The invention relates to the technical field of 
food safety detection, in particular to an 
analysis method and 
system for identifying the heat treatment degree and 
doping of 
liquid milk, and the method comprises three steps of sample pretreatment, 
mass spectrometry and 
data analysis. According to the invention, a 
machine learning technology and 
mass spectrometric detection are combined to establish a milk identification method and 
system with different heating degrees. In this way, a model is established by utilizing a 
machine learning 
algorithm according to the difference of 
polypeptide composition and content change of milk of different heated types in 
mass spectrum information, and information of a heat-sensitive 
peptide fragment is obtained. Then the model is continuously trained and optimized, a powerful prediction model is screened out, and therefore efficient, stable and accurate identification is achieved. The identification method and 
system are scientific and effective. The method established by the invention has the advantages of simplicity and convenience in operation, low 
organic solvent consumption, high 
throughput, high prediction accuracy and the like.