中文摘要：

结构在共振区附近,由于频响矩阵病态问题使得随机载荷识别结果误差较大。针对此问题,本文提出了两步加权算法来提高识别精度。首先基于率谱谱分解阐述了随机载荷识别的基本过程,根据谱分析的结果分析了识别误差来源的影响参数,接着引入两步加权思想,提出新方法完成识别过程。该方法能在一定程度上减小频响函数病态问题,提高识别精度。仿真对比结果验证了方法的识别效果。最后,依据误差对比发现的问题,提出了避免病态的建议。

英文摘要：

Random load identification results come up with large error because of ill-conditioned frequency response function near the resonance region. Aiming at this problem, a new two-step weighted algorithm is proposed in this paper to improve the identification accuracy. Firstly, the basic process of random load identification is expounded based on power spectral spectrum decomposition; then the source of the error influence parameters are analyzed according to the result of spectrum analysis and the novel method is put forwarded to complete the identification process by introducing the two-step weighted idea. The method can ease ill-posed problem and improve identification accuracy to some extent. The simulation results validate the effect of method above. Finally, according to the error comparison, some suggestions are put forwarded to avoid the ill-posed problem.