RBF神经网络在核能谱平滑中的应用

在γ能谱中,为对核素进行可靠的定性、定量分析,需对原始能谱进行平滑滤波。径向基函数(Radical Basis Function,RBF)由于具有网络全局逼近性质和最佳逼近性能,对核能谱进行函数逼近。根据平滑要求和处理速度,建立不同函数的RBF神经网络对核能谱进行平滑滤波。实验结果表明,RBF神经网络方法能达到更为灵活和合理的平滑滤波。

An Application of RBF Neural Networks in Nuclear Spectrum Smoothing Filtering

In order to make a reliable quantitative and qualitative analyses in γ spectrum, to do the smoothing filter arithmetic for the original spectrum is important. Due to global approximation properties and optimize ap-proximation performance of Radical Basis Function( RBF) , the function approximation is applied on nuclear en-ergy spectrum. According to the requirements of smoothing and processing speed, different functions of RBF neural network is established to do the smoothing filter arithmetic for nuclear energy spectrum. The experiment results show that the smoothing filter arithmetic can be more flexible and more reasonable by the method of RBF neural network.