In-situ prediction of α-phase volume fraction in titanium alloy using laser ultrasonic with support vector regression

Abstract

A laser ultrasonic based non-contact detection method combined with support vector regression (SVR) was proposed for the first time for in-situ predicting the volume fraction of α-phase in titanium alloy (Ti-10V-2Fe-3Al). The alloy specimens were heated with incremental temperatures from 820 to 1100 °C to obtain different microstructures and volume fraction of α-phase, and the nonlinear relationship between the parameter of Rayleigh ultrasonic wave (velocity, amplitude and peak frequency) and the volume fraction of α-phase was established and analyzed by the combination of laser ultrasonic experiment and metallographic analysis. The SVR, with ultrasonic parameters and heat treatment parameters (heating and cooling rate, holding time and holding temperature) as input features, was implemented to predict the volume fraction of α-phase, achieving a relative mean error less than 5%. The results indicate that the SVR-based laser ultrasonic technique can be applied as a reliable and effective method for in-situ characterization of volume fraction of α-phase in titanium alloy.