Search

CN-122020881-A - Blade profile optimization method and device based on Gaussian process and electronic equipment

CN122020881ACN 122020881 ACN122020881 ACN 122020881ACN-122020881-A

Abstract

The disclosure provides a blade profile optimization method, a blade profile optimization device and electronic equipment based on a Gaussian process, wherein a target Gaussian process model corresponding to aerodynamic parameters is built based on preset blade design parameters and corresponding preset aerodynamic parameters, initial blade design parameters are selected in a design range, the initial aerodynamic parameters are predicted based on the target Gaussian process model, multiple groups of initial solutions corresponding to the initial blade design parameters and the initial aerodynamic parameters are obtained, the multiple groups of initial solutions are optimized based on a multi-target optimization model, and an optimal solution is obtained, wherein the optimal solution comprises optimal blade design parameters and optimal aerodynamic parameters. According to the blade profile optimization method, device and electronic equipment based on the Gaussian process, through rationalizing sampling of the training samples and the test samples, the proxy model is quickly constructed, the target parameters are quickly predicted, then multi-target optimization is performed, the blade profile design can be quickly obtained through iterative optimization, and the robustness and the accuracy of the blade profile optimization are improved.

Inventors

  • FU WENHAO

Assignees

  • 宁波方太厨具有限公司

Dates

Publication Date
20260512
Application Date
20260109

Claims (10)

  1. 1. A blade profile optimizing method based on Gaussian process is characterized in that, Constructing a target Gaussian process model corresponding to the aerodynamic parameters based on the preset blade design parameters and the corresponding preset aerodynamic parameters; selecting initial blade design parameters in a design range, and predicting initial aerodynamic parameters based on the target Gaussian process model to obtain a plurality of groups of initial solutions, wherein the initial solutions comprise the initial blade design parameters and the initial aerodynamic parameters corresponding to the initial blade design parameters; And optimizing a plurality of groups of initial solutions based on the multi-objective optimization model to obtain an optimal solution, wherein the optimal solution comprises optimal blade design parameters and optimal pneumatic parameters.
  2. 2. A blade profile optimization method as defined in claim 1, wherein the constructing a target Gaussian process model of the corresponding aerodynamic parameters based on the preset blade design parameters and the corresponding aerodynamic parameters comprises, Selecting preset blade design parameters in a preset range, sampling based on the preset blade design parameters, and performing simulation to obtain corresponding preset pneumatic parameters; Constructing an initial Gaussian process model based on the preset blade design parameters and the preset aerodynamic parameters, and constructing a training sample and a testing sample based on the preset blade design parameters and the preset aerodynamic parameters; And training the initial Gaussian process model based on the training sample and the test sample to obtain a target Gaussian process model.
  3. 3. The blade profile optimizing method as claimed in claim 2, wherein the selecting of the predetermined blade design parameters within the predetermined range, sampling based on the predetermined blade design parameters and performing simulation to obtain the corresponding predetermined aerodynamic parameters, specifically comprises, Latin hypercube sampling is carried out within a preset range to obtain preset blade design parameters, and simulation is carried out on the preset blade design parameters to obtain corresponding preset pneumatic parameters; And/or the number of the groups of groups, Training the initial Gaussian process model based on the training sample and the testing sample to obtain a target Gaussian process model, wherein the method specifically comprises the following steps of, After the sample with the maximum uncertainty is obtained for the first time in each iteration, the new sample is directly put into a covariance matrix of a Gaussian process, optimization is not performed in a training mode, and then sampling is performed again.
  4. 4. The method of optimizing a leaf pattern of claim 1 wherein said optimizing initial aerodynamic parameters based on a multi-objective optimization model determines an optimal solution from a plurality of sets of initial solutions, comprising, And (3) taking the pneumatic parameters as targets, taking the design parameters and the flow parameters as constraints, performing multi-target parameter optimization, and determining an optimal solution from a plurality of groups of initial solutions.
  5. 5. A method of profile optimization as in claim 4, wherein the flow parameter is a preset parameter, or And selecting initial blade design parameters in a design range, and predicting flow parameters based on a correspondingly constructed target Gaussian process model to obtain the target Gaussian process model.
  6. 6. A method of optimizing a blade profile as in claim 4, wherein the design parameters include at least one of leading and trailing edge radii, inlet and outlet geometry angles, inlet upper wedge angle, inlet lower wedge angle, axial chord length, effective gas outlet angle, and outlet deflection angle; The aerodynamic parameters comprise at least one of flow, full pressure efficiency, load factor, total pressure loss factor and outlet air flow angle; the flow parameters include at least one of inlet total pressure, inlet airflow angle, turbulence, and outlet static pressure.
  7. 7. A profile optimizing apparatus, comprising: the model construction module is used for constructing a target Gaussian process model corresponding to the pneumatic parameters based on the preset blade design parameters and the corresponding preset pneumatic parameters; The initial solution acquisition module is used for selecting initial blade design parameters in a design range, predicting initial aerodynamic parameters based on the target Gaussian process model, and obtaining a plurality of groups of initial solutions corresponding to the initial blade design parameters and the initial aerodynamic parameters; And the optimization module is used for optimizing the initial aerodynamic parameters based on the multi-objective optimization model, and determining an optimal solution from a plurality of groups of initial solutions, wherein the optimal solution comprises optimal blade design parameters and optimal aerodynamic parameters.
  8. 8. An electronic device comprising a memory, a processor and a computer program stored on the memory for execution on the processor, characterized in that the processor implements the blade profile optimization method of any one of claims 1 to 6 when executing the computer program.
  9. 9. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the leaf profile optimization method according to any one of claims 1 to 6.
  10. 10. A computer program product comprising a computer program which, when executed by a processor, implements the leaf profile optimization method according to any one of claims 1-6.

Description

Blade profile optimization method and device based on Gaussian process and electronic equipment Technical Field The disclosure relates to the field of fans, and in particular to a fan blade profile optimization method. Background In the fan blade profile optimization process, the blade profile is generated through parametric modeling, and according to the actual simulation or test result, the traditional optimization method needs to rely on experience of a designer to adjust part of parameters of the blade profile so as to obtain the blade profile with better performance, and the method often needs longer iteration time. With the development of technology, many researchers begin to adopt intelligent algorithms to optimize leaf type parameters, such as genetic algorithms, gradient accompanying methods, proxy model methods and the like. With the development of machine learning algorithms, machine learning is also applied to proxy model construction, such as support vector machine regression (SVR), artificial Neural Network (ANN), and Gaussian Process (GP). The traditional blade profile optimization method relies on manual experience, has long design iteration time and low efficiency, and is difficult to find out the parameter with optimal performance. Once an accurate model is built, the agent model method can rapidly predict the target, but in practical application, a plurality of design parameters are often required, so that the agent model is provided with a problem of dimension explosion, and the model training difficulty is high. Disclosure of Invention The technical problem to be solved by the present disclosure is to overcome the defects in the prior art, and provide a leaf profile optimization method with convenient sample acquisition, and high accuracy and high robustness by rapidly iterating a model. The technical problems are solved by the technical proposal that a blade profile optimizing method based on a Gaussian process, Constructing a target Gaussian process model corresponding to the aerodynamic parameters based on the preset blade design parameters and the corresponding preset aerodynamic parameters; selecting initial blade design parameters in a design range, and predicting initial aerodynamic parameters based on the target Gaussian process model to obtain a plurality of groups of initial solutions corresponding to the initial blade design parameters and the initial aerodynamic parameters; And optimizing the initial aerodynamic parameters based on the multi-objective optimization model, and determining an optimal solution from a plurality of groups of initial solutions, wherein the optimal solution comprises optimal blade design parameters and optimal aerodynamic parameters. Optionally, the constructing a target gaussian process model of the corresponding aerodynamic parameters based on the preset blade design parameters and the corresponding aerodynamic parameters, specifically comprises, Selecting preset blade design parameters in a preset range, sampling based on the preset blade design parameters, and performing simulation to obtain corresponding preset pneumatic parameters; Constructing an initial Gaussian process model based on the preset blade design parameters and the preset aerodynamic parameters, and constructing a training sample and a testing sample based on the preset blade design parameters and the preset aerodynamic parameters; And training the initial Gaussian process model based on the training sample and the test sample to obtain a target Gaussian process model. Optionally, selecting preset blade design parameters within a preset range, sampling based on the preset blade design parameters, and performing simulation to obtain corresponding preset aerodynamic parameters, Latin hypercube sampling is carried out within a preset range to obtain preset blade design parameters, and simulation is carried out on the preset blade design parameters to obtain corresponding preset pneumatic parameters; And/or the number of the groups of groups, Training the initial Gaussian process model based on the training sample and the testing sample to obtain a target Gaussian process model, wherein the method specifically comprises the following steps of, After the sample with the maximum uncertainty is obtained for the first time in each iteration, the new sample is directly put into a covariance matrix of a Gaussian process, optimization is not performed in a training mode, and then sampling is performed again. Optionally, the optimizing the initial aerodynamic parameters based on the multi-objective optimizing model determines an optimal solution from a plurality of groups of initial solutions, specifically including, And (3) taking the pneumatic parameters as targets, taking the design parameters and the flow parameters as constraints, performing multi-target parameter optimization, and determining an optimal solution from a plurality of groups of initial solutions. Optionally, the flow para