The invention provides an intelligent electricity stealing criterion method and device based on power grid parameter key feature extraction, and the method comprises the steps: firstly, calculating d-distance and d-outlier local density of a user by using monthly electricity consumption data of user, and further calculating d-distance outlier factor and d-distance neighborhood transformation ratiofactor of the user; then, using user monthly electricity consumption data, d-distance outlier factor and d-distance neighborhood transformation ratio factor as electricity stealing user identification characteristic input quantities and training neural network and obtaining parameters for analyzing electricity stealing suspicion of the user; and finally, inputting to-be-detected user power consumption data into the trained neural network. The feature design of the electricity stealing user detection method based on the neural network and the design of the neural network are simple and feasible. The influence caused by data imbalance in the electricity stealing identification technology is effectively solved. The precision and generalization ability of the electricity stealing identification model based on the neural network are improved.