An Algorithm of Nuclide Number Identification for Environmental Radioactivity Monitoring

The traditional nuclide identification algorithm based on the characteristic peak has high requirements for users’ specialization, many steps and slow recognition speed. In recent years, although the new methods such as artificial neural network and Bayesian theory based on full spectrum analysis have high recognition speed and low requirements for user specialization, they rely on a large number of training samples, and the accuracy of measurement results will be affected when the measurement conditions of training samples and measured samples deviate. To solve this problem, an algorithm of nuclide number identification based on singular value method is designed. The algorithm is to analyze the time series of the collected data, which is mainly divided into the steps of acquiring array signals, constructing incidence matrix, calculating and sorting eigenvalues. The experiments show that the nuclide recognition time is short and the recognition is accurate. It can provide fast and accurate information in case of abnormal nuclear radiation.