Network flow abnormality detection method based on adjustable segmented Shannon entropy

The invention discloses a network flow abnormality detection method based on adjustable segmented Shannon entropy and mainly provides a network flow abnormality detection method based on adjustable segmented Shannon entropy and suitable for the abnormality detection requirement of a large-scale network. The detection method disclosed by the invention comprises the following specific steps: selecting an original sample space; obtaining high probability entropy of a high probability sample space and low probability entropy of a low probability sample space based on an adjustable segmented entropy implementation method; respectively judging whether the high probability entropy and the low probability entropy are abnormal, if the high probability entropy is smaller than the preset high probability entropy, the high probability entropy is abnormal, otherwise the high probability entropy is normal; if the low probability entropy is larger than the preset low probability entropy, the low probability entropy is abnormal, otherwise the low probability entropy is normal; and determining that the sample space which corresponds to the abnormal entropy is a network flow abnormality sample space, namely determining that the network flow abnormality occurs.