Marine radioactive gamma energy spectrum preprocessing method based on generative adversarial network

Abstract

The invention discloses a marine radioactive energy spectrum resolution optimization method. The method comprises steps of carrying out the training of a generative adversarial network for a marine radioactive gamma energy spectrum, reducing the full width at half maximum of an energy peak through the trained generative adversarial network, and carrying out the peak searching nuclide recognition test. According to the method, marine radioactive gamma-ray energy spectrum processing is carried out on the energy spectrum based on the generative adversarial network, the calculation speed is high, nuclide identification accuracy is high, the test result is stable, and the method is not easily influenced by shielding substances.

Description

technical field [0001] The invention relates to an energy spectrum optimization method, in particular to an optimization processing method for marine radioactivity gamma energy spectrum. Background technique [0002] It is very necessary to monitor the radioactive pollution in the coastal waters of nuclear power plants. Studies have shown that when domestic nuclear power plants are in normal operation, radioactive substances that are significantly higher than the environmental background can be detected in the seawater of the surrounding sea areas. The key to monitoring the marine radioactive environment is the analysis of the low-activity energy spectrum. High-resolution spectrometers (such as high-purity germanium) cannot be deployed for real-time monitoring at sea. It is difficult to distinguish correctly and needs to be analyzed by complex spectral analysis methods. Traditional methods use deconvolution methods (such as GOLD deconvolution) to improve the resolution of c...

AI Technical Summary

Problems solved by technology

[0003] Aiming at the low-activity gamma energy spectrum of marine radioactivity, the present invention proposes an energy spectrum preprocessing method based on the confrontation generation network, and performs

Images