论文部分内容阅读
研究了小波变换时频特性的信号识别及其在复合材料损伤检测中的应用．根据小波变换的框架重构理论及时频相空间理论，提取信号的时频域特征，通过比较原信号的时频空间和小波变换相空间的相同部分，得到能反映同样时频特征的小波级数展开项和的个数，并用误差函数的最小化提取能反映时频性质的小波系数．以此作为小波神经网络的学习参数，经过学习后，使之能对信号进行识别．应用此方法对复合材料试验过程中的复杂曲线进行了实验识别，效果很好．从小波时频特性提取的信号特征，在时间和频率方面都能体现原信号所包括的本质信息，供助Ｂ样条小波神经网络的识别结果，达到了预期目的．
The signal identification of time-frequency characteristics of wavelet transform and its application in composite material damage detection are studied. According to the frame reconstruction theory of wavelet transform and the theory of time-frequency phase space, the time-frequency feature of the signal is extracted. By comparing the time-frequency space of the original signal with the same part of the phase space of the wavelet transform, the wavelet series Expand the sum of the number of terms, and use the error function to extract the wavelet coefficients that reflect the time-frequency properties. As a wavelet neural network learning parameters, after learning, so that it can identify the signal. Applying this method to experimentally identify the complex curves of composite materials has good effect. The signal characteristics extracted from the time-frequency characteristics of wavelet can reflect the essential information included in the original signal in terms of time and frequency, and help the identification results of B-spline wavelet neural network to achieve the expected purpose.