论文部分内容阅读
从计算机模拟和１４Ｎ远程ＮＱＲ实验上研究了神经网络在对已知信号的自动识别问题．证实了多层网络的反向传播算法对解决这一问题的有效性．实验结果表明．在信噪比很低和存在很强的干扰条件下，Ｂ－Ｐ算法仍然保持了很高的识别率（≥９０％）．这在远程ＮＱＲ实际应用中是非常有意义的．
From the computer simulation and the 14N remote NQR experiment, the neural network is used to automatically recognize the known signals. Confirmed the effectiveness of multi-layer network backpropagation algorithm to solve this problem. Experimental results show. The B-P algorithm still maintains a high recognition rate (≥90%) with very low signal-to-noise ratio and strong interference. This is very meaningful in the practical application of remote NQR.