Approach Based on Cloud Model and SVM for Emitter Identification
Yang Cheng-zh · Modern Radar · 2013
For radar emitter recognition,support vector machine can handle only quantitative data,while it can not deal with qualitative data,besides it will not resolve the multi-classification problems,resulting in a problem that the radiation source can not be identified correctly. So this paper proposes a recognition algorithm of support vector machine( SVM) and multi-classification based on cloud model and kernel function. The algorithm achieves the conversion from the qualitative concept to quantitative interval value by using the cloud model and the nonlinear mapping from the vector input of interval type to the model output of interval type by using the modified kernel function,and using SVM of decision tree modified the problem of multi-class classifier. The simulation results show that this method can not only deal with input vector of the interval type,but also handle the input vector of the scalar type. What's more,for radar emitter signals which have the high similarity in characteristic parameters,it has better classification results.