Bayesian Hierarchical Assessment of Proficiency Test Item Stability in Radiated Disturbance Proficiency Testing

In electromagnetic compatibility (EMC) proficiency testing, the stability of the proficiency test item used is of critical importance. Currently, most stability assessments for proficiency test items used in radiated emission field strength proficiency testing rely on the standard deviation of repeated measurements as the statistical method. However, this approach assumes normally distributed data and is susceptible to outliers. The main objective of this study is to propose a machine learning-based method to evaluate the stability of proficiency test items. A radiated emission field strength proficiency testing project involving 15 laboratories was conducted, during which stability tests were performed before, during, and after sample transfer. The test data were evaluated using Bayesian hierarchical analysis and compared with traditional methods such as standard deviation. The Bayesian hierarchical analysis leverages prior knowledge, improves efficiency with small sample sizes, and directly provides the probability distribution of stability parameters, thereby enhancing the reliability of radiated emission field strength proficiency testing.