Telemetry Fault-Detection Algorithms: Applications for Spacecraft Monitoring and Space Environment Sensing

Abstract

Algorithms have been developed that identify unusual behavior in satellite health telemetry. Telemetry from solid-state power amplifiers and amplifier thermistors from 32 geostationary Earth orbit communications satellites from 1991 to 2015 are examined. Transient event detection and change-point event detection techniques that use a sliding window-based median are used, statistically evaluating the telemetry stream compared to the local norm. This approach allows application of the algorithms to any spacecraft platform because there is no reliance in the algorithms on satellite-or component-specific parameters, and it does not require a priori knowledge about the data distribution. Individual telemetry data streams are analyzed with the event detection algorithms, resulting in a compiled list of unusual events for each satellite. This approach identifies up to six events of up to six events that affect 51 of 53 telemetry streams at once, indicative of a spacecraft system-level event. In two satellites, the same top event date (4 December 2008) occurs over more than 10 years of telemetry from both satellites. Of the five spacecraft with known maneuvers, the algorithms identify the maneuvers in all cases. Event dates are compared to known operational activities, space weather events, and available anomaly lists to assess the use of event detection algorithms for spacecraft monitoring and sensing of the space environment.

Novel statistical method for data drift detection in satellite telemetry

“…In this case, the mission ops team and critical design review document 21 of the subject satellite have indicated that for payload OFF condition, the maximum tolerance and minimum tolerance limits of BUS CURRENT are revised to 0.886 and 0.601 Amps respectively after a period of 2 years of in-orbit operation and are given by Equations ( 3) and (4). Similarly, for a payload ON condition, the maximum tolerance and minimum tolerance limits are revised to 1.19 and 1.43 Amps, respectively after a period of 2 years of orbit operation 21 and are given by Equations ( 5) and (6).…”

“…Data that is used in the initial or preliminary stages for building a model that accurately represents a system's behaviour slowly deviates from its pattern or distribution. This is due to the fact the harsh outer space gradually degrades the performance and properties of the materials and the components 5,6 which ultimately represents a change in the recorded telemetry data. Therefore, models that are built and trained using the heuristic data at the beginning of mission life may not accurately represent the data state during the mid‐mission life or end of life of the mission.…”

Human Factors Considerations in Satellite Operations Human-Computer Interaction Technologies: A Review of Current Applications and Theory

“…Satellites tend to operate for long periods between anomalies or non-nominal events, resulting in trending data being stored for very long periods until the data is needed to be analyzed [7,125]. Satellite telemetry data can be considered Big Data due to the size of the data, the speed at which the data transmits from a satellite to the ground station, and its relative complexity if not for the proper software infrastructure to allow for proper data analysis [129].Satellite Big Data processing may be a viable option to help mitigate human error by better equipping the HITL with the correct information ahead of time if descriptive, predictive, and prescriptive analytics are employed [124].…”

Univariate exploratory data analysis of satellite telemetry

“…The outer space effects like ionisation and solar winds produce undesirable effects like single event upsets (SEU), gradual performance degradation of components, etc. in a satellite manifest as faults or anomalies 3,4 . Timely response is the goal so that anomalies/faults do not manifest as failures so as to guarantee availability, reliability and performance 5,6 …”