Research And Implementation Of Morse Signal Recognition Algorithm Based On AI

In recent years,with the rapid development of artificial intelligence(AI)and deep learning technology,Morse signal has ushered in a new opportunity in automatic recognition.Traditional Morse signal recognition mainly relies on manual listening and recognition,and also relies on traditional machine learning methods for recognition.However,the performance of Morse signal recognition is insufficient in code rate,accuracy and noise adaptive.At present,special buglers are still reserved in the army,listening and translating the Morse signals received,which takes up a high number of manpower.Deep learning has been widely used in image,speech,natural language and so on.Morse signal recognition and the current application of deep learning have many similar places,so applying deep learning to Morse signal recognition can not only solve the shortcomings of traditional methods to identify Morse signal,but also broaden the application field of deep learning.In this study,we mainly use the pytorch deep learning framework,full connected neural network(FCNN),convolutional neural network(CNN)and gate recurrent unit(GRU)neural network structure to build a set of Morse signal recognition algorithm based on deep learning.The data set generated in this experiment is generated according to the specified Morse signal frequency,sampling rate,code rate,amplitude and interference noise of Morse signal generated randomly within the specified range,and is used for training and testing.In this study,two kinds of Morse signal recognition network structure are built for 42 characters commonly used in Morse signal.By comparing the two network structure models,the better performance model is selected.The first is Morse signal recognition model based onthe convolutional recurrent neural network(CRNN),and the accuracy of the model is 97%in the test set.The second is the Morse signal recognition model based on the Gru.The accuracy rate on the test set is 94%left and right,and the model with high accuracy is selected for deployment.In this study,a decoding service subsystem which can identify Morse signal is built by using the better model.The test of the subsystem shows that the application can recognize Morse signal in real time by calling the interface provided by the subsystem.