The techniques of minimizing losses in a smart grid system need a fast algorithm to estimate the conditions of the active distribution system. Excessive losses threat the reliability and security of the smart grid system. This paper presents a novel method for estimating power loss in a real-time of each line in the active distribution system. The proposed method, which is called, a neural network power loss estimation (NN-PLE) is a computational method for estimation the line losses using an artificial neural network. The proposed method provides a fast calculation with high accuracy comparing to other traditional methods that take a very long execution time. Simulation results are presented to demonstrate the performance of (NN-PLE) for a 33-bus distribution system with different data resolutions.