论文部分内容阅读
在对诱发铀部件裂变信号的测量原理及特点分析的基础上,开展了基于BP神经网络的诱发铀部件裂变时间关联信号特征参量分析处理的研究工作。采用无偏估计方法,计算信号的自相关函数和互相关函数,再利用比较法和导数法两种特征量提取方法,提取出不同状态下裂变信号的特征参量,借助于BP神经网络模式识别应用原理进行训练和预测。理论分析和研究结果表明:基于比较法和导数法获得的特征参量能较好地反映诱发铀部件裂变信号的特征;用BP神经网络对裂变信号进行模式识别,取得了较高的正确率,验证了此方法的有效性和合理性。
Based on the analysis of the measuring principle and characteristics of fission signals induced uranium components, a research work on the analysis and processing of characteristic parameters of fission time correlated signals of induced uranium components based on BP neural network was carried out. The unbiased estimation method is used to calculate the autocorrelation function and the cross correlation function of the signal. Then the comparison method and the derivative method are used to extract the characteristic parameters of the fission signal under different states. With the help of BP neural network pattern recognition Principles for training and prediction. Theoretical analysis and research results show that the characteristic parameters obtained based on comparative method and derivative method can well reflect the characteristics of induced fission signals of uranium parts. The pattern recognition of fission signals is carried out by BP neural network, and a high correctness rate is obtained. The validity and rationality of this method.