Research On Micro-Doppler Effects Applied In The Classification Of Aircraft Targets

As one of the main operational tool in the modern war, aircraft plays a vital role to get the victory of the war. The micro motion of rotating parts on the plane produces periodic frequency modulation on the radar echo, which result in different aircrafts showing different micro-Doppler features. This has important guiding significance and practical value on the research of aircraft targets classification.In the dissertation, we research the method of feature extracting applied in aircraft targets classification based on the analysis of the micro-Doppler effect mechanism. The main work is as follows:First, for helicopter, propeller airplane and jet aircraft, we deduced the ideal parameter modulation echo model of their rotating parts. And also we studied some practical parameters effect on three kinds of plane rotating parts, such as the aircraft tail rotor, engine number, blade angle, pulse repetition frequency, beam dwell time and emission wavelength and so on. From the view of characteristic differences between class and class aggregation, we extracted the time domain second moment, entropy, time domain amplitude relative amounts and phase relative amounts to be the feature vector. Then, we classified the three types of aircraft using support vector machine. Experiment results showed that the extracted feature vector could effectively classify the three types of plane in high signal-to-noise ratio, and the performance along with the elevating of signal-to-noise ratio. The four features can realize easily and their computational complexity is low, so these are helpful for engineering implementation.Second, in practice, the extraction of the JEM feature requested the echo including at least one complete cycle. It was often difficult to meet the condition when the radar working under the scanning state. In this dissertation, a method was proposed for estimating the micro-Doppler parameter of scanning-beam radar targets based on a fractional period echo. This method transformed the signal to the time-frequency domain by the time-frequency analysis technology. The fractional period micro-Doppler signal was extracted from the time-frequency domain of echo signal. The micro-Doppler signal of a rotation target could be expressed by a sinusoid. By using the properties of sine, we could estimate the rotation rate and rotation amplitude of the target. With these two parameters, the rotation radius also could be given. As the intrinsic characteristics of the target, the rotation radius reflected the structure information and motion characteristics of the target and provides effective basis for the recognition and classification of target. Thus, we could describe the rotation target movement in more detail. Compared with the Hough transform, the simulation experiment showed that this method had high precision and lower complexity. In addition, we also verified the effectiveness and robustness of this method using the measured data.