Nonlinear dynamic probabilistic design of turbine disk-radial deformation using extremum response surface method-based support vector machine of regression

Abstract

In order to improve the computational efficiency of nonlinear dynamic probabilistic design for aeroengine typical components, a probabilistic design method–extremum response surface method-based support vector machine of regression was proposed. By taking support vector machine of regression as an extremum response surface function, the mathematical model of surface method-based support vector machine of regression was established. The probabilistic design of turbine disk-radial deformation was accomplished based on the surface method-based support vector machine of regression fully considering the influences of the nonlinearity of material property and the dynamic of heat load and mechanical load. The analysis results show that the probabilistic distribution and inverse probabilistic features of input–output parameters and the major factors (rotor speed and gas temperature) are gained legitimately, which provide the useful reference for disk design and blade-tip clearance control more effective of high-pressure turbine). Through the comparison of methods, surface method-based support vector machine of regression is demonstrated to hold high efficiency and high precision in nonlinear dynamic probabilistic design of aeroengine typical components. Moreover, the proposed surface method-based support vector machine of regression is promising to provide a useful insight for disk dynamic optimal design and blade-tip clearance control of aeroengine high-pressure turbine.