学术报告第201534期

报告题目：Advanced  real-time fault diagnosis and fault-tolerant control: from paper to  planes

报 告 人：万一鸣博士（Delft  University of Technology）

报告时间：2015年6月10  日下午  3:00

报告地点：南一楼中311室

Abstract：

Analytical  redundancy based fault diagnosis (FD) and fault-tolerant control (FTC)  techniques have been studied in academia over the past 20 years, but they have  not achieved expected maturity in the aircraft industry. The European Framework  7^th project “REconfiguration of CONtrol in Flight for Integral Global  Upset REcovery (RECONFIGURE)” combines efforts from both academia and industry,  aiming at enabling the benefits of advanced FD and FTC methods in commercial  aircrafts.

This  talk includes three different fault scenarios (air data sensor faults, stall  load on elevators, and ice accretion on the wings) investigated in the  RECONFIGURE project. We adopt moving horizon estimation (MHE) and model  predictive control (MPC) techniques, and show how their capacity of constraint  handling can 1) lead to better trade-off between sensitivity to fault and  robustness to disturbance for detecting air data sensor faults; 2) allow easy  integration between control reconfiguration and active diagnosis of elevator  faults. We also develop data-driven moving horizon unknown input estimation  methods as a component of data-driven predictive control in the icing condition.    

Estimation  and control algorithms in a flight control system need to satisfy real-time  requirements in a rather limited computing platform. The related implementation  constraints impose a grand challenge for the computationally intensive MHE and  MPC techniques. This talk also includes our preliminary efforts in bringing  nonlinear MHE into fast real-time applications.

Speaker’s  Short Bio:

Dr.  Yiming Wan received the B.S. degree from Huazhong University of Science and  Technology, Wuhan, in 2007, and the Ph.D. degree from Tsinghua University,  Beijing, in 2013, both in control engineering. Currently, he is a postdoc  researcher in Delft Center for Systems and Control, Delft University of  Technology, The Netherlands. His research interests include model-based and  data-driven fault diagnosis and fault-tolerant predictive control, with  applications in aircraft systems and chemical  processes.