【摘要】 通过制作理论模型，提取出时域、频域内最能反应薄层厚度的多种参数：反射波的总能量、视振幅、正半周面积、负半周面积、相关函数第１个零点的时移值、振幅谱面积、主频、中心频率、频带宽度、频谱的二阶矩及三瞬剖面等，它们与薄层厚度的关系都是非线性的。采用ＢＰ法神经网络，通过对模型数据的学习、记忆、识别，预测薄层厚度，取得了较满意的结果。更多还原

【Abstract】 Representative parameters which respond to thin-bed thickness in time domain and frequency domain are derived by making theoretical models.They are total energy of reflected waves, visible amplitude, plus half-periodic area,minus half-periodic area,the time-shift corresponding to the first zero point of the correlationfunction,area of the amplitude spectrum,dominant frequency, center frequency,spectralband width,the second moment of the spectrum and three-instantneous section,etc.The relations between these parameters and the thin-bed thickness are not linear.We predict the thin-bed thickness by using back propagation(BP)neural network to learn,memorize,discriminate to model data,and the results are satisfactary.更多还原