On the morning of 13 September 2026, the National Rail Transit Electrification and Automation Engineering Technology Research Center (Hong Kong Branch) (CNERC-Rail) at The Hong Kong Polytechnic University (PolyU) held a research seminar at PolyU Z414. Prof. Ying Lei from the School of Architecture and Civil Engineering at Xiamen University was invited to deliver a seminar entitled “Some Progresses on Deep Learning Enhanced Missing Data Imputation and KF-UI Algorithms.” Prof. Ying Lei’s research interests include structural health monitoring, system identification, structural control, and stochastic dynamics. He has authored more than 300 academic papers and received the National Natural Science Award of China and other honours.
The seminar focused on the application of deep learning to structural monitoring data processing and system identification. Prof. Ying Lei first introduced physics-guided unsupervised deep learning methods for random and continuous missing data imputation, including simultaneous continuous missing data, multi-mode missing data, and missing data with consistent probability distributions. He then presented recent developments in deep learning-enhanced KF-UI algorithms, covering self-adaptive estimation of process and measurement noise covariance matrices, pseudo-measurement-assisted state-input identification, and in-field identification of cable MR damper and multi-damper failures under unknown wind loads.
The seminar highlighted recent advances in integrating deep learning with state estimation and system identification techniques, providing new perspectives for missing data processing, state-input identification, and fault diagnosis in structural health monitoring.
