Year
1982
Abstract
In a previous paper a radiation monitor was described as a typical device for the control of the movement of nuclear material. The statistical distribution of the measured data was determined, and a test procedure was described and validated experimentally, where ten data were grouped together and tested with a fixed false alarm probability. As it is needed to detect any major undeclared movement of material immediately, in this paper sequential test procedures are described and applied to radiation monitor data. In case of the first sequential procedure each observed difference between background and material movement measurement data is tested separately such that the avarage run length under the null hypothesis (no material movement) is fixed a priori. In case of the second sequential procedure (Page's test) appropriately added measured data are tested after each new measurement again for a fixed run length under the null hypothesis. Both procedures permit the immediate detection of undeclared movements of large amounts of material. The second procedure also permits the detection of longterm undeclared movements of small amounts. In this paper the experimental equipment, the measurement methods and the data processing procedures are described. The sequential tests are formulated and analysed, and their relative merits are discussed in the light of experimental results.