Automated on-ground FDIR for ESA’s XMM-Newton mission

Abstract

The flagship of European X-ray astronomy, ESA's XMM-Newton space observatory, was launched in 1999 with an expected lifetime of up to ten years. Through an operational change of the AOCS system, we extended its potential lifetime until the late 20ies of this century. Spacecraft and instruments are performing without major degradation. Because of the highly elliptical orbit, the spacecraft is permanently visible from the ground station network. For this and other reasons, the originally chosen on-board autonomy design is very limited. Fifteen years of inflight experience created a deep understanding of the spacecraft behavior including elaborated strategies and procedures to recover from reoccurring non-critical anomalies on board. Based on this experience, an automated ground failure detection system has been developed to provide reliable fault isolation and adapted reaction. We present here the system put in place to conduct automatically some of the nominal XMM-Newton operations and share our experience with the automatic contingency recovery tool after half a year of usage. The on-ground Failure Detection, Isolation and Recovery (FDIR) system is able to read various sorts of anomaly messages, either from the spacecraft (telemetry being out-of-limit), from the ground segment (link loss) or from the mission control software (an application stopped). Upon detection of anomaly, the automation system starts the execution of a procedure to confirm its consistency and persistency to avoid over reaction to minor events. Once the anomaly has been confirmed and identified, the recovery is kicked off under the supervision of the operator or autonomously. The targeted level of automation shall release the operator from activities that were so far manually performed, let him concentrate on other activities running in parallel, reduce the stress in contingency situations and lessen the probability of human error.

Evolution of the Eclipse Operations Concept for ESA’s X-ray Observatory XMM-Newton

“…As it is described in the work of Godard and al. some criteria had to be defined to decide which operational scenario has the priority to be automated first 4 . Figure 4 shows the usual sequence of events in an eclipse timeline.…”

“…As can be seen from Table 3, eclipse operations score fairly high in both workload and length of operations, hence best suited to be automated (number of eclipses vary over the life time of XMM-Newton, hence a range is depicted as eclipse frequency in Table 3). 300 Table 3 -Criteria to select activities to be automated 4 The activities mentioned above are already used operationally or under development and about to be validated. The criteria to select the candidate for automation are the frequency of occurrence, the workload for the human operator and its duration.…”

Optimization of the battery usage during eclipses using a machine learning approach

“…The automating platforms [2] have been tailored to the mission's needs to perform, from ground, what newer platforms do on-board. Repeating manual activities for both nominal situation and contingency recovery that were previously executed manually by the operator are now triggered and executed autonomously [3,4].…”