Research On Analysis Of Signals Based On Deep Learning Method

With the rapid development of artificial intelligence technology,Western military powers led by the United States have begun to develop cognitive radio and cognitive electronic warfare projects based on machine learning theory.At the same time,facing the needs of academia and industry,modern machine learning methods represented by in-depth learning have begun to emerge and develop rapidly in the field of radio communication.Under the background of massive data based on high-speed computing capability of hardware,the combination of related work with deep learning model represented by deep convolution neural network and deep recurrent neural network can pioneer in mining,searching and summarizing electromagnetic signal characteristics at the original data level,for the problem of modulation recognition and terminal recognition of emitter in radio signal analysis.The target recognition accuracy index under the corresponding electromagnetic environment could be realized.Thus,in the absence of prior information,it helps operators mine the important information contained in the target data structure in a short time,and reduces the time cost of traditional methods in the stage of artificial feature extraction.The simulation results show that some typical deep neural networks can improve the performance of radio signal analysis tasks and promote the essential efficiency of electromagnetic signal reconnaissance from the perspective of deep learning technology.Based on this,a new signal analysis and recognition algorithm with deep learning as the core technology is of great significance to the field of electronic countermeasures.