Foundations for model‐based systems engineering and model‐based safety assessment

Abstract

This article is a contribution to the INCOSE initiative for model-based engineering transformation.Its material has been presented at the ALSEE tour event in Oslo in September 2016. The ideas developed here come from the practical and theoretical experience of the authors in both industrial and academic frameworks. We organize the discussion around 6 theses that aim at establishing robust conceptual foundations for the model-based engineering transformation. We focus on model-based systems engineering, model-based safety assessment, and the relationship between these 2 disciplines. We report on active research initiatives that implement these 6 theses via the S2ML+X paradigm. We conclude with suggestions about future research and teaching activities. K E Y W O R D S model-based safety assessment, model-based systems engineering 146

Modeling and structuring design rationale to enable knowledge reuse

“…The knowledge generated during this process is referred to as “design rationale” and these are recorded as test reports. These reports inform future engineers on the evolution of design space and prevent them from repeating the errors committed in the past . (p146) Design rationale could serve as a means to train learning algorithms that can predict the performance of a system for a given choice of design/environment variables.…”

“…We provide an overview of some existing failure assessment models—refer to Rauzy and Haskins for a comprehensive review. One category of failure models stems from a simple causal chain of events, later developed into approaches such as Fault Tree Analysis (FTA), Fishbone diagram, Failure Modes Effect Analysis (FMEA), and Fuzzy‐FTA .…”

“…Most of these safety models are event‐driven, and they are categorized as pragmatic and formal models . (p149) Pragmatic models such as BMPN, SysML (eg, Relation‐based modeling for static properties (RMSP) proposed by Chen et al), and OPM do not use abstract elements—all the elements in the model are case‐specific.…”

“…(p149) Pragmatic models such as BMPN, SysML (eg, Relation‐based modeling for static properties (RMSP) proposed by Chen et al), and OPM do not use abstract elements—all the elements in the model are case‐specific. Formal models include a mathematical framework to describe a system's behavior and structuring paradigm to organize the model, which is generally absent in pragmatic models . (p148)…”

“…(p462) These are grounded in the state transition of the system, which is explained only using physical laws . (p463) Rauzy and Haskins (p151) state the following, “The most widely used modeling formalisms for safety analyses lack either expressiveness, for example, fault trees and event trees, or structure, for example, Markov chains and stochastic Petri nets. Consequently, they are far from system specifications.”…”