In modern warfare, the radar electronic warfare environment is more and more complicated. With the variety of radar types and the complexity of radar inter-pulse modulation, the difficulty in identifying signal sorting is increasing. This paper proposes a random forest algorithm to sort the characteristics of pulse descriptors, which can daptively select features and achieve classification. Random forest can make the error balance of unbalanced data. Meanwhile through the multi-decision tree voting method, it can quickly complete the rapid training of large amounts of data. In the case of a pulse loss, the recognition accuracy can still be maintained. Experimental results show that the proposed method is effective in presorting radar pulse descriptors.

random forest  /  decision tree  /  radar signal sorting  /  feature selection