Using Machine Learning to Minimize User Intervention in Theorem Proving based Dynamic Fault Tree Analysis

Dynamic fault trees (DFTs) have become one of the commonly used modeling techniques that capture the dynamic failure behavior of systems. Recently, DFTs have been formalized in higher-order logic (HOL), which allows performing DFT analysis within the sound core of a HOL theorem prover. However, due to the interactive nature of HOL theorem proving, the proof process involves significant user guidance. In this paper, we propose to use machine learning techniques to facilitate automating the proof of the subgoals. The machine learning can use the existing proofs of these goals as well as the verification steps being performed at runtime to come up with reasoning to verify the remaining subgoals. This kind of support from machine learning can lead to the creation of a tool for DFT analysis that requires minimum user intervention in the formal DFT analysis and thus can facilitate the industry to benefit from a sound DFT analysis approach.