Study On N/γ Discrimination Method Based On BP Neural Network

Neutron detection technology is widely used in material exploration,explosive safety detection,radioactive detection of environmental pollution,aerospace,nuclear industry applications and other fields.The wide application of neutron detection technology has led to the rapid development of neutron detectors.However,the commonly used neutron detectors are very sensitive to neutrons and γ-rays,so the primary of neutron detection is to eliminate the interference of γ-rays background after discriminating neutrons and γ-rays.Therefore,it is of great significance to study the discrimination method of neutrons and γ-rays for the application and development of neutron detection technology.Considering that the traditional neutrons and γ-rays signal discrimination algorithms have their own limitations,the BP neural network algorithm can not only realize the classifier function but also solve the limitations of the traditional algorithm.Therefore,the classification function of BP neural network algorithm and pulse shape discrimination technology are used to realize the recognition and classification process of neutrons and γ-rays particle.Firstly,the charge comparison algorithm and frequency domain gradient analysis algorithm are used to preprocess the neutrons and γ-rays pulse signals to judge the types of signals,and the neutrons and γ-rays pulse signals that meet the discrimination conditions of the two discrimination methods selected as the training sample set of BP neural network.By training BP neural network with training samples,the trained neural network can realize the recognition and classification function of neutrons and γ-rays.The accuracy of BP neural network applied to discrimination neutrons and γ-rays is verified by calculating the discrimination error rate of BP neural network.The results show that the discrimination error rate of the neutrons is 2.2%,and that of the γ-rays is 0.4%.The discrimination error rate is small enough,so the accuracy of BP neural network discrimination algorithm is high enough.Finally,by comparing with the discrimination results of charge comparison algorithm,rise time algorithm,frequency domain gradient analysis algorithm and K-means++ clustering algorithm,the effectiveness of BP neural network algorithm in particle species identification and classification in mixed radiation field is further proved.The training samples confirmed in the implementation process of BP neural network discrimination algorithm not only take into account the time domain and frequency domain waveform characteristics of pulse waveforms,but also solve the problem of insufficient training samples in the training process of neural network.What’s more,this paper can also provide basis for the selection and optimization of discrimination methods in actual discrimination work.The discrimination algorithm based on BP neural network can give consideration to the time domain and frequency domain characteristics of signals at the same time.The processing time for discriminating the same number of mixed signals is the fastest,and the discrimination error rate of the neutrons is 2.2%.The discrimination error rate of the γ-rays is 0.4%.Therefore,BP neural network can identify and classify neutrons and γ-rays well,which can provide a solid foundation for the development and application of neutron detection technology.