X-ray Pulsar Signal Recognition Based On ANN

X-ray Pulsar-based Navigation(XPNAV)is a new type of celestial navigation method with strong autonomy and high reliability.XPNAV has a good development prospect in deep space navigation because of X-ray pulsar has a wide spatial distribution in the universe and stable rotation period.The recognition algorithms of X-ray pulsar is one of the key technologies of XPNAV.A fast recognition algorithm has a great signigicance in promoting the engineering realization of XPNAV.A common method for the recognition of X-ray pulsar signal is to extract features with translational invariance in the signal and to match the features of the standard signal.In order to improve the efficiency of X-ray pulsar signal identification and save computing resources and storage space.Some scholars have introduced the method of machine learning into the X-ray pulsar signal identification algorithm,and the pulsar signal can be successfully identified.In recent years,artificial neural networks have achieved good application results in the fields of pattern recognition and artificial intelligence.Pulsar has a high stability rotation period.If X-ray pulsar observation signals are folded according to the correct pulsar period by the periodic folding algorithm.Then the pulsar integrated pulse profil has certain features.If the folding period is not correct,the pulsar integrated pulse profil does not have obvious features.According to the above characteristics of the X-ray pulsar integrated pulse profil,This paper presents an X-ray pulsar signal identification method based on Artificial Neural Network.Two classification programs of Long Short-Term Memory Network and Convolutional Neural Network are constructed respectively.According to the difference of the processing data format between Long Short-Term Memory Network and Convolutional Neural Network,two methods of establishing training data set and test data set are proposed.In order to verify the ability of Convolutional Neural Network and Long Short-Term Memory Network to recognize the X-ray pulsar integrated pulse profil,X-ray pulsar observation signals with different intensity noises are simulated through numerical simulation.The identification effect of CNN and LSTM is compared from two aspects of identification accuracy and program running time.The relationship between the identification accuracy of CNN and LSTM and the duration of the signal is compared under different noise intensities.It shows that when the pulsar integrated pulse profil has a certain signal-to-production ratio,the recognition accuracy of the pulsar integrated pulse profil by CNN and LSTM will be very high.Compared with the traditional pulsar signal identification algorithm based on bispectral,The method introduced in this paper is obviously superior to the traditional method in terms of computational complexity.