中文摘要：

系统响应可表示为单位脉冲响应函数与激励载荷的卷积，将其离散化一组线性方程组，则载荷识别问题即转化为求解线性方程组的反问题。针对响应中带有噪音时载荷识别的困难，提出了联合奇异熵去噪修正和正则化预优的共轭梯度迭代识别方法。一方面对含噪信号进行基于奇异熵的去噪处理，提高反问题求解中输人数据的精度。另一方面利用正则化方法对共轭梯度迭代算法进行预优，改善反问题的非适定性。由于从输入的响应数据去噪和正则化算法两方面同时改善动态载荷识别反问题的求解，因此可以有效地抑制噪声，提高识别精度。通过数值算例分析，表明在不同的噪声水平干扰下，其识别精度均优于常规的正则化方法，能够实现有效稳定地识别动态载荷。最后通过实验研究进一步验证了该方法的正确性和有效性。

英文摘要：

Due to the difficulty of measuring directly the dynamic load with the increasing of complexity of structures and excita- tions, it is therefore of great importance how to identify accurately the dynamic load by indirect method. As is well known, , the response of the linear system can be obtained by the convolution integral of the unit pulse response function and the dynamic loads. Hence, by discretizing the convolution integral into a set of linear algebraic equations, the load identification problem can be transformed into an inverse problem of solution of linear algebraic equations. To smooth out the difficulty of the load identi- fication in the presence of noises, a dynamic load identification method is proposed based on the combination of the singular en- tropy denoising method and the regularization--preconditioned conjugate gradient iteration method, of which the former is to improve the accuracy of input data in the process of load identification, and the latter is to avoid ill--posedness of the inverse problem.. The numerical simulation result for a given example indicates that the proposed methods are more accurate and stable in the identification of dynamic loads with the interference of different noise levels, compared with the conventional regulariza- tion methods. Finally the validity and the effectiveness of the proposed methods are demonstrated by experiment.