一、报告题目：Time-frequency Methods for Separation of Non-stationary Multi-component Signals
二、报告人：蒋庆堂 教授
三、时间：2025年4月14日（周一）15:30—16:30
四、地点：闻理园A4-309
报告摘要：In the real world, the majority of signals appear as multi-component signals. To streamline the processing of these signals, it is essential to extract the unknown components of the multi-component signal of interest from blind-source data. When dealing with non-stationary signals—where the frequencies of the components are functions of the time variable—no rigorous mathematical methodology existed until Ingrid Daubechies and her colleagues introduced the synchrosqueezing transform (SST) approximately a decade ago. Unfortunately, despite the extensive efforts of numerous researchers, SST is applicable only under highly restrictive conditions.
In this talk, we will explore two enhancements to SST. First, we introduce a direct time-frequency method that employs a linear chirp local approximation. This method aims to provide more accurate recovery of signal components. Second, we will discuss the chirplet transform and our recently developed time-scale-chirp_rate (TSC_R) approach. The TSC_R approach, like the chirplet transform, is capable of separating multi-component signals, even those with crossover instantaneous frequency curves. Unlike the traditional mapping of a signal into a two-dimensional space of time and scale (or frequency), TSC_R (and the chirplet transform respectively) maps a signal into a three-dimensional space encompassing time, scale (or frequency respectively), and chirp-rate.
报告人简介：蒋庆堂，浙江师范大学的杰出教授，博士生导师，于北京大学取得博士学位，并在该校数学系先后担任讲师和副教授。2002年加入密苏里大学圣路易斯分校（UMSL）之前，曾在新加坡国立大学做NSTB博士后工作，之后又担任研究员，并曾于加拿大阿尔伯塔大学和美国西弗吉尼亚大学担任访问学者。2005年至2024年，他担任UMSL数学与统计系的教授。蒋教授曾合著一部著作，并已发表90余篇论文。他当前的研究兴趣包括信号分离、图像恢复和基于深度学习的信号分类。
欢迎广大师生参加！联系人：房启全
