Crack monitoring data anomaly classification and identification method

The invention provides a crack monitoring data anomaly classification and identification method. The method comprises the steps of segmenting an original data segment into N sub-data segments based on an absolute value of a difference value of all two adjacent data in the original data segment; the sub-data segments with the length larger than 24 serve as initial normal sub-data segments, and the sub-data segments with the length smaller than or equal to 24 serve as suspected abnormal sub-data segments; combining the adjacent suspected abnormal sub-data segments to obtain combined suspected abnormal sub-data segments; respectively judging that each merged suspected abnormal sub-data segment is in a flying spot abnormal state, a back-and-forth jumping abnormal state or a continuous jumping abnormal state; and respectively judging that two adjacent preliminary normal sub-data segments or two preliminary normal sub-data segments separated by the flying spot abnormal form are in a step abnormal form or a normal d