Session: 04-02-02: Advances in Aerodynamics & Novel Aerospace Propulsion Systems

Paper Number: 68541

Start Time: Tuesday, 10:05 AM

68541 - Combined Time- and Frequency-Domain Aircraft System Identification Using Pareto Optimization 

There are two broad categories of system identification used for aircraft aerodynamic modelling, each with distinct advantages and disadvantages: the time-domain approach and the frequency-domain approach. An aircraft aerodynamic model forms a crucial component of aircraft development programs for control law development, flying qualities assessments, aircraft performance analysis, and pilot training. More recently, aerodynamic models are being explored for the safety-critical application of certification-by-simulation, where a model is used to assess the safety of a maneuver or a new aircraft configuration prior to resource-intensive and time-consuming flight test programs.

The time-domain system identification approach works to directly minimize the error between the simulation and flight data time-histories. Matching the time-history is important for applications where analysis is completed in the time-domain. This includes performance analysis and assessing the limit loads and dynamics necessary for certification-by-simulation. A model developed in the time-domain is also strong at capturing static aircraft behaviour, thus facilitating a good trim solution, which is critical to a range of analyses such as evaluating performance in steady flight, evaluating handling qualities about specific flight conditions, and initializing pilot-in-the-loop simulations.

Perhaps unintuitively, good time-domain matches do not necessarily translate to a model that reproduces the target aircraft handling qualities as perceived by a pilot in a flight simulator. For example, despite extensive efforts to obtain objective criteria for handling quality assessment from time-domain models, experts have not been able to obtain criteria that correlates with pilot opinion leading them to converge on frequency-domain criteria instead. This points to the importance of considering the frequency response when developing an aerodynamic model for pilot-in-the-loop simulation.

The frequency-domain aircraft modelling approach seeks to match the frequency response of the model to the frequency response estimate gathered from flight data. Accordingly, matching frequency responses builds similar handling qualities into the aerodynamic model such that the simulator feels like the real aircraft when evaluated by a human pilot. A model that matches the frequency response of the aircraft is also crucial for control law development. However, a model developed in the frequency-domain neglects higher-ordered dynamics, and when combined with random and systematic error in the collected flight data and frequency response estimate, such a model does not necessarily translate to good time-history matches.

A combined system identification approach that includes both time- and frequency-domain information can combine the strengths of both methods. This combination forms a multi-objective optimization problem. This study applies Pareto optimization for the system identification of a forward flight model for the National Research Council of Canada's Bell 412 helicopter. The generated Pareto fronts showed the existence of a trade-off and overspecialization where moving from the compromise solution to either the isolated time- or frequency-domain solutions resulted in a small improvement in one cost while the other suffered relatively more. The compromise solution between the time- and frequency-domain matches was found to avoid overspecialization in either of the domains.

Presenting Author: Joseph Ricciardi National Research Council Canada

Authors:

Terrin Stachiw National Research Council Canada
Joseph Ricciardi National Research Council Canada
Alexander Crain National Research Council Canada