Model-based reasoning using answer set programming

Diagnosis, i.e., the detection and identification of faults, provides the basis for bringing systems back to normal operation in case of a fault. Diagnosis is a very important task of our daily live, assuring safe and reliable behavior of systems. The automation of diagnosis has been a successful research topic for several decades. However, there are limitations due to complexity issues and lack of expressiveness of the underlying reasoning mechanisms. More recently logic reasoning like answer set programming has gained a lot of attention and practical use. In this paper, we tackle the question whether answer set programming can be used for automating diagnosis, focusing on industrial applications. We discuss a formalization of the diagnosis problem based on answer set programming, introduce a general framework for modeling systems, and present experimental results of an answer set programming based diagnosis algorithm. Past limitations like not being able to deal with numerical operations for modeling can be solved to some extent. The experimental results indicate that answer set programming is efficient enough for being used in diagnosis applications, providing that the underlying system is of moderate size. For digital circuits having less than 500 components, diagnosis time has been less than one second even for computing triple fault diagnoses.