| Summary: | 博士 === 國立臺灣海洋大學 === 電機工程學系 === 95 === The objective of this thesis is to propose a systematic design method to enhance system diagnosablity, and to design a diagnosable system. This design method includes diagnosability analysis and diagnosability enhancement.
In prior work of diagnosability analysis, it is often to find a necessary and sufficient conduction to verify system diagnosability. The state enumeration technique was adopted, and caused state explosion. Thus this kind of method for complex off-line and on-line diagnosis can not be achieved within limited time and limited space. For this reason, the first goal of this thesis is to develop a diagnosably analysis algorithm that does not require state enumeration. In this thesis, we interpret and formulate the diagnosability problem as a binary integer linear programming problem that may have a feasible solution. By adopting the linear programming software LINGO that uses the Branch and Bound method to implement our method, we can alleviate the above-mentioned problem.
Next, in order to enhance system diagnosability, we propose an iterative systematic design algorithm. When the system is known to be non-diagnosable in the design phase, our approach tries to add sensors to enhance its diagnosablity. This design eventually establishes a diagnosis knowledge base for on-line diagnosis based on Petri nets.
In this thesis, it is assumed that the cost of a sensor is relatively low compared to the whole system. In areas as national defense, aeronautics, semiconductor equipment, the assumption is well justified. To illustrate the applicability of our approach, we use a subsystem of a Metal-Organic Vapor Phase Epitaxy (MOVPE) system.
|
|---|
