A new step towards the integration of probabilistic μ in the aerospace V V process

Probabilistic μ -analysis was introduced 20 years ago as a control system validation means able to quantify the probability of rare and potentially critical events. However, for a long time, no practical tool offering both good reliability and reasonable computational time was available, making this technique hardly usable in an industrial context. The STOchastic Worst-case Analysis Toolbox (STOWAT) was introduced a few years ago to bridge this gap between theory and practice. It has been significantly improved since then, thanks to the addition of new features, but above all to increasingly efficient implementations, resulting in a dramatic reduction in CPU time. However, until recently, it could only be applied to small-scale models, with up to 4 or 5 uncertainties. In the perspective of analyzing systems with a larger number of uncertain parameters, a time-consuming and tedious process was carried out. This led to a complete rewrite of the STOWAT, which is now optimized down to the sub-function level, and whose performance is assessed in this paper on a series on benchmarks of increasing complexity with up to about 20 states and 20 uncertainties. This work represents a new step towards the development of a consolidated tool that could reasonably be integrated in the aerospace Verification and Validation process in a near future, finding its place between Monte Carlo simulations – useful for quantifying the probability of sufficiently frequent phenomena – and worst-case μ -analysis – relevant for detecting extremely rare events.