Rainfall profile detection radar rainfall rate inversion method based on machine learning

The invention provides a rainfall profile detection radar rainfall rate inversion method based on machine learning. The method comprises the steps of obtaining reflection spectrum profile data of a micro-rain radar and rainfall data of a rain gauge in space-time matching; determining rainfall parameters based on the reflection spectrum profile data of the micro-rain radar, wherein the rainfall parameters comprise a raindrop falling end velocity, a raindrop spectrum profile, a backscatter cross section spectrum, a radar reflectivity factor and a liquid water content; inputting the rainfall parameters as a training set and Sigmod as an activation function into a preset rainfall rate inversion model for training, and outputting predicted rainfall data; and with an error between the predicted rainfall data and the actual spatio-temporal matched rainfall data of the rainfall gauge as a target, optimizing the preset inversion model parameters to obtain a final rainfall rate inversion model. The rainfall rate invers