Nuclide spectrum whole spectrum identification method and system based on recurrent neural network, and medium

The invention discloses a nuclide spectrum whole spectrum identification method and system based on a recurrent neural network, and a medium, and belongs to the technical field of nuclide spectrum identification. Comprising the following steps: S1, obtaining first nuclide spectrum data after background radiation is deducted; s2, dividing the first nuclide spectrum data into multiple segments, and reserving a channel address position corresponding to each energy value in the first nuclide spectrum data to obtain second nuclide spectrum data; s3, transmitting the second nuclide spectrum data into a trained LSTM-KAN model, extracting data features of the second nuclide spectrum data through an LTSM layer, and then performing nuclide identification through a KAN layer; and S4, outputting an identification result of the second nuclide spectrum data to judge the nuclide type of the first nuclide spectrum data. A complete nuclide spectrum is segmented, and each segment is identified once, so that multi-nuclide ident