcting moving and stationary blades in the non-design point performance prediction correction model. After the process begins, it first determines whether the current processing object is a moving blade. If it is, the process proceeds to the left branch, first checking if the distribution data of the outlet total pressure along the blade height meets the convergence condition. If not, the loss model correction module is invoked to adjust the correction coefficients to match the target total pressure distribution until convergence. After the total pressure distribution converges, the process further checks if the distribution data of the outlet total temperature along the blade height meets the convergence condition. If not, the lag angle model correction module is invoked to adjust the correction coefficients to match the target total temperature distribution until convergence. If the current processing object is a stationary blade, the process proceeds to the right branch. Similarly, it first checks if the distribution data of the outlet total pressure along the blade height meets the convergence condition. If not, the loss model correction module is invoked for matching. After the total pressure distribution converges, the process further checks if the distribution data of the outlet airflow angle along the blade height meets the convergence condition. If not, the lag angle model correction module is invoked to adjust the correction coefficients to match the target airflow angle distribution until convergence. The process ends when the loss models and lag angle models for both the moving and stationary blades are corrected and the convergence conditions are met.

[0088] Step 104: Using the design point performance calculation correction model and the non-design point performance prediction correction model, the performance of the axial flow turbine under the new operating conditions is predicted to obtain the performance prediction results.

[0089] After calibration at both design and non-design points, the resulting corrected model exhibits high generalization ability. For new unknown operating conditions (such as variable speed, variable operating conditions, variable adjustable stator angle, variable working fluid, etc.), the calibrated model can be directly used for performance prediction. During prediction, the correction coefficients can be interpolated and adjusted according to the characteristic parameters of the operating condition to obtain calculation results adapted to the new operating conditions. This step achieves accurate prediction across all operating conditions from the design point to non-design conditions, significantly improving the efficiency and accuracy of turbine aerodynamic design.

[0090] In this embodiment, the new operating condition includes, but is not limited to, operating states that differ from the coverage of existing performance data, such as variable speed, variable operating conditions, variable adjustable stator angle, and variable working fluid. During prediction, the similarity or distance between the new operating condition and existing calibration operating points is first determined based on the characteristic parameters of the new operating condition (such as speed, pressure ratio, inlet temperature, and working fluid properties). Then, the correction coefficients in the design point model and non-design point model are interpolated and adjusted to adapt the model to the flow characteristics of the new operating condition. Subsequently, the adjusted correction coefficients are substituted into the flow calculation model, and the complete streamline curvature method solution process is invoked to calculate the performance parameters under the new operating condition, including efficiency, flow rate, expansion ratio, and detailed flow field distribution data.

[0091] Understandably, since the design point model and the non-design point model have undergone dual calibration based on high-precision data, their internal correction coefficients possess good physical rationality and generalization ability. Therefore, predictions for new operating conditions can achieve high accuracy while ensuring computational efficiency. This step enables extrapolation predictions from existing operating conditions to unknown operating conditions, fully leveraging the value of the calibration model and providing reliable tool support for turbine aerodynamic design and performance evaluation.

[0092] According to the calibration method for the axial turbine flow calculation model proposed in this disclosure, by acquiring axial turbine performance data covering both design and non-design conditions, a comprehensive and high-precision data foundation is provided for model calibration. Based on the design condition data, an intelligent optimization algorithm is used to correct the empirical model in the flow calculation, effectively reducing the systematic deviation of the performance prediction at the design point and significantly improving the calculation accuracy under the design condition. Then, based on the non-design condition data, the design point correction model is further corrected, enabling the model to accurately capture the complex flow characteristics under varying conditions and ensuring its prediction reliability under non-design conditions. Finally, the calibrated design point and non-design point correction models are used to predict the performance of new conditions, achieving accurate prediction across the entire operating range from the design point to non-design conditions. This significantly improves the generalization ability and prediction accuracy of the flow calculation model, thereby reducing the number of design iterations and improving the efficiency of turbine aerodynamic design.

[0093] Figure 6 This is a block diagram of a calibration device for an axial turbine flow calculation model, according to an exemplary embodiment. (Refer to...) Figure 6 The device includes an acquisition unit 601, a first correction unit 602, a second correction unit 603, and a prediction unit 604.

[0094] Among them, the acquisition unit 601 is used to acquire the performance data of the axial flow turbine in the gas turbine, including design condition data and non-design condition data; The first correction unit 602 is used to correct the empirical model in the flow calculation based on the performance data under the design conditions using an intelligent optimization algorithm, so as to obtain the design point performance calculation correction model of the axial flow turbine. The second correction unit 603 is used to correct the performance calculation correction model at the design point based on the performance data under non-design conditions, so as to obtain the non-design point performance prediction correction model of the axial turbine. The prediction unit 604 is used to predict the performance of the axial turbine under new operating conditions by using the design point performance calculation correction model and the non-design point performance prediction correction model, and obtain the performance prediction results.

[0095] In some embodiments of this application, the first correction unit 602 is specifically used for: Establish an integrated model that includes a streamline curvature method flow calculation model and an intelligent optimization algorithm; The empirical model used in the streamline curvature method flow calculation model is replaced with a parameterized empirical model that includes undetermined correction coefficients; Using performance data under design conditions as the optimization target and undetermined correction coefficients as optimization variables, an intelligent optimization algorithm is used for iterative optimization. The correction coefficients obtained after iterative convergence are determined as the correction model for performance calculation at the design point.

[0096] In some embodiments of this application, the parameterized empirical model for the undetermined correction coefficients is as follows:

[0097] in, The overall undetermined correction factor is... Based on the leaf shape loss of the basic model, for Undetermined correction coefficients Based on the secondary flow loss of the fundamental model, for Undetermined correction coefficients Shock wave loss in the basic model for Undetermined correction coefficients Based on the trailing edge loss of the basic model, for Undetermined correction coefficients Based on the tip leakage loss of the basic model, for Undetermined correction coefficients Losses due to cold air mixing for The undetermined correction coefficient.

[0098] In some embodiments of this application, the first correction unit 602 is specifically used for: An artificial neural network is constructed as a proxy model for the flow calculation model to evaluate the calculation error under different combinations of correction coefficients; Using performance data under design conditions as the target, the Grey Wolf optimization algorithm is used to perform global optimization in the solution space of the correction coefficient to obtain the target optimal correction coefficient; among which, the fitness evaluation is completed by an artificial neural network surrogate model. The target correction coefficient was substituted into the streamline curvature method flow calculation model for verification, and the surrogate model was updated based on the verification results. Repeat the steps of global optimization in the solution space of the correction coefficients using the Grey Wolf optimization algorithm until the surrogate model meets the convergence condition and outputs the target correction coefficients.

[0099] In some embodiments of this application, the first correction unit 602 is specifically used for: The optimal correction coefficient determined by the Grey Wolf Optimization Algorithm in the current iteration is input into the streamline curvature method flow calculation model to obtain the corresponding performance prediction results. The actual error between the predicted performance results and the performance data under design conditions is calculated. If the actual error is less than the preset error value, the iteration is terminated, and the design point is corrected based on the current correction coefficient. If the actual error is greater than or equal to the preset error value, the current correction coefficient and the corresponding performance prediction result will be added as new sample points to the training dataset. The artificial neural network agent model was retrained using the expanded training dataset; Return to the Grey Wolf optimization algorithm for the next round of optimization until the actual error is less than the preset error value, and obtain the updated surrogate model.

[0100] In some embodiments of this application, the second correction unit 603 is specifically used for: The design point performance calculation correction model is used as the initial model, and the geometric parameters and preset boundary conditions of the non-design working conditions are input. For the moving blades of axial turbines, the correction coefficients corresponding to the loss model in the design point performance calculation correction model are adjusted so that the calculation results match the distribution data of the total outlet pressure along the blade height under non-design conditions; the correction coefficients corresponding to the lag angle model in the design point performance calculation correction model are also adjusted so that the calculation results match the distribution data of the total outlet temperature along the blade height under non-design conditions. For the stator blades of axial turbines, the correction coefficients corresponding to the loss model in the design point performance calculation correction model are adjusted to make the calculation results match the distribution data of the total outlet pressure along the blade height under non-design conditions; the correction coefficients corresponding to the lag angle model in the design point performance calculation correction model are also adjusted to make the calculation results match the distribution data of the outlet airflow angle along the blade height under non-design conditions. The adjusted model was determined as the modified model for predicting the performance of axial turbines at non-design points; the distribution data was obtained from the performance data under non-design conditions.

[0101] In some embodiments of this application, the first correction unit 602 is specifically used for: The correction coefficients are adjusted using the first configuration of the flow solver. The first configuration of the flow solver does not perform streamline relocation or boundary layer blockage coefficient updates. The adjusted correction coefficients are passed to the second configuration flow solver, which performs streamline relocation and boundary layer blockage coefficient updates. If the calculation result of the second configuration flow solver does not meet the corresponding convergence condition, the calculation result is fed back to the first configuration flow solver to continue adjusting the correction coefficient until the calculation result meets the corresponding convergence condition.

[0102] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0103] The axial turbine flow calculation model calibration device proposed in this disclosure provides a comprehensive and high-precision data foundation for model calibration by acquiring axial turbine performance data covering both design and non-design conditions. Based on the design condition data, an intelligent optimization algorithm is used to correct the empirical model in the flow calculation, effectively reducing the systematic deviation of the performance prediction at the design point and significantly improving the calculation accuracy under the design condition. Then, based on the non-design condition data, the design point correction model is further corrected, enabling the model to accurately capture the complex flow characteristics under varying conditions and ensuring its prediction reliability under non-design conditions. Finally, the calibrated design point and non-design point correction models are used to predict the performance of new conditions, achieving accurate prediction across the entire operating range from the design point to non-design conditions. This significantly improves the generalization ability and prediction accuracy of the flow calculation model, thereby reducing the number of design iterations and improving the efficiency of turbine aerodynamic design.

[0104] Figure 7 This is a block diagram illustrating an apparatus for a method of calibrating a flow calculation model for an axial turbine, according to an exemplary embodiment. For example, apparatus 700 may be an electronic device, such as a mobile phone, computer, digital broadcasting terminal, messaging device, tablet device, personal digital assistant, etc.

[0105] Reference Figure 7 The device 700 may include one or more of the following components: a processing component 702, a memory 704, a power component 706, a multimedia component 708, an audio component 710, an input / output (I / O) interface 712, a sensor component 714, and a communication component 716.

[0106] Processing component 702 typically controls the overall operation of device 700, such as operations associated with display, telephone calls, data communication, camera operation, and recording. Processing component 702 may include one or more processors 720 to execute instructions to complete all or part of the steps of the methods described above. Furthermore, processing component 702 may include one or more modules to facilitate interaction between processing component 702 and other components. For example, processing component 702 may include a multimedia module to facilitate interaction between multimedia component 708 and processing component 702.

[0107] Memory 704 is configured to store various types of data to support the operation of device 700. Examples of this data include instructions for any application or method operating on device 700, contact data, phonebook data, messages, pictures, videos, etc. Memory 704 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0108] The power supply component 706 provides power to the various components of the device 700. The power supply component 706 may include a power management system, one or more power sources, and other components associated with generating, managing, and distributing power to the device 700.

[0109] Multimedia component 708 includes a screen that provides an output interface between device 700 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touchscreen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensors may sense not only the boundaries of touch or swipe actions but also the duration and pressure associated with the touch or swipe operation. In some embodiments, multimedia component 708 includes a front-facing camera and / or a rear-facing camera. When device 700 is in an operating mode, such as a shooting mode or a video mode, the front-facing camera and / or rear-facing camera may receive external multimedia data. Each front-facing camera and rear-facing camera may be a fixed optical lens system or have focal length and optical zoom capabilities.

[0110] Audio component 710 is configured to output and / or input audio signals. For example, audio component 710 includes a microphone (MIC) configured to receive external audio signals when device 700 is in an operating mode, such as call mode, recording mode, and voice recognition mode. The received audio signals may be further stored in memory 704 or transmitted via communication component 716. In some embodiments, audio component 710 also includes a speaker for outputting audio signals.

[0111] I / O interface 712 provides an interface between processing component 702 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to, home buttons, volume buttons, power buttons, and lock buttons.

[0112] Sensor assembly 714 includes one or more sensors for providing state assessments of various aspects of device 700. For example, sensor assembly 714 may detect the on / off state of device 700, the relative positioning of components such as the display and keypad of device 700, changes in the position of device 700 or a component of device 700, the presence or absence of user contact with device 700, the orientation or acceleration / deceleration of device 700, and temperature changes of device 700. Sensor assembly 714 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. Sensor assembly 714 may also include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, sensor assembly 714 may also include an accelerometer, a gyroscope, a magnetometer, a pressure sensor, or a temperature sensor.

[0113] Communication component 716 is configured to facilitate wired or wireless communication between device 700 and other devices. Device 700 can access wireless networks based on communication standards, such as WiFi, 2G, or 3G, or combinations thereof. In one exemplary embodiment, communication component 716 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, communication component 716 also includes a near-field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0114] In an exemplary embodiment, the apparatus 700 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the methods described above.

[0115] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 704 including instructions, which can be executed by a processor 720 of the device 700 to perform the above-described method. For example, the non-transitory computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

[0116] In an exemplary embodiment, a computer program product is also provided, including a computer program that implements the above-described method when executed by the processor 720 of the device 700.

[0117] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This disclosure is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the following claims.

[0118] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method of calibrating a throughflow calculation model of an axial turbine, characterized in that, include: Obtain performance data of the axial flow turbine in the gas turbine, including design condition data and non-design condition data; Based on the performance data under the aforementioned design conditions, an intelligent optimization algorithm is used to correct the empirical model in the flow calculation, resulting in a corrected model for the design point performance calculation of the axial turbine. Based on the performance data under the non-design conditions, the performance calculation correction model at the design point is corrected to obtain the non-design point performance prediction correction model for the axial flow turbine. The performance of the axial flow turbine under the new operating conditions is predicted using the design point performance calculation correction model and the non-design point performance prediction correction model, and the performance prediction results are obtained.

2. The calibration method of the axial flow turbine throughflow calculation model according to claim 1, characterized in that, Based on the performance data under the design conditions, an intelligent optimization algorithm is used to correct the empirical model in the flow calculation, resulting in a corrected performance calculation model for the design point of the axial turbine, including: Establish an integrated model that includes a streamline curvature method flow calculation model and an intelligent optimization algorithm; The empirical model used in the streamline curvature method flow calculation model is replaced with a parameterized empirical model that includes undetermined correction coefficients; Using the performance data under the design conditions as the optimization target and the undetermined correction coefficient as the optimization variable, the intelligent optimization algorithm is used for iterative optimization. The correction coefficients obtained after iterative convergence are determined as the performance calculation correction model for the design point.

3. The calibration method of the axial flow turbine throughflow calculation model according to claim 2, characterized in that, The parameterized empirical model for the undetermined correction coefficients is: in, The overall undetermined correction factor is... Based on the leaf shape loss of the basic model, for Undetermined correction coefficients Based on the secondary flow loss of the fundamental model, for Undetermined correction coefficients Shock wave loss in the basic model for Undetermined correction coefficients Based on the trailing edge loss of the basic model, for Undetermined correction coefficients Based on the tip leakage loss of the basic model, for Undetermined correction coefficients Losses due to cold air mixing for The undetermined correction coefficient.

4. The calibration method for the axial turbine flow calculation model according to claim 2, characterized in that, The method of using intelligent optimization algorithms to correct the empirical model in flow calculation includes: An artificial neural network is constructed as a proxy model for the flow calculation model to evaluate the calculation error under different combinations of correction coefficients; Using the performance data under the design conditions as the target, the Grey Wolf optimization algorithm is used to perform global optimization in the solution space of the correction coefficient to obtain the target optimal correction coefficient; wherein, the fitness evaluation is completed by the artificial neural network surrogate model. The target correction coefficient is substituted into the streamline curvature method flow calculation model for verification, and the surrogate model is updated based on the verification results. Repeat the steps of using the Grey Wolf optimization algorithm to perform global optimization in the solution space of the correction coefficients until the surrogate model meets the convergence condition and outputs the target correction coefficients.

5. The calibration method for the axial turbine flow calculation model according to claim 4, characterized in that, The step of substituting the target correction coefficient into the streamline curvature method flow calculation model for verification, and updating the surrogate model based on the verification results, includes: The optimal correction coefficient determined by the gray wolf optimization algorithm in the current iteration is input into the streamline curvature method flow calculation model to obtain the corresponding performance prediction results. Calculate the actual error between the performance prediction results and the performance data under the design conditions; If the actual error is less than the preset error value, the iteration is terminated, and the design point correction model is determined based on the current correction coefficient. If the actual error is greater than or equal to the preset error value, the current correction coefficient and the corresponding performance prediction result will be added as new sample points to the training dataset. The artificial neural network agent model was retrained using the expanded training dataset; Return to the Grey Wolf optimization algorithm for the next round of optimization until the actual error is less than the preset error value, and obtain the updated agent model.

6. The calibration method for the axial turbine flow calculation model according to claim 1, characterized in that, The process of modifying the design point performance calculation correction model based on the performance data under the non-design operating conditions to obtain the non-design point performance prediction correction model for the axial flow turbine includes: Using the design point performance calculation correction model as the initial model, input the geometric parameters and preset boundary conditions for non-design working conditions; For the moving blades of axial turbines, the correction coefficients corresponding to the loss model in the design point performance calculation correction model are adjusted so that the calculation results match the distribution data of the total outlet pressure along the blade height under non-design conditions; the correction coefficients corresponding to the lag angle model in the design point performance calculation correction model are also adjusted so that the calculation results match the distribution data of the total outlet temperature along the blade height under non-design conditions. For the stator blades of axial turbines, the correction coefficients corresponding to the loss model in the design point performance calculation correction model are adjusted so that the calculation results match the distribution data of the total outlet pressure along the blade height under non-design conditions; the correction coefficients corresponding to the lag angle model in the design point performance calculation correction model are also adjusted so that the calculation results match the distribution data of the outlet airflow angle along the blade height under non-design conditions. The adjusted model was determined as the off-design point performance prediction correction model for axial turbines; the distribution data was obtained from the performance data under the off-design conditions.

7. The calibration method for the axial turbine flow calculation model according to claim 6, characterized in that, The correction factor is adjusted to match the calculated result with the corresponding distribution data, including: The correction coefficients are adjusted using the first configuration of the flow solver, which does not perform streamline relocation or boundary layer blockage coefficient updates. The adjusted correction coefficients are passed to the second configured flow solver, which performs streamline relocation and boundary layer blockage coefficient updates. If the calculation result of the second configuration flow solver does not meet the corresponding convergence condition, the calculation result is fed back to the first configuration flow solver to continue adjusting the correction coefficient until the calculation result meets the corresponding convergence condition.

8. A calibration device for a flow calculation model of an axial turbine, characterized in that, include: The acquisition unit is used to acquire performance data of the axial flow turbine in the gas turbine, including design condition data and non-design condition data. The first correction unit is used to correct the empirical model in the flow calculation based on the performance data under the design conditions using an intelligent optimization algorithm, so as to obtain the corrected model for the design point performance calculation of the axial turbine. The second correction unit is used to correct the design point performance calculation correction model based on the performance data under the non-design conditions, so as to obtain the non-design point performance prediction correction model of the axial flow turbine. The prediction unit is used to predict the performance of the axial flow turbine under new operating conditions using the design point performance calculation correction model and the non-design point performance prediction correction model, and obtain the performance prediction results.

9. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 7.

Images