Research On Radar Signal Sorting Technology Of Multi-parameter Clustering

Radar signal sorting plays an important role in the field of electronic countermeasures,at the same time is the important part of passive radar seeker.Radar signal sorting completes the process of detecting the target radars and extracting the featured parameter in the complicated environment.The performance of radar signal sorting will affect performance of radar reconnaissance system directly.With the complexity of electromagnetic signal environment and the increase of radar signal diversity and pulse density,the traditional radar signal sorting algorithm has encountered a bottleneck because of tolerance,meanwhile moving to the clustering algorithm slowly.This content mainly focuses on the fuzzy clustering radar signal sorting algorithm and the support vector clustering sorting algorithm of multi-parameter clustering sorting algorithm.The fuzzy clustering sorting algorithm is widely used in clustering sorting algorithm.However,the calculation of the fuzzy similar matrix and the judgment of signal type are major difficulties.Aiming to solve computation of fuzzy similarity matrix increases with the increase of pulse density,the featured samples extracted method is put forward which uses the featured samples to calculate the fuzzy similarity matrix instead of the whole data,thus reducing the computation greatly.The fuzzy clustering sorting algorithm can amplify the difference between the radar signal parameter,so an agile radar signal can be divided into many pulse flows and the pulse repetition period can not be extracted normally.On this basis,a pulse repetition period matching method is proposed.According to the fundamental wave and multiple harmonics,it can be judged whether the pulse flow in different clustering centers belongs to the same pulse radar signal and the pulse repetition period is extracted again.The dithering radar signal and the sliding radar signal have the similar radar signal variation rules,which leads to misjudgment of the dithering radar signal and the sliding radar signal.On this basis,the second pulse repetition period processing method is proposed to judge two signal types according to number of peaks of periodic difference of neighboring pulse repetition.Support Vector Clustering(SVC)algorithm is a more accurate classification method,whose algorithm maps radar data points to high-dimensional space and makes clustering judgment on sample data points in high-dimensional space.Aiming at the partition of non-support vector data points in cluster calibration process,an improved method is proposed based on the double-centroid cluster calibration algorithm.The vector relationship between support vector points,non-support vector points and double-centroid is used to judge.The simulation results show that the improved algorithm improves the accuracy of non-support vector points partition and reduces the sensitivity to the numerical selection of Gaussian kernel width.