挙動・事象因果の定量的表現に基づく製品の不具合検出

Abstract

With recent globalization in industries, the number of failures and troubles of products caused by using them in unexpected ways has increased. In order to avoid such troubles, it is necessary not only to assume various ways of use thoroughly, but also to verify whether the design plan can fulfill required functions when the product is utilized in those ways. From this point of view, the authors proposed a functional verification method considering ways of use based on qualitative modeling of behavior of entities and cause-and-effect relationships among physical phenomena using Petri nets. It is, however, impossible to detect failures concerning to specification which requires dealing with quantitative information. This paper provides a method for quantitative modeling of behavior of entities and cause-and-effect relationships among physical phenomena. Two types of tokens were defined for dealing with positive and negative values and four types of arcs were for controlling changes of those values. These new elements of Petri net made it possible to represent behavior of entities and cause-and-effect relationships quantitatively. Application of this new modeling method to the functional detection method enables automatic detection of failures concerning to both functions and specification. The detection method using this modeling method was applied to an example, and its effectiveness was proven.