Radar Echo Automatic Classification Algorithm Using Support Vector Machine
In this paper, an AgileDARN (Agile Dual Auroral Radar Network) radar echo classification method based on support vector machine is proposed. AgileDARN radar echo includes ionospheric backscattering echo, meteor echo, noise interference, etc. With the continuous operation of AgileDARN radar, the amount of data increases rapidly, requiring efficient and reliable classification methods. In order to efficiently classify the echoes of AgileDARN radar, this paper proposes an echo classification method based on support vector machine. By analyzing the characteristics of the autocorrelation function (ACF) of the sampled data and extracting the features, the support vector machine(SVM) classification method is adopted to classify AgileDARN echo into ionospheric backscattering echo, meteor echo and noise interference. The data analysis shows that the classification accuracy of training data set is more than 99%, and that of test data set is more than 95%. Using this classification model to classify 1800 echo data of AgileDARN radar, the classification accuracy is more than 91% compared with the result of manual interpretation.