近期，我院本科生林子轩（第一作者），教师范伟（通讯作者）等的研究成果Multirate ensemble probabilistic sparse identification of nonlinear dynamics for industrial process monitoring在人工智能领域TOP期刊《Engineering Applications of Artificial Intelligence》（IF=8）上发表
论文简介如下：
Modern industrial processes face a core challenge posed by heterogeneous data collected at multiple sampling rates with nonlinear dynamics. Traditional single-rate process monitoring methods often experience significant performance degradation when applied to such multirate systems. This paper proposes a novel probabilistic sparse identification of nonlinear dynamics method for multirate industrial process monitoring (ME-PSINDy). ME-PSINDy effectively integrates multirate observation data into a unified probabilistic state-space framework. By utilizing an ensemble strategy within the framework of Probabilistic Sparse Identification of Nonlinear Dynamics (PSINDy), the approach achieves enhanced robustness. The method captures dominant nonlinear dynamic patterns in a compact sparse form, which improves model robustness and transparency for reliable fault detection and diagnosis. Validation on a controlled synthetic multirate nonlinear benchmark and two real-world industrial cases demonstrates that ME-PSINDy yields superior fault detection rates and enhanced model robustness when handling multirate nonlinear systems compared to existing monitoring methods.
