An important property of wavelet multiresolution analysis is the capability to represent functions in a dynamic multiscale manner, so the solution in the wavelet domain enables a hierarchical approximation to the exact solution. The typical problem that arises when using Daubechies wavelets in numerical analysis, especially in finite element analysis, is how to calculate the connection coefficients, an integral of products of wavelet scaling functions or derivative operators associated with these. The method to calculate multiscale connection coefficients for stiffness matrices and load vectors is presented for the first time. And the algorithm of multiscale lifting computation is developed. The numerical examples are given to verify the effectiveness of such a method.

The method of constructing any scale wavelet finite element (WFE) based on the one-dimensional or two-dimensional Daubechies scaling functions was presented, and the corresponding WFE adaptive lifting algorithm was given. In order to obtain the nested increasing approximate subspaces of multiscale finite element, the Daubechies scaling functions with the properties of multi-resolution analysis were employed as the finite element interpolating functions. Thus, the WFE could adaptively mesh the singularity domain caused by local cracks, which resulted in better approximate solutions than the traditional finite element methods. The calculations of natural frequencies of cracked beam were used to check the accuracy of given methods. In addition, the results of cracked cantilever beam and engineering application were satisfied. So, the current methods can provide effective tools in the numerical modeling of the fault prognosis of incipient crack.

小波变换是非平稳信号处理的有力工具，小波性质取决于它的基函数。在模态分析、机械监测诊断等领域,冲击响应信号十分普遍。一种单边衰减复指数型的Laplace小波基函数具有分析冲击响应信号的优势，采用Laplace小波相关滤波方法可提取振动信号中的冲击响应分量，成功地诊断出内燃机进气阀的磨损故障和识别出大型水轮机轴系的一阶固有频率。

In this paper we present an experimental investigation of the identification of crack location and size. By providing the first three natural frequencies through vibration measurements, curves of crack equivalent stiffness versus crack location are plotted, and the intersection of the three curves predicts the crack location and size. In the experiments, the cracked specimens were made using a wire-cut electrical discharge machine, and the cantilever beams were excited next to the free end by means of an impulse force hammer. In order to obtain the accurate natural frequencies from the transient signal measured, the method of zoom fast Fourier transform is adopted to improve frequency resolution. From experimental results, it is observed that the identification errors of crack location and size are less than 2% and 4%, respectively. The effectiveness of crack identification through vibration measurements is verified.

为了克服在无先验知识的情况下，人为选择时域无量纲指标作为故障敏感特征的盲目性，提出了一种基于特征评估和径向基函数（RBF）神经网络的机械故障诊断模型。该模型分别采用小波包和经验模式分解方法对原始振动信号进行分解，分别提取原始信号和各分解信号的时域无量纲指标组成联合特征，然后对联合特征进行评估，计算评估因子，并根据评估因子的大小选取敏感特征作为RBF神经网络的输入，实现对机器不同状态的自动识别。实验结果和工程应用表明，这种集成了小波包、经验模式分解、特征评估方法和RBF神经网络的机械故障诊断模型能够精细地获取故障信息，从大量的故障特征中筛选出敏感特征，因而减小了网络规模，提高了分类准确率，具有很强的鲁棒性。