The invention provides a GNSS multipath effect correction method based on a BP neural network technology. The method sequentially comprises the following steps of multipath error extraction, sample pre-selection, sample preprocessing, sample training, network topology determination, data storage, data post-processing and model updating, wherein modeling is carried out on multipath errors by meansof forward propagation of signals and reverse propagation of the errors of a BP neural network algorithm, a motion condition of a satellite, a satellite elevation, a signal-to-noise ratio and multipath effect errors caused by ambient conditions around monitoring points can be effectively weakened, and meanwhile, periodic fluctuations in directions X, Y and Z of a monitoring station are remarkablyimproved through trained results, a model is dynamically updated according to the new multipath error data, and the periodic errors caused by multipath effects on the monitoring points can be weakenedto the maximum extent.