Theoretical Research, The Observer Method Of Delay Systems-based Robust Fault Diagnosis

With the great developments of the science and technology, there is an increasing demand for the equipments and controlled systems to be more safe, effective and reliable. The subject of fault diagnosis (including fault detection and isolation, FDI) for the complex systems is receiving more and more attention and has become one of the hot topics in the world automatic control field.Model-based fault diagnosis theory was developed in the early 1970s 20th century and a large amount of mature theoretical knowledge on model-based fault diagnosis has been accumulated during the last three decades such as: parameter estimation approach, unknown input observer approach, eigenstructure assignment approach, parity space (relation) approach, H_∞ optimization approach, strongtracking filter approach and differential geometry approach etc. Among them observer-based approach is an important one which makes full use of the mathematical model. The benefits of the observer-based fault diagnosis scheme are with high robustness, high reliability and safety. This thesis summarized the basic principle of observer-based FDI approach with examples of unknown input observer and fault detection filter (FDF) approaches and fault detection and fault isolation simulations research are carried out for the inverted pendulum system.However, most of the achievements in the FDI field are for delay-free systems. More and more attention is given to the research on time delay systems in the world automatic control field, but only few researches on FDI have been carried out for them. Time delay is an inherent characteristic of many practical systems, such as rolling mills processes, networked control systems, chemical processes, paper manufacturing industry and traffic control systems. FDI theory for the time delay systems can be applied to the practical industrial processes. So the research on rapid and effective FDI theory for the time delay systems has both important theoretical value and wide practical applicationBased on above analysis, this thesis proposed the research on the FDI design problem for several classes of linear time delay systems based on observer-based approach. A robust FDF is designed for a class of linear timedelay systems with both L2 -norm bounded unknown inputs and modeluncertainty. An observer-based FDF is introduced to generate residual and the FDF design problem is formulated into a model-matching problem so that the influence of uncertainty and disturbance is minimized to increase the robustness to unknown inputs, while the effect of fault is maximized to enhance thesensitivity to fault. Using the Hm -optimization technology, a linear matrixinequality (LMI) approach to design FDF is proposed. Generalized coordinate change is introduced for linear time delay systems such that in the new coordinates the systems could be formulated into a new form with only input and output delays and without state delay. Based on the new obtained .system expression, an observer-based FDF is considered as the residual generator, while the dynamics of the residual is changed into a general form of linear time invariant (LTI) with unknown input and fault time-delay terms. As a result, the FDF problem can be solved by applying approaches to general LTI systems. Two approaches: FDFdesign Hx optimization approach and observer-based LMI approach in LTI fieldare extended into time delay systems separately. Both of the two simulation results illustrate that the FDF approaches in the LTI field can be applied into the time delay systems effectively if the generalized coordinate change condition is met.In the end, a brief conclusion of this thesis is given- and further research in this field is expected.