Flight testing of small remotely piloted aircraft for system identification

Aaron Wypyszynski, Purdue University

Abstract

Flight testing of small unmanned aircraft (span 4-8ft, weight 5-15lbs, speed 30-100ft/s) in both the academic and small company fields is sparse at best. Most test programs consist of little more than attempting to perform a mission and then determining if objectives were met. Often, a flight test program may determine an aircraft drag polar at most. In order to truly determine the qualities of an aircraft, it is important to determine the aerodynamic, stability, and control parameters of the aircraft system, so that increased quantitative analysis and comparison may be performed versus the low-level analysis that is typically performed. When qualitative analysis is used to improve a design, the process can be lengthy as a trial and error a process is used to determine what improves or degrades performance. The use of system identification can immediately and quantitatively determine the aircraft parameters. This greatly shortens both the time and the cost of improving the aircraft design. Three factors often prevent the use of system identification in academic and small company settings; complexity, lack of a process that minimizes the challenges of testing small unmanned aircraft and cost. Processes currently used in large flight testing programs in industry provide excellent results, yet take several years of experience to become familiar with and multiple tests in order to tune the processes for best results. Testing small unmanned aircraft also poses several problems, both due to the fact that the pilot is removed from the aircraft and that aircraft in the size range of this study are more susceptible to atmospheric turbulence due to their low wing loading. Cost of a data acquisition system capable of performing system identification is also prohibitive, costing several thousands of dollars when the maximum limit for many small projects can be as little as $1000 for data acquisition hardware. This thesis discusses a method that addresses all three of these problems to enable aircraft system identification on a low budget and short time period for someone who has limited experience in aircraft system identification.

Degree

M.S.A.A.

Advisors

Sullivan, Purdue University.

Subject Area

Aerospace engineering

COinS