New approach to safeguards accounting. [Kalman filter method for detecting continual small losses

Abstract

In recent years there has been widespread concern over the problem of nuclear safeguards. Due to the proliferation of nuclear reactors throughout the world, the concern about the loss or diversion of nuclear materials at various points in the fuel cycle has greatly increased. To minimize the possibility of material loss, the nuclear industry relies on physical protection systems and materials accountability procedures at licensed facilities. Present techniques of material accountability rely on double-entry accounting systems. For various reasons, only noisy observations of on-hand inventory are available. Hence one is forced to use statistical techniques in an attempt to detect the existence of missing material. Current practice is to use control charts as the basis for detecting significant material losses. Control charts may aid in detecting large material losses but are insensitive to small quantities of material loss, even if these small losses occur repeatedly over a long period of time. The purpose of this research is to show the feasibility of using linear state estimation theory in nuclear material accountability. The Kalman Filter is used as the state estimation technique. The state vector which consists of on-hand inventory and material losses is estimated recursively.