Frank Ortmeier, Michael Lipaczewski, Matthias Güdemann: Practical Experiences in Model-Based Safety Analysis. proceedings: International Workshop on Digital Engineering, ACM Proceedings, 2011.

Abstract

The formal modeling framework Safety Analysis and Modelling Language (SAML) allows for a combined specification of qualitative (non-deterministic) and quantitative (probabilistic) aspects. Using semantically founded model transformations, SAML models can be analyzed using current stateof- the art verification tools while guaranteeing that the analysis results of the different tools are computed on equivalent models. The significance of the results, in particular of computed quantitative occurrence probabilities heavily depends on the accuracy of the modeling. The most important factors are the temporal resolution of the model, the considered mission time for the analysis and if applicable also factors like spatial resolution. The overall results always have to be interpreted depending on the character of the system, the properties under consideration and choice of modeling parameters. In general, unfortunately, an increase in system accuracy is strongly correlated to a more complex, in particular resource intensive, analysis. In this paper we discuss qualitatively, how the dependencies between the relevenat model parameters of SAML models afflict the accuracy of the quantitative analysis results. We conduct some experiments with different parameters and survey the effects on the overall results. This work is triggered primarily with regard to optimization of such systems, where the analysis of many different system variants is necessary. In this use case, in early phases of optimization, a compromise between accuracy and running time can be used to explore many variants, identify the most promising ones and then analyze these with more accuracy.

BibTeX (Download)

@conference{IWDE2011,
title = {Practical Experiences in Model-Based Safety Analysis},
author = { Frank Ortmeier and Michael Lipaczewski and Matthias G\"{u}demann},
url = {https://www.researchgate.net/profile/Frank_Ortmeier/publication/265976291_Practical_Experiences_in_Model-Based_Safety_Analysis/links/55ae256208ae98e661a4d26c.pdf},
year  = {2011},
date = {2011-01-01},
booktitle = {proceedings: International Workshop on Digital Engineering},
publisher = {ACM Proceedings},
abstract = {The formal modeling framework Safety Analysis and Modelling Language (SAML) allows for a combined specification of qualitative (non-deterministic) and quantitative (probabilistic) aspects. Using semantically founded model transformations, SAML models can be analyzed using current stateof- the art verification tools while guaranteeing that the analysis results of the different tools are computed on equivalent models. The significance of the results, in particular of computed quantitative occurrence probabilities heavily depends on the accuracy of the modeling. The most important factors are the temporal resolution of the model, the considered mission time for the analysis and if applicable also factors like spatial resolution. The overall results always have to be interpreted depending on the character of the system, the properties under consideration and choice of modeling parameters. In general, unfortunately, an increase in system accuracy is strongly correlated to a more complex, in particular resource intensive, analysis. In this paper we discuss qualitatively, how the dependencies between the relevenat model parameters of SAML models afflict the accuracy of the quantitative analysis results. We conduct some experiments with different parameters and survey the effects on the overall results. This work is triggered primarily with regard to optimization of such systems, where the analysis of many different system variants is necessary. In this use case, in early phases of optimization, a compromise between accuracy and running time can be used to explore many variants, identify the most promising ones and then analyze these with more accuracy.},
keywords = {model-based, safety, safety analysis, SAML},
pubstate = {published},
tppubtype = {conference}
}