GNSS perceptible water vapor pressure prediction method and device based on multilayer perceptron, and medium

The invention provides a GNSS perceptible water vapor pressure prediction method based on a multi-layer perceptron algorithm. The GNSS perceptible water vapor pressure prediction method comprises the steps that a troposphere delay time sequence and a perceptible water vapor pressure data set are obtained through calculation according to GNSS observation data flow; on the basis of the troposphere delay time sequence and the perceptible water vapor pressure data set, abnormal value detection and missing data are adopted for data preprocessing, and the preprocessed troposphere delay time sequence and the perceptible water vapor pressure data set are obtained; taking the troposphere delay time sequence and the observation station spatio-temporal information data set as input parameters, taking the perceptible water vapor pressure as an output parameter, and training a multi-layer perceptron neural network; gNSS troposphere delay solved in real time and observation station space-time information data are accessed