The invention discloses a method for finding an abnormal
electric energy meter based on a gray GM (1, 1) model. The method comprises the following steps that first, normally-measuring power utilization data of a user
electric energy meter are obtained from a user power utilization
system database; second, obtained data samples are arrayed according to the
time sequence,
original data are accumulated to form new regular data, a linear first-order
differential equation is established for the new data, the model is subjected to identifying and parameter
estimation, a prediction equation is obtained, and finally a short-term predicted value is obtained by regressive restoring; and third, an obtained load predicted value and a value collected by the user
electric energy meter are compared, whether the user electric energy meter is normal is judged, and if an anomaly exists, a user is suspected of stealing
electricity. The method for finding the abnormal electric energy meter based on the gray GM (1, 1) model has the advantages that a large number of samples are of no need, the high regularity of the samples is of no need, computing
workload is low, operation cost is low, a mounted electric energy meter does not need to be transformed, manpower investment of an
electric power company is lowered, and judging is accurate.