论文部分内容阅读
该文给出了一种利用Ｌｅｖｅｎｂｅｒｇ－Ｍａｒｑｕａｒｄｔ最优化算法来改进ＢＰ神经网络的方法。通过仿真试验，证明了改进的ＢＰ算法具有收敛速度快和精度高的特点。为了进一步提高ＢＰ网络拾取地震波初至的抗干扰能力，该文提出了应用小波变换对地震波信号进行预处理来消除噪声影响。该方法经实际应用取得了满意的效果。
This paper presents a method to improve BP neural network by using Levenberg-Marquardt optimization algorithm. Through simulation experiments, it is proved that the improved BP algorithm has the characteristics of fast convergence and high precision. In order to further improve the anti-jamming ability of the BP network at the beginning of picking up the seismic wave, this paper proposes to apply the wavelet transform to pre-process the seismic wave signal to eliminate the influence of noise. The method has achieved satisfactory results through practical application.