In order to solve the problem that the recognition rate of traditional nuclide recognition methods is low due to the lack of strong adaptability, a nuclide recognition prediction model based on back propagation (BP) neural network is established. Taking the measured signals of seven nuclides of Americium, Cadmium, Plutonium, Radon, Palladium, Cobalt and Cesium as an example, the nuclide identification model is established. The results show that the recognition model can quickly and accurately identify the above nuclides, and has a broad application prospect.