A modeling method for a hybrid drive gas turbine gas path system
By establishing a mechanistic model of the gas turbine gas path system and combining it with Bayesian optimization algorithm and turbine cooling effect correction, the modeling accuracy and parameter coupling problems of gas turbines under complex operating conditions are solved, and high-precision simulation and optimization are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANGHAI JIAOTONG UNIV
- Filing Date
- 2026-01-27
- Publication Date
- 2026-06-05
AI Technical Summary
Existing gas turbine gas path system modeling methods are difficult to accurately describe real behavior under complex operating conditions, and the mechanism-driven and data-driven modeling methods fail to fully consider the coupling relationship of multiple component parameters, resulting in high computational complexity and low parameter identification accuracy.
Based on thermodynamic theory, a mechanistic model of the gas turbine compressor, combustion chamber and turbine is established. Combined with actual operating data, parameters are identified and optimized through Bayesian optimization algorithm. A turbine cooling effect correction factor is introduced to construct a hybrid mechanistic data driven model.
It improves the modeling accuracy of gas turbines, reduces the complexity and computational cost of parameter identification, and is suitable for scenarios with limited computing resources or online calibration.
Smart Images
Figure CN122154399A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of gas turbine technology, and in particular to a method for modeling a hybrid-driven gas turbine gas path system. Background Technology
[0002] Gas turbines, as highly efficient thermodynamic power plants, possess advantages such as high power output, rapid start-up, and stable operation, and are widely used in marine propulsion systems, power generation equipment, and aerospace propulsion systems. With the transformation of the energy structure and the requirements for energy conservation and emission reduction, gas turbines need to achieve higher operating efficiency, reliability, and controllability under complex operating conditions. Therefore, constructing a high-fidelity model that reflects the operating status of gas turbines is of great significance for achieving gas turbine performance evaluation, operation optimization, and health management.
[0003] Existing modeling methods for gas turbine gas path systems mainly include mechanism-based modeling methods and data-based modeling methods. Mechanism-based modeling methods typically rely on fundamental theories such as thermodynamics, fluid mechanics, and heat transfer to mathematically model key components of the gas turbine, such as the compressor, combustion chamber, and turbine. However, in practical engineering applications, these methods often require the introduction of many assumptions and empirical parameters, making it difficult to accurately describe the actual behavior of the gas turbine under complex operating conditions, degradation, or failures.
[0004] Modeling methods based on operational data utilize a large amount of experimental data collected during the actual operation of gas turbines. Through machine learning, they establish input-output mapping relationships, enabling them to better fit the operating characteristics of gas turbines. However, these methods typically lack physical constraints, rely heavily on large amounts of training data, and suffer from low model stability and generalization ability when the amount of data is limited.
[0005] To address the shortcomings of single-method modeling approaches, hybrid mechanism- and data-driven modeling methods have emerged in recent years. These methods, while maintaining physical consistency, incorporate operational data to refine the model, thereby improving modeling accuracy. However, existing hybrid mechanism- and data-driven modeling methods often employ simple parameter adjustments or model weighting, failing to fully consider the parameter coupling relationships among multiple components of the gas turbine. Consequently, they still suffer from high computational complexity and low parameter identification accuracy.
[0006] Therefore, there is an urgent need for a gas turbine modeling method that can maintain physical consistency, effectively integrate actual operational data, and achieve high-precision modeling through reasonable parameter identification and optimization strategies, so as to improve the application effect of gas turbine simulation models in performance prediction and operation optimization. Summary of the Invention
[0007] To address the limitations of existing gas turbine gas path system modeling methods, this invention provides a hybrid-driven gas turbine gas path system modeling method, which includes the following steps: S1. Based on thermodynamic theory and the characteristics of gas turbine components, establish the mechanistic models of the gas turbine compressor, combustion chamber and turbine, and obtain the mechanistic model of the gas turbine gas circuit system; S2. Compare and analyze the output of the mechanism model with the actual operating data of the gas turbine, and construct an unknown parameter identification model with the deviation between the output of the mechanism model and the actual operating data as the constraint. S3. Using the Bayesian optimization algorithm, the compressor efficiency and turbine efficiency in the unknown parameter identification model are optimized in the first stage to obtain the optimal parameter combination that minimizes the deviation between the mechanism model results and the actual operating data; S4. Based on the optimal parameter combination described in S3, a correction factor characterizing the turbine cooling effect is introduced to establish a turbine outlet temperature correction model. The correction factor is then optimized in the second stage using a Bayesian optimization algorithm to further reduce the simulation deviation of the turbine outlet temperature. S5. Based on the optimal parameter combination described in S3 and the correction factor for the turbine cooling effect described in S4, a mechanism data hybrid driving model of the gas turbine gas path system is constructed.
[0008] Furthermore, the process of establishing the mechanistic model of the gas turbine gas path system described in S1 includes: S11. Based on the fundamental laws of thermodynamics and component characteristic parameters, a modular mechanism model of the compressor, combustion chamber and turbine is established to describe the compression, combustion and expansion processes of gas in each component of the gas turbine gas circuit system; S12. The gas turbine compressor outlet temperature is calculated using the aforementioned mechanism model. Combustion chamber outlet temperature Turbine exhaust temperature Gas turbine output power This provides a model foundation for subsequent identification and optimization of unknown parameters.
[0009] Furthermore, in S2, under given gas turbine operating conditions, based on the gas turbine gas path system mechanism model established in step S1, the simulation output result vector is calculated: In the formula, The vector of unknown parameters to be identified, including compressor efficiency. and turbine efficiency , The output power of the gas turbine is the output of the mechanistic model simulation. The compressor outlet temperature is the output of the mechanistic model simulation. The combustion chamber outlet temperature is the output of the mechanistic model simulation. The turbine exhaust temperature is the output of the mechanism model simulation. Simultaneously, actual operating data of the gas turbine under the same operating conditions are collected to form a measurement data vector: In the formula, To measure the actual operating output power of the gas turbine, The measured compressor outlet temperature during actual operation. The measured combustion chamber outlet temperature during actual operation. The measured turbine exhaust temperature during actual operation.
[0010] Furthermore, based on the difference between the output results of the mechanistic model and the actual operational measurement data, an objective function measured by relative error is constructed to describe the influence of unknown parameters on the model output deviation. The objective function is expressed as: in, For parameters The first mechanistic model under the condition Each simulation output quantity For the corresponding actual operational measurement data, The weights corresponding to the output results. The number of simulation output parameters, To prevent the use of pre-defined positive numbers with a denominator of zero; By minimizing the objective function This enables the identification of unknown parameters.
[0011] Furthermore, S3 specifically includes the following steps: S31. Let the parameter vector to be identified in the first stage be... for: in, For compressor efficiency, For turbine efficiency; To ensure that the parameters meet physical feasibility requirements, a search range is set: In the formula, This represents the minimum range of compressor efficiency. This represents the maximum value within the compressor efficiency range. This represents the minimum value within the turbine efficiency range. This represents the maximum value within the turbine efficiency range. S32. Using the output power of the gas turbine Compressor outlet temperature Turbine exhaust temperature The relative error is used as the joint optimization objective; for the parameter set Define relative error: in, These are the simulated values for the parameter set. For the measured values of the parameter set, It should be an extremely small positive number to prevent errors from exploding due to small amounts or zero values; S33. Taking the weighted average of the three errors in the first-stage objective function, the optimization problem is expressed as: in, This represents the optimal parameter solution obtained in the first stage of optimization. The objective function for the first stage is... The weights of the three errors; S34. Using a Bayesian optimization framework, a sampler based on a tree-structured Pareto estimator (TPE) is employed for parameter proposal; let the historical sample set be: In the formula, For the first phase One solution parameter, Error evaluation index determined by unknown parameters The number of simulation output parameters; TPE classifies the conditional probability density function of unknown parameters into good value distributions and bad value distributions: In the formula, Represents the parameter vector to be identified The probability density function within the good value interval of the objective function. Represents the parameter vector to be identified The probability density function within the inferior range of the objective function. For indicator functions, The threshold is determined based on the quantile criterion of the objective function value.
[0012] Furthermore, it also includes constructing and maximizing a combined sampling criterion function based on the probability density ratio to obtain the optimal parameter solution for the core mechanism parameters. : In the formula, , These are control parameters used to adjust the trade-off between exploration and exploitation.
[0013] Furthermore, a median pruning strategy is introduced to calculate the median of the objective function. : In the formula, This is the median calculation function. For the first The objective function value obtained from this calculation This is the set of objective function values that have been calculated. The calculation was terminated early when the calculated objective function value was significantly worse than the historical median, ultimately yielding the optimal parameter combination. .
[0014] Furthermore, S4 specifically includes the following steps: S41. Suppose that the turbine outlet temperature calculated by the mechanistic model under given parameters is... Introducing a turbine cooling effect correction factor The turbine outlet temperature is corrected to: In the formula, For correction functions, This is a correction value for the turbine outlet temperature; S42. Determine the turbine cooling effect correction factor And set its physically feasible search range as follows: In the formula, This is the minimum value in the range of the turbine cooling effect correction factor. This represents the maximum value within the range of the turbine cooling effect correction factor. S43. The objective function is the measured actual turbine exhaust temperature. Turbine outlet temperature correction value The relative error is used as the joint optimization objective, and the relative error is defined as: The second-stage optimization problem can be written as: In the formula, This represents the optimal turbine cooling effect correction factor solution obtained in the second stage of optimization; S44. Using a Bayesian optimization framework, a TPE-based sampling strategy is employed to generate candidate samples. Perform a fixed number of calculations directly: In the formula, The feasible parameter space for the turbine cooling effect correction factor. The sampling criterion function is constructed based on a tree-structured Pareto estimator. Indicates the first Candidate values for turbine cooling effect correction factors generated in sub-Bayesian optimization calculations For historical calculation sample set, This represents the maximum number of calculations.
[0015] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: 1) This invention establishes a mechanistic model of the gas turbine compressor, combustion chamber and turbine based on the basic laws of thermodynamics and component characteristics. On this basis, actual operating data is introduced to identify and correct the key parameters of the model. This enables the model to more accurately reflect the real state of the gas turbine under actual operating conditions while maintaining clear physical meaning and structural interpretability, thus effectively improving the overall accuracy of gas circuit system modeling.
[0016] 2) This invention decomposes the global optimization problem of multi-parameter coupling in the gas turbine gas path system into a phased, low-dimensional parameter optimization process with physical constraints. First, the core mechanism parameters such as compressor efficiency and turbine efficiency are jointly optimized. Then, the turbine cooling effect correction factor is optimized separately on this basis. This effectively reduces the dimensionality of the search space in the parameter identification process, reduces the non-unique solution problem caused by parameter coupling, and improves the stability and convergence efficiency of the parameter identification process.
[0017] 3) This invention employs a Bayesian optimization algorithm based on a tree-structured Pareto estimator. By probabilistically modeling historical test samples, it guides the search for unknown parameters within a high-probability good value region, reducing the number of invalid or low-value test point samplings. While ensuring optimization accuracy, it significantly reduces the computational cost required for model calibration, making it suitable for scenarios with limited computing resources or online calibration. Attached Figure Description
[0018] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0019] Figure 1 This is a flowchart of the mechanism data hybrid-driven modeling method for calculating the thermodynamic response of a gas turbine gas path system according to the present invention.
[0020] Figure 2 This is a flowchart of the Bayesian hierarchical combinatorial optimization strategy of the present invention. Detailed Implementation
[0021] The following is in conjunction with the instruction manual appendix. Figure 1-2This paper further describes in detail a modeling method for a hybrid-driven gas turbine gas path system provided by the present invention. The method includes the following steps: S1. The gas turbine gas path system is modeled using a modular mechanism model; The mechanistic model is based on thermodynamic theory and the characteristics of gas turbine components, and establishes functional modules of compressor, combustion chamber and turbine to describe the compression, combustion and expansion processes of gas in each component.
[0022] The compressor module is based on the inlet air temperature. ,pressure and mass flow rate And in conjunction with the set export pressure and efficiency Calculate the gas state changes during the compression process and output the compressor outlet temperature. and compression power ; The combustion chamber module is based on the principles of component conservation and energy conservation, combined with fuel mass flow rate. Calculate the state of the gas mixture after combustion and output the combustion chamber outlet temperature. and release of heat ; The turbine module operates at a given outlet pressure. and efficiency Under these conditions, the expansion of the combustion chamber outlet gas is calculated to obtain the turbine exhaust temperature. and turbine output power And based on the compressor compression power and turbine output power Calculate the output power of the gas turbine ; Based on the above mechanism model, the key state parameters of the gas turbine gas path system are obtained, including the compressor outlet temperature. Combustion chamber outlet temperature Turbine exhaust temperature and gas turbine output power This provides a model foundation for subsequent identification and optimization of unknown parameters.
[0023] S2. Compare and analyze the output of the mechanism model with the actual operating data of the gas turbine, and construct an unknown parameter identification model with the deviation between the output of the mechanism model and the actual operating data as the constraint. Specifically, under given gas turbine operating conditions, based on the gas turbine gas path system mechanism model established in step S1, the simulation output result vector is calculated as follows: In the formula, The vector of unknown parameters to be identified, including compressor efficiency. and turbine efficiency , The output power of the gas turbine is the output of the mechanistic model simulation. The compressor outlet temperature is the output of the mechanistic model simulation. The combustion chamber outlet temperature is the output of the mechanistic model simulation. The turbine exhaust temperature is the output of the mechanism model simulation. Simultaneously, actual operating data of the gas turbine under the same operating conditions are collected to form a measurement data vector: In the formula, To measure the actual operating output power of the gas turbine, The measured compressor outlet temperature during actual operation. The measured combustion chamber outlet temperature during actual operation. The measured turbine exhaust temperature during actual operation.
[0024] Based on the difference between the output of the mechanistic model and the actual operational measurement data, an objective function is constructed using relative error as a metric to describe the influence of unknown parameters on the model output deviation. The objective function is expressed as: in, For parameters The first mechanistic model under the condition Each simulation output quantity For the corresponding actual operational measurement data, The weights corresponding to the output results. The number of simulation output parameters, To prevent the use of pre-defined positive numbers with a denominator of zero; By minimizing the objective function This enables unified identification and modeling of unknown parameters in the gas turbine gas path system, thereby transforming mechanistic model calibration into physically constrained parameter optimization, and providing an objective function basis for subsequent parameter identification based on Bayesian optimization.
[0025] S3. Using a Bayesian optimization algorithm, the compressor efficiency and turbine efficiency in the unknown parameter identification model are optimized in the first stage to obtain the optimal parameter combination that minimizes the deviation between the mechanism model results and the actual operating data. S3 specifically includes the following steps: S31. Let the parameter vector to be identified in the first stage be... for: in, For compressor efficiency, For turbine efficiency; To ensure that the parameters meet physical feasibility requirements, a search range is set: In the formula, This represents the minimum range of compressor efficiency. This represents the maximum value within the compressor efficiency range. This represents the minimum value within the turbine efficiency range. This represents the maximum value within the turbine efficiency range. S32. Using the output power of the gas turbine Compressor outlet temperature Turbine exhaust temperature The relative error is used as the joint optimization objective; for the parameter set Define relative error: in, These are the simulated values for the parameter set. For the measured values of the parameter set, It should be an extremely small positive number to prevent errors from exploding due to small amounts or zero values; S33. Taking the weighted average of the three errors in the first-stage objective function, the optimization problem is expressed as: in, This represents the optimal parameter solution obtained in the first stage of optimization. The objective function for the first stage is... The weights of the three errors; S34. Using a Bayesian optimization framework, a sampler based on a tree-structured Pareto estimator (TPE) is employed for parameter proposal; let the historical sample set be: In the formula, For the first phase One solution parameter, Error evaluation index determined by unknown parameters The number of simulation output parameters; TPE classifies the conditional probability density function of unknown parameters into good value distributions and bad value distributions: In the formula, Represents the parameter vector to be identified The probability density function within the good value interval of the objective function. Represents the parameter vector to be identified The probability density function within the inferior range of the objective function. For indicator functions, The threshold is determined based on the quantile criterion of the objective function value. Let be the probability density function.
[0026] Construct and maximize a combined sampling criterion function based on probability density ratio to obtain the optimal parameter solution for the core mechanism parameters. : In the formula, , These are control parameters used to adjust the trade-off between exploration and exploitation.
[0027] Meanwhile, to reduce unnecessary computation, a median pruning strategy is introduced to calculate the median of the objective function. : In the formula, This is the median calculation function. For the first The objective function value obtained from this calculation This is the set of objective function values that have been calculated. The calculation was terminated early when the calculated objective function value was significantly worse than the historical median, ultimately yielding the optimal parameter combination. .
[0028] S4. Based on the optimal parameter combination described in S3, a correction factor characterizing the turbine cooling effect is introduced to establish a turbine outlet temperature correction model. The correction factor is then optimized in the second stage using a Bayesian optimization algorithm to further reduce the simulation deviation of the turbine outlet temperature. Specifically, the following steps are included: S41. Suppose that the turbine outlet temperature calculated by the mechanistic model under given parameters is... Introducing a turbine cooling effect correction factor The turbine outlet temperature is corrected to: In the formula, For correction functions, This is a correction value for the turbine outlet temperature; S42. Determine the turbine cooling effect correction factor And set its physically feasible search range as follows: In the formula, This is the minimum value in the range of the turbine cooling effect correction factor. This represents the maximum value within the range of the turbine cooling effect correction factor. S43. The objective function is the measured actual turbine exhaust temperature. Turbine outlet temperature correction value The relative error is used as the joint optimization objective, and the relative error is defined as: The second-stage optimization problem can be written as: In the formula, This represents the optimal turbine cooling effect correction factor solution obtained in the second stage of optimization; S44. Using a Bayesian optimization framework, a TPE-based sampling strategy is employed to generate candidate samples. Perform a fixed number of calculations directly: In the formula, The feasible parameter space for the turbine cooling effect correction factor. The sampling criterion function is constructed based on a tree-structured Pareto estimator. Indicates the first Candidate values for turbine cooling effect correction factors generated in sub-Bayesian optimization calculations For historical calculation sample set, This represents the maximum number of calculations.
[0029] S5. Based on the optimal parameter combination described in S3 and the correction factor for the turbine cooling effect described in S4, a mechanism data hybrid driving model of the gas turbine gas path system is constructed.
[0030] After completing the first phase of core mechanism parameters Optimization and Second Stage Turbine Cooling Effect Correction Factor Based on the optimization, the optimization results are jointly incorporated into the gas turbine gas path system mechanism model to construct a high-precision mechanism data hybrid-driven model of the gas turbine gas path system. The compressor outlet temperature is then calculated using this model under given operating conditions. Combustion chamber outlet temperature Turbine exhaust temperature and gas turbine output power Turbine exhaust temperature The cooling effect correction function is used to compensate for the influence of cooling airflow not explicitly described in the mechanism model, thereby achieving high-precision simulation and prediction of the gas turbine gas path system performance.
[0031] The thermodynamic parameters of the actual operating state of the gas turbine gas path system were compared and analyzed with the thermodynamic parameters output by the hybrid drive model of the mechanism data established by the method of this invention. Specific calculation results and error comparisons are shown in Table 1. The analysis results show that the hybrid drive gas turbine gas path system modeling method adopted in this invention has high analytical accuracy, with errors in key output thermodynamic parameters all within 1%, meeting the accuracy requirements of actual machine simulation.
[0032] Table 1 In this specification, the same or similar parts between the various embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the descriptions of the embodiments described later are relatively simple, and relevant parts can be referred to the descriptions of the foregoing embodiments.
[0033] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
Claims
1. A method for modeling a hybrid-driven gas turbine gas path system, characterized in that, Includes the following steps: S1. Based on thermodynamic theory and the characteristics of gas turbine components, establish the mechanistic models of the gas turbine compressor, combustion chamber and turbine, and obtain the mechanistic model of the gas turbine gas circuit system; S2. Compare and analyze the output of the mechanism model with the actual operating data of the gas turbine, and construct an unknown parameter identification model with the deviation between the output of the mechanism model and the actual operating data as the constraint. S3. Using the Bayesian optimization algorithm, the compressor efficiency and turbine efficiency in the unknown parameter identification model are optimized in the first stage to obtain the optimal parameter combination that minimizes the deviation between the mechanism model results and the actual operating data; S4. Based on the optimal parameter combination described in S3, a correction factor characterizing the turbine cooling effect is introduced to establish a turbine outlet temperature correction model. The correction factor is then optimized in the second stage using a Bayesian optimization algorithm to further reduce the simulation deviation of the turbine outlet temperature. S5. Based on the optimal parameter combination described in S3 and the correction factor for the turbine cooling effect described in S4, a mechanism data hybrid driving model of the gas turbine gas path system is constructed.
2. The method according to claim 1, characterized in that, The process of establishing the mechanistic model of the gas turbine gas path system described in S1 includes: S11. Based on the fundamental laws of thermodynamics and component characteristic parameters, a modular mechanism model of the compressor, combustion chamber and turbine is established to describe the compression, combustion and expansion processes of gas in each component of the gas turbine gas circuit system; S12. The gas turbine compressor outlet temperature is calculated using the aforementioned mechanism model. Combustion chamber outlet temperature Turbine exhaust temperature Gas turbine output power This provides a model foundation for subsequent identification and optimization of unknown parameters.
3. The method according to claim 2, characterized in that, In step S2, under given gas turbine operating conditions, based on the gas turbine gas path system mechanism model established in step S1, the simulation output vector is calculated: In the formula, The vector of unknown parameters to be identified, including compressor efficiency. and turbine efficiency , The output power of the gas turbine is the output of the mechanistic model simulation. The compressor outlet temperature is the output of the mechanistic model simulation. The combustion chamber outlet temperature is the output of the mechanistic model simulation. The turbine exhaust temperature is the output of the mechanism model simulation. Simultaneously, actual operating data of the gas turbine under the same operating conditions are collected to form a measurement data vector: In the formula, To measure the actual operating output power of the gas turbine, The measured compressor outlet temperature during actual operation. The measured combustion chamber outlet temperature during actual operation. The measured turbine exhaust temperature during actual operation.
4. The method according to claim 3, characterized in that, Based on the difference between the output of the mechanistic model and the actual operational measurement data, an objective function is constructed using relative error as a metric to describe the influence of unknown parameters on the model output deviation. The objective function is expressed as: in, For parameters The first mechanistic model under the condition Each simulation output quantity For the corresponding actual operational measurement data, The weights corresponding to the output results. The number of simulation output parameters, To prevent the use of pre-defined positive numbers with a denominator of zero; By minimizing the objective function This enables the identification of unknown parameters.
5. The method according to claim 3, characterized in that, S3 specifically includes the following steps: S31. Let the parameter vector to be identified in the first stage be... for: in, For compressor efficiency, For turbine efficiency; To ensure that the parameters meet physical feasibility requirements, a search range is set: In the formula, This represents the minimum range of compressor efficiency. This represents the maximum value within the compressor efficiency range. This represents the minimum value within the turbine efficiency range. This represents the maximum value within the turbine efficiency range. S32. Using the output power of the gas turbine Compressor outlet temperature Turbine exhaust temperature The relative error is used as the joint optimization objective; for the parameter set Define relative error: in, These are the simulated values for the parameter set. For the measured values of the parameter set, It should be an extremely small positive number to prevent errors from exploding due to small amounts or zero values; S33. Taking the weighted average of the three errors in the first-stage objective function, the optimization problem is expressed as: in, This represents the optimal parameter solution obtained in the first stage of optimization. The objective function for the first stage is... The weights of the three errors; S34. Using a Bayesian optimization framework, a sampler based on a tree-structured Pareto estimator (TPE) is employed for parameter proposal; let the historical sample set be: In the formula, For the first phase One solution parameter, Error evaluation index determined by unknown parameters The number of simulation output parameters; TPE classifies the conditional probability density function of unknown parameters into good value distributions and bad value distributions: In the formula, Represents the parameter vector to be identified The probability density function within the good value interval of the objective function. Represents the parameter vector to be identified The probability density function within the inferior range of the objective function. For indicator functions, The threshold is determined based on the quantile criterion of the objective function value.
6. The method according to claim 5, characterized in that, It also includes constructing and maximizing a combined sampling criterion function based on the probability density ratio to obtain the optimal parameter solution for the core mechanism parameters. : In the formula, , These are control parameters used to adjust the trade-off between exploration and exploitation.
7. The method according to claim 6, characterized in that, A median pruning strategy is also introduced to calculate the median of the objective function. : In the formula, This is the median calculation function. For the first The objective function value obtained from this calculation This is the set of objective function values that have been calculated. The calculation was terminated early when the calculated objective function value was significantly worse than the historical median, ultimately yielding the optimal parameter combination. .
8. The method according to claim 1, characterized in that, S4 specifically includes the following steps: S41. Suppose that the turbine outlet temperature calculated by the mechanistic model under given parameters is... Introducing a turbine cooling effect correction factor The turbine outlet temperature is corrected to: In the formula, For correction functions, This is a correction value for the turbine outlet temperature; S42. Determine the turbine cooling effect correction factor And set its physically feasible search range as follows: In the formula, This is the minimum value in the range of the turbine cooling effect correction factor. This represents the maximum value within the range of the turbine cooling effect correction factor. S43. The objective function is the measured actual turbine exhaust temperature. Turbine outlet temperature correction value The relative error is used as the joint optimization objective, and the relative error is defined as: The second-stage optimization problem can be written as: In the formula, This represents the optimal turbine cooling effect correction factor solution obtained in the second stage of optimization; S44. Using a Bayesian optimization framework, a TPE-based sampling strategy is employed to generate candidate samples. Perform a fixed number of calculations directly: In the formula, The feasible parameter space for the turbine cooling effect correction factor. The sampling criterion function is constructed based on a tree-structured Pareto estimator. Indicates the first Candidate values for turbine cooling effect correction factors generated in sub-Bayesian optimization calculations For historical calculation sample set, This represents the maximum number of calculations.