Method Analysis Of Frequency-domain System Identification For A Small Size Unmanned Aircraft

The research object in this paper is system identification for a small size unmannedaircraft. Building model for an aircraft is of vital important for design of aircraft controlsystem. System identification is an accurate tool for model building tasks, since aircraftsystem only has some unknown parameter. System identification method depend on a largenumber of flight data, and identified the response between input signal and output signal ofthe system. Compared with calculate method, system identification needs less measure thatcan`t insure accuracy. Compared with wind tunnel method, system identification is moreinexpensive. The result of wind tunnel method and system identification can validate eachother.However, there are some problems in building model for small size unmanned aircraft,such as the short development period and hard to be calculated. The system identificationmethod is used to deal with the problems.There are two ways of system identification: the time-domain method and thefrequency domain method. The two methods are compared according to properties of smallsize aircraft. Frequency domain system identification, which takes advantage of frequencyresponse method, is applied to derive nonparametric model to improve the model accuracy.However, traditional ways of aircraft identification are mainly designed for bigger aircraftsthat can collect flight data with enough dynamic information. So it is necessary to find away to solve the problems raised by bad carrying capability and short time of endurance.To find the method special for small size unmanned aircraft system identification,frequency response identification method is introduced, and a data collection experimentwith special input signal is designed to collect credible flight data for the identificationmethod. The parametric values of longitudinal and lateral dynamics equation are identifiedrespectively. Time domain identification method is employed to validate the estimationresults. Calculation result of parametric estimation is used as reference. The validationresults indicate that the estimated model can be well accepted.