Multiexperiment data processing in identifying model helicopter's yaw dynamics

摘要

The multi-experiment data is usually needed in identifying a model helicopter's yaw dynamics. In order to strengthen the information of the dynamics and reduce the effect of the noise, a new kind of least square method by using a weighted criterion is investigated to estimate the model parameters. To calculate the factors of the weighted criterion, a neural perceptron is trained to determine the factors automatically. The simulated outputs of the model derived by this kind of method fit the measured outputs well. It is suggested that this kind of data processing method is useful in identifying the yaw dynamics and processing the multi-experiment data for the system identification.