Anwendung der indefinit quadratischen Optimierungsmethoden zur Lösung der Fehlererkennungsprobleme für lineare zeitvariante Systeme

• Applicant: Professor Dr.-Ing. Steven Xianchun Ding
• Subject Area: Automation, Mechatronics, Control Systems, Intelligent Technical Systems, Robotics
• Term: from 2010 to 2013
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 144939979

• Final Report Year: 2013

Final Report Abstract

The overall goal of this project is to develop new control theoretical methods for linear time-varying (LTV) systems and new FD (fault detection) methodologies by using the indefinite quadratic optimization theory. There are four research objectives formulated in the propose: • integrated design of observer-based fault detection systems for LTV processes with unknown inputs (disturbances) in the framework of the indefinite quadratic optimization theory; • application of the projection technique to solving the integrated design problems in a recursive computation algorithm; • formulation and solution of fault detection problems in terms of a trade-off between the FAR (false alarm rate) and FDR (fault detection rate) for LTV processes and finally • development of fault detection schemes for LTV systems with model uncertainties. In order to demonstrate and illustrate the achieved theoretical results, the major results should be implemented and tested on a laboratory wireless NCS (networked control system) platform. The major results achieved in this project are summarized as follows: • An integrated design scheme is proposed for fault detection in LTV processes. Different from the existing FD schemes, FAR and FDR are defined based on online estimation of the disturbances and faults and their computations are integrated into the core of the FD system, which allows an optimized trade-off between FAR and FDR and thus improves the FD performance in comparison with the existing methods considerably. • With the aid of the Krein space technique, H2- and H∞-fault and disturbance estimation problems are solved for LTV processes with (deterministic) unknown inputs. Recursive algorithms have been developed, which allow the integration of the fault and disturbance estimations into the FD system. This is the first FD scheme with integrated computations of FAR and FDR and thus offers high FD performance. • Uncertainties like changes in system matrices and delays in measurements have been addressed and the achieved results have been integrated into the FD scheme. • The FD scheme and the major algorithms have been successfully tested and demonstrated on the laboratory wireless NCS platform WiNC. In this project, an advanced design framework has been established, based on which the following research work in the future promises new scientific contributions: • Integrated design of FD systems for hybrid processes, including multiple mode systems, switching (or jump) systems; • FD for time-varying nonlinear processes; • applications in designing fault-tolerant control systems. The most potential application is the design and optimization of fault detection systems for the industrial NCSs (including wireless NCSs) and their application to fault-tolerant control systems, as demonstrated by the case study on the laboratory system WiNC-platform. Moreover, LVT systems can be found e.g. in aerospace industry, in chemical industrial systems, in mechatronic and transport systems as well as in power systems. In all these systems, safety, reliability and availability are fundamental requirements for the system design, implementation and operation. The new FD methods and design framework for LTV systems developed in this project can be fully applied for this objective.

Publications

• Integrated design of fault tolerant networked control systems, the 8th IFAC Symposium on Fault Detection, Supervision and Safety of Technical Processes, SafeProcess 2012, Mexico City, Mexico, 2012
  Y. Wang, S. X. Ding, D. Xu, S. Yin, B. Shen and P. Zhang
  
• A projection-based method of fault detection for linear discrete time-varying systems, International Journal of Systems Science, 44(5): 820-830, May. 2013
  M. Zhong and S. X. Ding
  
• Finite-horizon H∞ fault estimation for linear discrete time-varying systems with delayed measurements, Automatica, 49(1): 293- 296, Jan. 2013
  B. Shen, S. X. Ding and Z. Wang
  
• Quadratic optimal fault tolerant control on wireless networked control system, IEEE ICCA 2013, Hangzhou, China, Jun. 12-14, 2013
  D. Xu, S. X. Ding, Y. Wang and B. Shen