Data-Driven Modeling for Analysis, Fault Detection, Optimization and Control of Dynamic Systems
author: time:2019-04-15 clicks:
Time and place: 2019.4.16, 14:30 pm, Wuhan National High Magnetic Field Center B206
Presenter: Jian-Qiao Sun
Title: Data-Driven Modeling for Analysis, Fault Detection, Optimization and Control of Dynamic Systems
Abstract:
In this talk, we review efforts of the past few decades in developing mathematical models for analysis, fault detection, optimization and control of various civil, mechanical and biological systems. Examples with “small” number of data are first discussed. These include the data driven modeling of acoustic materials, fatigue life prediction of metallic structures, modeling, fault detection and control of HVAC systems in office buildings, and surgical outcome prediction. The methods for data-driven research are discussed including statistical analysis, principal component analysis, correlation analysis and neural networks. We then discuss the implications of the availability of “big” data. We then review the recent advances of methods in artificial intelligence, and discuss their potentials and challenges for applications to civil and mechanical systems operating in complex environment. An example of our preliminary study of fault detection of rotor dynamic systems with deep learning is then presented to conclude the talk. It is hoped that this talk will stimulate the interests in artificial intelligence and its application to traditional engineering disciplines.