Digital monitoring and data analysis method for optical fiber link

The invention discloses an optical fiber link digital monitoring and data analysis method, and relates to the technical field of optical fiber link abnormity monitoring, and the method comprises the steps: collecting station abnormal sample feature data, station abnormal sample label data and source point topological positioning sample data in advance, and constructing source point positioning sample feature data and source point positioning sample label data; training a site anomaly prediction model, then training a source point positioning map model, loading the site anomaly prediction model for a system background of each optical fiber site, outputting a single-point anomaly label for each optical fiber site, and if a secondary traceability condition is not met, checking an abnormal source point according to the time when the output single-point anomaly label is abnormal; if yes, calculating the comprehensive source point probability of each optical fiber site, and performing abnormal source point troubles