Rapid Parameter Inference with Uncertainty Quantification for a Radiological Plume Source Identification Problem
Christopher Edwards, Ralph C Smith
TL;DR
This work tackles real-time localization of instantaneous radiological releases using downwind detector arrays by comparing regression, categorical classification, and Bayesian neural networks against DRAM-based MCMC for uncertainty quantification. A 2-D advection–diffusion dispersion model and a Poisson sensor model generate training data (up to 400{,}000 samples) enabling fast inference of release parameters $x_c$, $y_c$, and $m_c$ with uncertainty estimates. The Bayesian NN provides posterior densities via weight distributions and sampling, the classification NN offers probabilistic localization, and the regression NN delivers fast point estimates; all NN approaches are significantly faster than DRAM, with densities from the Bayesian NN approaching DRAM-like uncertainty. The results demonstrate that real-time, uncertainty-aware inference is feasible for radiological plume source identification, with extensions to 3-D, multi-time data, and urban domains proposed for future work.
Abstract
In the event of a nuclear accident, or the detonation of a radiological dispersal device, quickly locating the source of the accident or blast is important for emergency response and environmental decontamination. At a specified time after a simulated instantaneous release of an aerosolized radioactive contaminant, measurements are recorded downwind from an array of radiation sensors. Neural networks are employed to infer the source release parameters in an accurate and rapid manner using sensor and mean wind speed data. We consider two neural network constructions that quantify the uncertainty of the predicted values; a categorical classification neural network and a Bayesian neural network. With the categorical classification neural network, we partition the spatial domain and treat each partition as a separate class for which we estimate the probability that it contains the true source location. In a Bayesian neural network, the weights and biases have a distribution rather than a single optimal value. With each evaluation, these distributions are sampled, yielding a different prediction with each evaluation. The trained Bayesian neural network is thus evaluated to construct posterior densities for the release parameters. Results are compared to Markov chain Monte Carlo (MCMC) results found using the Delayed Rejection Adaptive Metropolis Algorithm. The Bayesian neural network approach is generally much cheaper computationally than the MCMC approach as it relies on the computational cost of the neural network evaluation to generate posterior densities as opposed to the MCMC approach which depends on the computational expense of the transport and radiation detection models.