Application of Fuzzy Decision Tree for the Construction of Structure Function for System Reliability Analysis Based on Uncertain Data

The reliability analysis of any system is based on the mathematical representation (mathematical model) of the investigated system. Different types of mathematical models are used in reliability engineering. The use of a particular model is determined by the specifics of the system under study and the requirements for reliability analysis (which of the reliability indices should be calculated). One of the often used models which allow using of simple mathematical methods for reliability evaluation is the structure function. The structure function determines system performance depending on its components states. But this mathematical model can be constructed if all system states depending on components states can be indicated or measured. Therefore the structure function can not be used for incompletely specified and uncertain data about system behavior. In this paper, the new method of structure function construction based on uncertain data is considered. This method uses the classification property of the structure function of dividing of all system states into a set of classes which agree with the number of system performance levels. The Fuzzy Decision Tree is used to implement system states classification based on uncertain data and transform the result of this classification into the structure function.