Towards Identifying Useful Sorts of Knowledge and Strategies for Diagnosing Faults : A Framework for Hypothesis Generation

概要

論文の詳細を見る
As can be found in, model-based diagnosis (MBD) can be considereda method of diagnosis that ises the notjons ofstructure and behavior of the device being diagnosed to detect faults from sn observed symptom.Although stated jn slightly different terms in (~,in general, MBD can be divided into three subtasks..Hypo~esis Geseregion at which possible fault hypotheses are generated, given a symptom;Hypoihesii Tes~-ing where the fault hypotheses are tested if each of them can account for the symptom or not, and then those which can not account for the symptom wili be discarded; and H~pothesis Clessiflceiioe where the fault hypotheses that passed the test will be clsssified into an expected order. in each subtask, the problerns are: What sorts of knowledge can be used ?What kinds of strategies can be applied ?In this paper, we focus on the subtask liypothesis generation. We propose an approach to tlie subtask, providing a frainework at wliich crucial sorts of knowiedge and strategies for diagnosing faults, structurai faults as weII ss behavioral faults, are covered.
一般社団法人情報処理学会の論文
1993-03-01