Automatic Detection of Anomalies in Post-Processed Data Applied to UTC Time Transfer Links.

Area of Science:

• Data Science
• Metrology
• Signal Processing

Background:

• Data anomalies can compromise system reliability and accuracy.
• Accurate anomaly detection is crucial for understanding data integrity.
• Existing methods may lead to unnecessary deletion of valuable data.

Purpose of the Study:

• To present a novel Kalman filter-based method for identifying anomalies in time-step data.
• To retain maximum data while detecting anomalies, avoiding the deletion of valuable information.
• To accurately determine the occurrence and magnitude of anomalies, focusing on time steps.

Main Methods:

• Utilizing a Kalman filter optimized for post-processed data.
• Developing an algorithm to detect anomalies without discarding significant data points.
• Applying the method to time links in Coordinated Universal Time (UTC) data from BIPM.

Main Results:

• The developed tool effectively identifies anomalies in time-step data.
• The method successfully avoids the unnecessary deletion of valuable data.
• The algorithm accurately determines the date of occurrence and magnitude of anomalies.

Conclusions:

• The novel Kalman filter approach enhances data validation and problem identification for systems like UTC.
• This method improves the reliability and accuracy of critical timekeeping systems.
• Rapid anomaly detection is essential for maintaining the integrity of Coordinated Universal Time (UTC).