Turbine gas thermal performance uncertainty quantification method and system based on universal Kriging model

Abstract

A turbine gas thermal performance uncertainty quantification method based on a universal Kriging model comprises the following steps: generating a to-be-solved polynomial chaos expansion through a polynomial chaos theory, and generating to-be-calculated sparse / dense sample point data based on a low-order / high-order Symolak sparse grid technology; using a genetic algorithm to automatically plan the calculation sequence of all sparse samples, and obtaining gas heat parameters of all the samples through multi-machine different-place asynchronous distributed calculation; solving the coefficient of the polynomial chaos expansion, using the obtained explicit expression as a regression function of a generic Kriging model building module to construct a generic Kriging model, and solving the expression of the generic Kriging model; calculating gas heat parameters of each dense sample point through an expression of the universal Kriging model; and solving the coefficient of the polynomial chaos expansion by using a Galerkin projection method, the uncertainty mean value and deviation of the turbine gas heat parameters can be obtained, and the sample size of the polynomial chaos method in turbine gas heat performance uncertainty quantitative calculation can be reduced.

Description

technical field [0001] The invention belongs to the technical field of turbine uncertainty quantification design, in particular to a method and system for quantifying uncertainty of turbine gas-thermal performance based on a universal Kriging model. Background technique [0002] At present, the mainstream research fields of turbine gas thermal performance at home and abroad are all in the framework of deterministic research. However, there are many uncertainties in engineering practice. For example, the depth of the turbine groove will be randomly distributed due to manufacturing errors, and the actual working conditions of the turbine, such as the total inlet pressure, also have certain uncertainties. According to the research of D'Ammaro et al. (D'Ammaro A,Montomoli F.Uncertainty quantification and film cooling[J].Computers&Fluids,2013,71:320-326.), these geometric and operating condition deviations will significantly change the The shape of the flow field thus affects it...

AI Technical Summary

Problems solved by technology

However, in practice, researchers found that the Monte Carlo method needs to conduct uncertainty quantitative analysis through a large number of random samples, which consumes a lot of computing resources.

Although the polynomial chaotic method can achieve the same calculation accuracy with less sample size than the Monte Carlo method, the polynomial chaotic method still needs a certain number of samples for uncertainty quantification calculation.

For example, using the polynomial chaos method to study a 3-dimensional problem requires 64 samples, and it takes about two months to calculate these samples using the mainstream Inter core server. Such a large calculation expenditure is not allowed for engineering design.

And as the dimension of the research problem increases, the polynomial chaos method will encounter the so-called disaster of dimensionality, that is, the required sample size will increase exponentially with the increase of the dimension of the research problem.

Images