The invention relates to the security monitoring technology field and more particularly, to a south-to-north water diversion security data abnormal mode discovering method and a system, which are centered by three respects: the sensor security threshold determining, the monitoring data predicting and abnormal mode discovering mechanism construction. Going further, based on the self-servicing method for sensor security threshold calculation, based on the monitoring data predicting by a machine learning and multi-mode method and based on the research on the abnormal mode discovering mechanism construction for information processing intelligently, scientific theoretical basis and auxiliary decisions can be made for the south-to-north water diversion security data abnormal mode discovering, including: dividing the sensors into groups; determining the security thresholds of each sensor; determining the already fixed threshold of each group of sensors; determining whether the data received by any sensor exceed the security threshold of the sensor or not; and determining whether the ratio of sensors with abnormity in the sensor groups exceeds the already fixed threshold or not.