Compound Model of Navigation Interference Recognition Based on Deep Sparse Denoising Auto-encoder

Abstract

For the navigation problem that has been affected by interference signals for a long time, a compound classification model algorithm based on a deep sparse denoising auto-encoder network is proposed. Firstly, frequency conversion and preprocessing are performed on several typical interference signals listed in this article, and then a deep sparse denoising auto-encoder is used for training sample data. After fine adjustment,final encode layer output the training data features. In the case of removing redundant information, maximize the retention of the original sample information. Finally, by comparing the recognition accuracy of three different classification models, it is concluded that the composite model proposed in this article has the advantages of fast convergence and high recognition rate, and it can get more than 2dB performance gains compared to the other two algorithm. It further demonstrates the advantages of deep learning in the field of navigation interference recognition.