An embodiment of the invention provides an electric power scheduling flow data abnormity detecting method based on an isolation forest algorithm. The method comprises the steps of sampling a data setthrough a system sampling method, establishing sub-forests, and forming a base forest abnormity detector; determining an abnormity condition of data which enter a sliding window through the base forest abnormity detector; performing Bernoulli sampling on the data which enter the sliding window, determining whether the data are stored in a buffer area, and when the data of the sliding window are full, determining a window data abnormity rate in real time; according to a fact that a buffer area data volume and sliding window data abnormity rate exceed threshold values, selecting a model updatingstrategy with a relatively low or relatively high updating proportion; and calculating an abnormity rate difference between each sub-forest and the base forest based on the updated data set, eliminating the sub-forests with relatively large differences, and establishing a plurality of sub-forests for supplementing, and forming a new base forest abnormity detector, thereby realizing updating. Theelectric power scheduling flow data abnormity detecting method can improve power flow data abnormity detecting accuracy.