Research On Algorithm For Extracting The Period Of Astronomical Light Curve Signal Based On Graph Signal Processing

Astronomical variability signals are signals that measure the radiation flux of celestial bodies over time,characterized by multiple periods,non-uniformity,large intervals,and noise.The periodic extraction of light curve signals is conducive to revealing the physical mechanism behind celestial bodies.Traditional algorithms for extracting the period of light curve signals mainly include algorithms based on time domain analysis,frequency domain analysis,and time-frequency domain analysis.Although these algorithms can extract the period of the light curve signal,there are still some problems,such as the inability to utilize some large time interval signals,the difficulty in extracting some small period components,and the need for signal preprocessing,resulting in low efficiency in period detection.Therefore,in view of the shortcomings of traditional algorithms,it is necessary to study new period extraction methods.In this dissertation,using the methods such as path graph and visibility graph in graph signal processing,we extend the period extraction of light curve signals from the traditional analysis domain to the graph domain,and combine them with methods such as Markov distance and graph Fourier transform to propose a period extraction method for light curve signals based on graph signal processing.The main research content of this article is as follows:(1)Performance analysis of traditional period extraction algorithms.This paper introduces in detail three classical period extraction algorithms in the traditional analysis domain,namely LS,JK,and WWZ algorithms,and uses these algorithms to perform period extraction experiments on the light curve signals of celestial bodies.Finally,the advantages and disadvantages of these algorithms in performing period extraction of light curve signals are systematically analyzed.(2)Aiming at the problem that traditional algorithms cannot make full use of signals with large time intervals,resulting in information loss and the inability to extract small periodic components of light curve signals,a graph signal processing method is applied to the period extraction of light curve signals,and a period extraction algorithm based on Mahalanobis distance weighted path graph Fourier transform is proposed.This algorithm first converts the light curve signal into a path graph signal,and then uses Mahalanobis distance to weight it to characterize the non-uniformity of signal sampling.Finally,it uses Graph Fourier transform to obtain the spectral domain of the signal’s eigenvalues,and then extracts the periodic components of the signal.The analysis results of simulated and real data show that the period extraction algorithm based on GFT and Markov distance weighting can accurately identify all the periodic components of the light curve signal.(3)In order to improve the efficiency of period detection,an innovative weighted limited traversal visual graph algorithm was proposed from the perspective of complex networks to address the problem that traditional algorithms need to preprocess the light curve signal before performing period extraction to suppress the impact of noise and improve the accuracy of period detection.By transforming astronomical light curve signals into complex networks,while preserving the characteristics of the signal itself,the noise resistance performance is improved.Subsequently,a weighted finite traversal viewable model is obtained by applying Mahalanobis weighting to the adjacency matrix of the finite traversal viewable.Finally,an accurate light curve period is extracted using the method of graph Fourier transform.By applying this method to real light curve signal data,the results show that this method can improve the efficiency of period detection of light curve signals while accurately extracting period components.