[296664] Artigo – SEMETRO – Congresso Internacional de Metrologia Elétrica

Resumo: Anomaly detection in commercial cesium standards is crucial due to their accuracy as these standards are used on atomic time scales and in various technological applications. Anomalies or instability in cesium standards can have a negative impact on systems such as satellite navigation, military communication, network synchronization, and on time scales in many laboratories. The cesium standard is widely used as an atomic time standard due to its stability and precision, requiring continuous monitoring of its parameters to ensure signal integrity and avoid significant deviations. This study uses univariate and multivariate statistical control charts to monitor and detect anomalies in the cesium standards used in the Brazilian atomic time scale. These control charts use Hotelling’s T2 statistical test to detect anomalies and identify their onset before instability occurs in the cesium standard output signal. In addition, an exponentially weighted moving average (EWMA) plot is used to individually evaluate each parameter in the multivariate analysis of the control chart. The study confirms the abnormal behavior of the cesium standard at a specific point and successfully identifies the exact moment of the anomaly using Hotelling’s T2 and EWMA plots, before a significant stability change in the pattern occurs.