Research on Electronic Signal Feature Extraction and Classification Based on Machine Learning

This paper presents a new approach to extracting and classifying electronic signals based on deep learning. Combining the merits of CNN and LSTM, a hybrid neural network structure is proposed to extract features. Finally, the validity and innovation of this algorithm are proved by experiment. The experimental dataset includes many kinds of electronic signals, including communication and industry control. Experiments show that this method can improve the precision of feature extraction by 25%. The precision of classification is 92.5%, the recall rate is 91.8%, and the F1 score is 92.2%. Compared with other methods, the proposed algorithm has better anti-interference capability and higher computing efficiency.