The elements of hybrid electrical system diagnosis.
Thesis DisciplineElectrical Engineering
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
The subject of this thesis is the formulation of diagnostic techniques for detecting and isolating faults in hybrid electrical systems. The techniques are developed using model-variation studies. Diagnostic techniques are valuable because they permit the engineer to a priori assess the performance of a system containing a malfunction. Furthermore, by knowing a priori how a fault will affect output signals, the engineer can design programmed equipment which will automatically monitor the system signals and detect the occurrence of a malfunction. My primary overall objective in this investigation has been to formalize the approach used to develop diagnostic information for hybrid electrical systems. The approach is formulated in Chapter 1. The need for efficient diagnostic analysis methods has grown as systems have become more complex. The heuristic methods for system testing which have been applied in the past are no longer satisfactory. Diagnostic analysis covers the areas of test information generation and test procedure specification. These are complementary. Unless test information can be computed prior to the occurrence of actual system faults, there is no way of knowing for certain that the test procedures specified are effective. Applications of diagnostic analysis are to test equipment specification and diagnostic data development. The latter application is essential to all diagnostic studies and is particularly useful for compiling fault dictionaries or maintenance charts which list the likely faults, given a set of symptoms. The approach formulated in Chapter 1 is explicated in Chapters 2, 3, and 4. Techniques are developed which are useful for deriving fault detection and isolation information and for specifying efficient tests. Methods for applying diagnostic models to analogue, combinationa1 and sequential network diagnosis are developed in these chapters. During the draft stages of this thesis, each chapter was written to "stand alone". Unfortunately repetition has not been completely eliminated from this draft and some unnecessary overlap occurs in various places. In addition, the presentation is largely informal in the sense that it is devoid of mathematica1 development. Because premature mathematical formalism can sometimes obscure the simplicity of the ideas, I have used words where symbols might well have abbreviated the discussion. Owing to the original drafting method and the style of presentation, the number of pages is rather more than might be expected or desired. To the reader, I recommend that Chapter 0 be read quickly. Chapter 1 is long but the ideas are definitive and much of the material forms the background for subsequent chapters. In Chapter 2, the first part may be skimmed. It contains some of the material in Chapter 1 masticated for the analogue network digestive system. The first thirty pages can be scanned. In Chapter 3, Sections 3.2 and 3.6 - 3.7 are largely original. Chapter 4 is short and can be read rather more quickly than the previous chapters.