Abstract
This Paper analyzes nonlinearity of human operation by using neural networks. Manual control experiments based on PC were carried out to emulate the longitudinal control of an aircraft. By using neural networks, two-input and one-output pilot models are identified from experimental data. The obtained pilot models are analyzed through FFT method, hi the results, it is found that human operations have nonlinear behaviors. Conventionally, there is a large number of researches which described human operation as linear pilot models and analyze them to estimate the difficulty of operation. However, this research suggests a possibility that actual human operations include nonlinearity and the nonlinear characteristics enables us to explain how pilots control the stick and evaluate workload levels.
