Sequential Monte Carlo simulation for the estimation of small reachability probabilities for stochastic hybrid systems

Krystul, J. and Blom, H.A.P. (2006) Sequential Monte Carlo simulation for the estimation of small reachability probabilities for stochastic hybrid systems. In: 2006 Second International Symposium on Communications, Control and Signal Processing, 13-15 Mar 2006, Marrakech, Morocco.

PDF
Restricted to UT campus only: Request a copy
168Kb

Abstract:	The problem of estimating the probability that a system reaches a given set within some time horizon is considered. Standard Monte Carlo methods for reachability probability estimation do not require specific assumptions on the system under consideration. However, they are computationally demanding when the probability to be estimated is small. An Interacting Particle System (IPS) approach has been developed for the estimation of small reachability probabilities for diffusion processes. IPS has then been extended so as to estimate small reachability probabilities for a certain family of stochastic hybrid processes, namely switching diffusions. This contribution further improves the hybrid IPS method by adopting an importance sampling approach that uses the interaction equations characterizing stochastic hybrid systems.
Item Type:	Conference or Workshop Item
Faculty:	Electrical Engineering, Mathematics and Computer Science (EEMCS)
Research Group:	Stochastic Systems and Signals (SST)
Link to this item:	http://purl.utwente.nl/publications/63930
Export this item as:	BibTeX EndNote HTML Citation Reference Manager

Show download statistics for this publication

Repository Staff Only: item control page

Metis ID: 237932