34 Part I: Data-driven Inference of Fault Tree models I.2 Nomenclature Fault Trees: F Fault Tree model BE Basic Event (BE) AND Logic gate AND OR Logic gate OR V Set of nodes in a Fault Tree G Set of logic gates in a Fault Tree TE Top Event b Status vector C Minimal Cut Set (MCS) Inference of Fault Trees: D Failure dataset FD Inferred FT fromD MD MCS matrix obtained fromD M F MCS matrix obtained from a given F ωs Size of the Fault Tree model ωd Error computed fromD ωc Error computed between M F and MD ε Number of superfluous BEs Multi-Objective Evolutionary Algorithms: ps Population size uc Max. generations with unchanged best ng Max. number of generations I.3 Related work Di!erent methods for inferring Fault Trees (FTs) have been discussed in the literature. We categorise these methods into three main groups: knowledge-based, model-based, and data-driven. Knowledge-based approaches primarily use di!erent heuristics for knowledge representation and domain expertise (Latif-Shabgahi, 2002); model-based approaches translate existing system models and/or graphs into FTs; and data-driven approaches use structured databases as the main information source, aiming to identify causal relationships in a failure dataset with minimal domain expertise and human intervention. Carpignano and Poucet, 1994 provides a comprehensive review of knowledge-based approaches. An example of a model-based approach is found in Mhenni, Nguyen, and Choley, 2014, where the authors employ SysML System Models to derive FT models. However, a significant limitation of model-based approaches is the need for a pre-existing model (Dickerson, Roslan, and Ji, 2018). Our focus is on data-driven approaches, which encompass the applications of machine learning techniques and
RkJQdWJsaXNoZXIy MjY0ODMw