Diagnostic de Défauts Des Systèmes Dynamiques Par Optimisation Non Lisse PDF Download

Author: Jingwen Yang
Publisher:
ISBN:
Category :
Languages : en
Pages : 0

Get Book Here

Book Description
This thesis considers the application of nonsmooth optimization approach on several FDI problems. First, to overcome the drawback of classical methods, a nonsmooth optimization approach is proposed to solve a multiobjective fault detection problem in the worst case. An additional constraint of fast transients of residual responses could be added into the design, which could be solved by nonsmooth optimization approach. A framework of designing a unique observer gain and residual weighting matrix is proposed for a system with multiple models. When the exact model is unknown, a new framework of robust fault detection filter and an unchanged threshold are proposed. Second, a method is proposed to design an integrated fault detection observer for general case (unknown l_2 bounded faults and disturbances) and specific case (some specific faults) in frequency and time domain. The lower bound envelope is used to design a fast fault detection observer for the specific faults with a guaranteed ability of fault detection in the worst. By contrast, to decrease false alarms when fault disappearing, a constraint of an upper bound envelope is added into the design. Third, a new framework of active diagnosis with auxiliary signal is proposed. A criterion of peak amplitude is proposed to evaluate the worst effects from the auxiliary signal on the system. The effects of auxiliary signal on the outputs and control signals are considered in the design. The design is firstly shown with a case of two models, which is then extended to multiple models case.