Skip to main content

Innovative solution for FMS computed trajectories validation by means of pilot actions emulation, comparison with PANS-OPS criteria and data mining techniques.

Periodic Reporting for period 2 - FIVER (Innovative solution for FMS computed trajectories validation by means of pilot actions emulation, comparison with PANS-OPS criteria and data mining techniques.)

Reporting period: 2019-11-01 to 2021-01-31

On 21st November of 2018 Scalian, CGX AERO and Thales launched the FIVER project ("FMS Innovative Validation for Enhanced Robustness"), funded by the European CleanSky 2 Joint Undertaking program. This 27-month project shall enable Thales to accelerate the maturity growth of its Flight Management System (FMS), through innovative massive testing methods based on simulation and artificial intelligence, with the final objective to limit the environmental impact of aeronautics.

The overall performance of the FMS, its ability to adapt to a wide variety of conditions (including the diversity of flight profiles) is decisive for flight safety, but also for its environmental impact.
We aim to allow Thales's FMS to optimize flight paths and therefore fuel consumption, while continuing to ensure the safety of the flight and its crew.

To achieve this objective, CGX AERO and SCALIAN have worked together, continuously involving the expertise of Thales. The high-level distribution of the overall objectives within FIVER were:
• Thales provided in particular the FMS and its regular updates, their analysis tools developed internally in the past, the elements necessary to run the FMS, and the environment on which the solution is now implemented. Thales also built jointly with CGX AERO the simulation environment for massive and cloud-oriented FMS analysis on Amazon Cloud (AWS).
• CGX AERO contributed with Thales to built the interaction with the FMS through a Virtual Pilot and simulation allowing high-exposure of the FMS functions in representative flights to drive the generation of FMS-characteristic data on massive scale for big data analysis. Moreover, CGX AERO developed trajectory analysis tools aiming at detecting potential trajectory outliers and comparing FMS trajectories with reference ones produced by internal procedure designers. The collected data and computed analysis results were then packaged and send to the Data Science Studio for big data treatments.
• SCALIAN's mandate was to develop a Data Science Studio, a virtual environment capable of analyzing the data produced by simulation, using machine or deep learning methods. This environment had to be oriented towards business users and offer a certain understanding of model outputs, so that test engineers can interpret the anomalies detected by the models.

As a conclusion, both CGX AERO and SCALIAN modules have been integrated into the Thales test environment (Thales Virtual Aircraft solution on Amazon cloud).
Even if the whole chain is still perfectible, the global result obtained is satisfactory, especially if one considers the starting point, now Thales owns a massive testing environment that can be used in the frame of the IADP LPA to validate the FMS before the integration on the Airbus Disco Bench.

Another stake of the project was to be able to confront Instrument Flight Rules Trajectories designer work with the reality of a flight execution through FMS guidance. The solution brings the first analysis of FMS regarding the procedure data and the expectation still asks for a dedicated new project to continue the work between CGX AREO and Thales to bring more “machine-interpretable” results and alternative framework to complete the Massive Testing.
At the end of the project, the partners have successfully implemented their solutions in the Thales test environment. Although it is still too early to estimate whether Thales has achieved the overall project objective, they acknowledge satisfaction with the overall results of the project.

To achieve this result, the following work has been addressed by the partners:
• CGX AERO has developed in close collaboration with Thales, a Test Generator Module (allowing the configuration, organization and follow-up of the test campaigns), a Test Running Module (integrating a simulation engine based on different Virtual Pilot typologies) and a module for “in-live” analysis of simulation data (based on a set of rules, on comparisons with reference data obtained by automated computation, or on statistical metrics). The Virtual Pilot allows to address determined objectives on the FMS such as evaluating the accuracy, stability and robustness of the predictions but also confirming non-regression in the software.
• SCALIAN has developed the Data Science Studio, a platform integrating multiple storage instances (simulation data, pre-analysis data, analysis models, analysis results, docker images), a platform administration module and a module for analyzing the data produced by the test running module. The platform is based on a micro-service architecture, allowing a high scalability of the solution, and the re-learning of models, or the addition of new analysis services.
The Data Science studio does not completely give the expected results regarding analysis capabilities but some anomalies are well detected in trajectories computed by the FMS on kerosene consumption for example (a first step is reached in automatic trajectories analysis). The gap between the expected result and the one obtained is due to the complexity of the data and the field of use, however the analysis modules implemented in the Data Science Studio have enabled Thales to orientate itself in the direction of work to be conducted in the near future.

To date, we foresee the following channels to communicate on project’s results:
• Publication of the official text by the JU (Clean Sky 2 Journalist interview)
• Press release, at the initiative of SCALIAN's press department
• Joint communication on events in webinar format by SCALIAN
• CGX AERO YouTube channel supporting a video promoting R&D activities and mentioning FIVER associated to a post on LinkedIn
• Update of Consortium partners web site with project results post
Potential further actions may be proposed after consultation with each other's trade authorities.
The detection of anomalies in complex systems is historically based on the limitations imposed by a rule system. By complementing such systems with a priori free research, based on unsupervised learning algorithms, the FIVER project led to enhance anomaly detection performance from conventional methods.

Once the solution has been delivered to Thales, it will be up to them to conclude whether the desired performance has been achieved in real test situations. The expected enhanced robustness of the FMS validation shall improve FMS performance in terms of fuel consumption and environmental impact and also increase flight safety by shortening the validation phase to embed faster the associated technological developments.
Another stake of the project was to be able to confront Instrument Flight Rules Trajectories designer work with the reality of a flight execution through FMS guidance. The FIVER project demonstrates the interest of such analysis and brings first results that requires to be looked deeper into by considering more flight procedure design cases (massive framework) and by increasing the level of automation on flight procedure related data.
FIVER's partners at launch