Research On Detecting Transients And Variable Sources In AST3-2 Survey

Observation on transients and variable sources plays an important role in studies on time-domain astronomy,optical survey and follow-up observation are important methods in visual.large amount of valuable data can be obtained by systematic obser-vations on specified sky areas and follow up performed on sources we are instrested in.Advanced wide-field telescopes have been massively introduced to long-term and con-tinuous observations in modern astronomy research,with considerable data achieved,traditional human classification and threshold cuts can no longer satisfy the demand of data processing,introducing machine learning models to classification and forecast of data processing result appears very necessary.In this paper,We discussed methods to detect transients and variable sources using AST3-2 2016 dataset.AST3-2 Telescope locates in Dome A,loftiest ice dome on the Antarctic Plateau,the climate characteristics there makes it a perfect site for continuous time domain astronomical survey observations,the huge amount of observarion data it produced requires more efficient data reduction program to be developed.Also data transmission in Antarctica is much difficult,thus we hope to perform data reduction to detect variable sources and transient sources remotely and automaticaly in Antarctica,but this attempt is restricted by the poor computer performance in Antarcitca.For the realization of our aim,we developed a new method based on pre-existing image sub-traction method and random forest algorithm,taking AST3-2 2016 dataset as our test sample.We perform image subtraction on data set,then apply principle componet anal-ysis the extract features of residual images,random forest is used as a machine learning classifier,and a recall rate of 97%is resulted.Our work verified the feasibility and accuracy of our method and finally find out a batch of candidates for variable stars in AST3-2 2016 dataset.The first chapter is about international and domestic progress in facility,pipeline,and data reduction of large scale surveys,the second describes the nature of variable stars,these are background of our work.The third chapter is about CCD features and basics of photometry,the fourth illustrates algorithms we used and the fifth introduces software our work is based on,all these components consist base of our work.Finally,the sixth chapter introduces this method based on image subtraction and random forest in detail,and the result of our research,we prospect the furture development of this method at last.