The invention discloses a power consumption data abnormity detection and a model training method and device, and the model training method comprises the steps: obtaining the training power consumptiondata of power consumption equipment, and obtaining a training power consumption data set; extracting first training feature information and second training feature information corresponding to each piece of training power consumption data in the training power consumption data set, the first training feature information being used for indicating power consumption features of each piece of power consumption equipment, and the second training feature information being used for indicating analysis result features of the power consumption data; and training the neural network model by using the first training feature information and the second training feature information to obtain a power consumption data abnormity detection model. By implementing the method and the device, the electricity consumption data predicted value is predicted by utilizing the model, the difference value is calculated with the observed value, and the difference value is compared with the selected threshold valueto judge whether the electricity consumption data is abnormal or not. The long-term and short-term memory neural network is used for predicting the power consumption data, the data prediction accuracyis improved, and the calculation efficiency of processing large-scale data is improved.