Flight Data Matching

Flight Test data matching, the process which takes flight test data, analyses it and uses the outputs from this analysis to match the digital model too, is a time consuming, slow and highly iterative process.  The skills required to successfully develop and tune a model using flight test data are learned over a long apprenticeship, often in excess of 25 years. Mostly, this is because aircraft are closely coupled systems so that changing one aspect on the model has a knock-on effect to many other  characteristics and responses. Finding the model fit which is then able to meet the requirements of the applicable standards e.g. FAA Part 60 rev 2 or equivalent when producing a high fidelity Level D type model for Pilot training fidelity simulation is a difficult and very time consuming task.

j2 has already provided the information elsewhere in the website on how models can be developed using j2 Builder, j2 Elements, or external models or data integrated into j2 software using j2 Developer, but now let us show you how j2 Flight is revolutionising the whole flight test data matching process.

There are many challenges that have to be overcome when tuning a model using flight test data for simulator certification applications. Firstly, the regulatory requirements demands a significantly more varied spread of flight test data for high fidelity simulation than for certification purposes.  This immediately presents challenges when purchasing OEM data packs.

In addition, poor quality flight data has many inconsistencies which if left unchecked can introduce characteristics that are not part of the aircraft. When using existing tools, it is quite common for a range of ‘frig’ factors to be used to correct the problems caused by such inconsistencies. Under examination such factors are easily identifiable.

An on going challenge has been the the length of time it takes to analyse the flight test data and manually identify the necessary corrections to the ‘a priori’ model, ensuring that it matches the flight test data. This can have a significant and marked impact on the overall simulator build project cost.

Finally, the continued reduction in workforce/skillset with sufficient experience to continue to do things the “old way” is fast reducing and this presents a significant business risk for small, medium and even large simulator manufacturers as they look to protect that knowledge and move on from older legacy systems.

j2 has already vested significant time and energy in launching j2 Flight, its unique flight test data handling, analysis and simulation plug in module. j2 has already established a proven alternative methodology for tuning models that does not require an OEM data pack.  j2 can also adjust existing models to refine new characteristics on and off line within a single continuous Tool-Kit and without any requirement to write code.

The j2 process and methodology can assist with identifying what additional flight test data is required to develop full qualification test points for high fidelity simulation applications. This additional data can come from any source and can be used to tune the model as an alternative to the contents of OEM Data packs.

The j2 methodology is uniquely able to smooth flight test data and remove excessive noise/bias whilst leaving the real aircraft nuances untouched. J2 Flight includes a full Unscented Kalman Filter, Total Variation Regularized Numerical Differentiation and Rauch-Tung-Striebel Smoothing, whilst also removing noise and bias for Flight Path Reconstruction from the flight data.  In this process, a kinematically consistent set of all 18 aircraft states are calculated.  These are smoothed and the bias removed so that they provide a clean set of test points for further analysis against QTG’s. The j2 process built into j2 Flight also undertakes flightpath reconstruction, re-prediction, parameter identification (PID) and it is able to identify model corrections for model tuning and QTG acceptance steps.

A further significant advance in methodology is that j2 software works at the coefficient level as part of the analysis process. This can uniquely place the model in the same point in space and time as the flight test aircraft. This removes any need for additional integration steps in the analysis.  Each analytical time step within j2 software is 1/1000^th of a second. The benefit of this is that additional signal noise, a potential source of error from the analytical process is completely removed.

This proven and fully documented process is already commercial and provides a new approach that is robust, repeatable and teachable. Relatively inexperienced engineers can be trained in how to tune models in a period of weeks without the need of a 25-year apprenticeship. This breakthrough is already full supported and technically proven through j2 being awarded 2 x TAA licenses with major US based simulator OEM’s.

The analysis that is possible within j2 Flight, part of the j2 Universal Tool-Kit software, has significantly reduced the work time needed to complete the flight test data matching process. This improvement is only possible because the j2 method does not require a test by test iterative process, instead j2Flight can run all lateral and longitudinal tests in large batches, whilst automating how corrections are found. This reduces the overall process, usually measured in man years, down to one that is measured in man months. A large US Defence Simulator OEM ran the entire flight test programme including all lateral and longitudinal tests in a single analytical run, allowing the j2 software to automatically determine the necessary corrections needed to comply with the QTG tolerances to meet the FAA part 60 requirements.