Method for Detecting Material High-Temperature Phase Transition Dynamic Region Based on Dynamic Event Coding

A dynamic detection framework based on dynamic event coding is proposed to detect the solidification precipitation and growth of Al3Ni intermetallic compounds at Al/Ni-liquid/solid interface during heating process. This method uses the Mask RCNN model to accurately detect the material phase transition region, and uses a dynamic event coding strategy to explicitly capture mutational events such as splitting and merging of the Al3Ni intermetallic compound region, achieving accurate analysis of the microstructure during the solidification phase transition process at the liquid/solid interface. Accurate analysis of organizational dynamic evolution (such as sudden behaviors such as splitting and merging). This method can effectively deal with the complex dynamic changes in the data set of Al/Ni-liquid/solid interface solidification process, solve the problems of error accumulation and neglect of space-time dependence in traditional methods, and provide a new idea for data processing and quantitative characterization of liquid/solid interface microstructure evolution dynamic detection process.