Optimal estimates of inventory and losses using nuclear material accountancy data

State-estimation theory has been shown to be applicable in obtaining minimum variance unbiased estimates of inventory and loss of special nuclear materials. The loss estimates are far superior to those obtained by the ID/LEID approach and the CUSUM approach. The technique may be extended to other situations where more complex models are appropriate. For example, one can add additional state variables to the system model to represent holdup at various points in the MBA. One could also model several interactive unit processes. So long as the system can be described by a linear system model, it is possible to construct a Kalman Filter and Linear Smoother to obtain minimum variance unbiased estimates of the state variables. (author)