Method and system for detecting agricultural pollutants in water environment

The invention discloses a method and a system for detecting agricultural pollutants in a water environment. The method comprises the following steps: acquiring Raman spectrum data of a to-be-detected water sample in a target water environment area, and preprocessing the Raman spectrum data; the method comprises the following steps: constructing a training set through Raman spectrum data of different agricultural pollutants, and constructing a detection model based on an LSTM network to qualitatively estimate the agricultural pollutants in a to-be-detected water sample; performing pollutant concentration prediction on the water sample to be detected by using the convolutional neural network, and performing agricultural pollutant evaluation on the target water environment area; and acquiring environment characteristics and geographic characteristics of the target water environment area, determining main influence factors of agricultural pollutants in the target water environment area according to a pollutant ev