Realized volatility models are enhanced in this paper through the use of change-point detection and functional regression. We explore the time-varying intraday features of realized volatility in the morning and afternoon sessions in China's stock market. The empirical results reveal significant structural discontinuities in the realized volatility generated from high-frequency data in China's stock market: the coefficients of AR term and realized quarticity (RQ) term vary with time in the day and experience different patterns. From two perspectives\textemdashinformation digestion and investor behavior\textemdashwe examine how coefficients of the AR term and RQ term change over trading time with different levels of volatility. \textcopyright 2022, The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature.