| Keywords |
Bayesian inference ; sensor-linked simulation ; sensor ; steering ; emergency response |
| Abstract |
A sensor-linked fire simulation model has been developed which is capable of estimating unknown parameters by use of Bayesian Inference. The methodology is an extension of the zone model, CRISP, linking simulations to sensor measurements which govern the evolution of the parametric space in which new scenarios are generated. The model output can assist in providing fire rescue service and building occupants with an indication of the likelihood of various hazard scenarios. This paper presents the “steering” of the simulation by the sensor data, and a hypothetical example is described with a coupled model of fire development and human evacuation behaviour used to predict the locations in the building where casualties are most likely to occur. |