Method and apparatus for monitoring an operational state of a system on the basis of telemetry data.
|572 - Abstract:|
|The European Space Agency developed a new way to diagnose anomalous or unexpected behaviour on systems, through the analysis of that system’s telemetry data. For applications where a big number of parameters (20.000 or more) are relayed for analysis, this methodology presents clear advantages over previous monitoring methods, providing earlier anomalies detections and reducing false positive events. License agreement collaborations are sought.|
This invention was developed for the detection of anomalous or unexpected behaviour of satellites, space probes or planetary rovers. These spacecrafts and robots are usually in contact with operating centres, transmitting their telemetry data such as systems maintenance, diagnosis and mission data. Typical telemetry data for a space mission consists of the monitoring of a very large number of parameters (20.000 to 40.000) over a given time period, with a pre-determined sample rate of, for example, one reading per second. Temperatures, pressures, voltages and currents are some of these parameters, and their readings can be used to monitor the state of the system, allowing the detection of abnormal behaviours that can be an indicator of malfunctions or other situations that require correction inputs by the operator or the system itself.
Timely anomaly detection is of great importance for remote-controlled space vehicles, as they are usually very expensive equipment, impossible to physically repair or retrieve. As such, by anticipating malfunction detection, appropriate corrective actions can be taken sooner, minimizing downtime and mitigating the development into severe failures. Traditional monitoring systems as the out of limits method (OOL) require complex manual input by operating personnel, incurring in considerable engineering efforts and theoretical assumptions that are prone to defects. In OOL, thresholds are set for critical parameters and a warning message appears if they go out of those boundaries.
The present invention overcomes the limitations of prior art, providing an efficient method to monitor large amounts of telemetry data using existent equipment at current monitoring sites. It also increases the reliability of the monitoring by increasing the detection sensitivity while reducing the number of false alarms. These improvements are achieved by using a protocol based on determination of the probability of a set of data, taken at a given time point, being an outlier in the statistical sense. This means that if it deviates so much from other observations, a warning flag is raised.
In the image below it is possible to see the advantages of this methodology over previous iterations of monitoring schemes, as it accurately detects a malfunction way sooner than the OOL method, which only does so after complete system failure.
Innovations and advantages
With this technology it is possible to process a very high number of parameters, checking for abnormal behavior, allowing the earlier detection of anomalous behaviour, providing timely warnings to take appropriate actions.
There is a lower rate of false positives and reduced incurred engineering effort for the monitoring, avoiding human errors and overall improving the reliability of the method.
Domains of application
Monitoring of complex systems with many controlled parameters. High value operations could benefit from this improved methodology, such as offshore oil drilling platforms and every industry that has to deal with often inaccessible, expensive pieces of machinery. Due to the possibility to reduce equipment downtime, this technique proves invaluable for the industry in general.
Last update: 25 March 2013