Application of classical versus bayesian statistical control charts to on-line radiological monitoring
- 134 Downloads
False positive and false negative incidence rates of radiological monitoring data from classical and Bayesian statistical process control chart techniques are compared. The on-line monitoring for illicit radioactive material with no false positives or false negatives is the goal of homeland security monitoring, but is unrealistic. However, statistical fluctuations in the detector signal, short detection times, large source to detector distances, and shielding effects make distinguishing between a radiation source and natural background particularly difficult. Experimental time series data were collected using a 1″ × 1″ LaCl3(Ce) based scintillation detector (Scionix, Orlando, FL) under various simulated conditions. Experimental parameters include radionuclide (gamma-ray) energy, activity, density thickness (source to detector distance and shielding), time, and temperature. All statistical algorithms were developed using MATLAB™. The Shewhart (3-σ) control chart and the cumulative sum (CUSUM) control chart are the classical procedures adopted, while the Bayesian technique is the Shiryayev–Roberts (S–R) control chart. The Shiryayev–Roberts method was the best method for controlling the number of false positive detects, followed by the CUSUM method. However, The Shiryayev–Roberts method, used without modification, resulted in one of the highest false negative incidence rates independent of the signal strength. Modification of The Shiryayev–Roberts statistical analysis method reduced the number of false negatives, but resulted in an increase in the false positive incidence rate.
KeywordsCUSUM Shiryayev–Roberts Shewhart Gamma-ray monitoring
The work was funded under a DOE NNSA SBIR Phase II grant through ADA Technologies, Inc.
- 1.Walpole RE, Myers H (1998) Probability and statistics for engineers and scientists, 1st edn. Macmillan Publishing, New YorkGoogle Scholar
- 2.Montgomery DC (2001) Introduction to statistical quality control, 4th edn. Wiley, New YorkGoogle Scholar
- 4.Chen C (1999) Monitoring and detecting shifts in the mean in quality levels for a production environment with properties found in the geometric Poisson process. Thesis, Texas Tech University, Lubbock (TX), p 204Google Scholar
- 5.Scandol JP (2003) Fish Res 64:19Google Scholar
- 8.Kenett R, Zacks S (1998) Modern industrial statistics. Duxbury Press, Pacific GroveGoogle Scholar
- 9.Dong X (1999) Contributions to statistical process control and robust parameter design. Thesis, University of Michigan, Ann Arbor, MI, p 91Google Scholar
- 11.Miller G, Inkret WC, Martz HF (1993) Radiat Prot Dosim 48:251Google Scholar
- 14.Meyer PD, Ye M, Neuman SP, Cantrell KJ (2004) Combined estimation of hydrogeologic conceptual model and parameter uncertainty, Pacific Northwest National Laboratory NUREG/CR-6843, April 2004Google Scholar
- 16.Lai TL (1995) J Royal Stat Soc B 57:613Google Scholar
- 17.Wu Y (1991) Some contributions to online quality control. Thesis, University of Toronto, p 137Google Scholar