The invention relates to an electricity consumption abnormity detecting method based on an electricity consumption deviation rate for a low voltage consumer. The electricity consumption abnormity detecting method comprises the steps of: step (1), acquiring actual electricity consumption of the low voltage consumer on a day to be detected, and acquiring predicted electricity consumption of the lowvoltage consumer by adopting a neural network prediction technology; step (2), acquiring the electricity consumption deviation rate through comparing the predicted value and the actual electricity consumption; step (3), calculating weights w1, w2 and w3 of the electricity consumption deviation rate on the day to be detected and two days before the day to be detected by adopting an analytic hierarchy process; step (4), acquiring electricity consumption deviation rates on two days before the day to be detected, and calculating an electricity consumption abnormity rate according to the weights; step (5), and comparing the electricity consumption abnormity rate with a set threshold value, and determining that the electricity consumption of the low voltage consumer is abnormal if the electricity consumption abnormity rate exceeds the threshold value. The electricity consumption abnormity detecting method adopts the weight method to combine the electricity consumption deviation rates on theday to be detected and two days before the day to be detected, so as to prevent a detection error caused by a mechanical failure of an electric meter of the user or sudden electricity consumption, thus the final detection result is more precise.