Support Vector Clustering and Type-Entropy Based Radar Signal Sorting Method

国强,王常虹,李峥

2010-01-01

Abstract:Focusing on the problem that the conventional radar signal sorting methods based on the tolerance are difficult to accurate sorting in complex and dense pulse environment of modern electronic warfare,a new sorting method is presented based on delaminating coupling and support vector clustering(SVC).All pulse sequences of radar signals are determined using delaminating processing and then each sub-sequence is separately processed by SVC clustering.Then the idea of delaminating coupling is introduced.And re-SVC clustering is determined by using all centroids of the extracted sub-sequences so that the sub-sequences which belong to the same radar emitter in all pulse sequences of the various layers are connected to achieve the radar full pulse sequence sorting.The character that type-entropy value is getting bigger with the increasing of categories and complexity of pulse signals is used to calculate type-entropy on the multi-parameters clustering results.The type-entropy recognition is used to assist signal sorting.Experimental results show that the system can efficiently sort radar signals in the complex pulses environment and avoid the impact on signal sorting for the tolerance problem in traditional signal sorting methods.