AI for model based diagnostic at System level (AiMSys)

Exploit usage of System level Model Based Diagnostics for reliable anomaly detection as well as for pre-launch testing including pass/fail criteria evaluation for tests.

Previous activities (T709-503OS) have looked into the usage of Model based diagnostics at sub-system level (like Thermal) joined with AI for the identification of issues. This work expands to cover the complete system at system level. It uses results from ongoing Digital Twin activities (T708-505SW), but as well standard simulation modelling technology. One key issue to be tackled is the evaluation of the output to reliable detect anomalies.Here previous work are expanded to cover all available sources of data for evaluation, hence creating a powerful AI to support a three way comparison of:- Live data from the real mission - Simulated data from models- Archived data from previous pre-launch test or mission archives. The decision logic aims at becoming bi-directional, i.e. ability to define pass/fail criteria for tests during pre-launch, hence bringing autonomy to the testing process together with the ability to detect anomalies post launch. ;The activity encompasses the following tasks:- Review of existing technologies and output of previous work;- Setup a suitable laboratory setup allowing testing of various algorithms;- PoC of methods to correlate test, simulated and live data;- PoC for evaluation of test, simulated and live data for identify anomalies as well as providing root cause analyses at system level;- PoC for usage of Model based diagnostics for brining autonomy to testing processes including evolution of test results;- Identify the required fidelity needed for models in order to be suitable for Model based Diagnostics at system level;- Evaluate the performance of Model based diagnostics wrt. traditional rule based anomaly detection.;Software shall be delivered under an ESA Software Community Licence, so that any individuals or entities within ESA Member States can access to it and can provide update to the community of users.