在结构振动测试中，由于测试环境影响、试验设备、操作人员以及模态分析过程中的简化，试验获取的结构模态参数中包含了不确定因素。为了正确的估计结构物理参数的统计特性，本文在贝叶斯估计方法中选择共轭先验分布，结合试验数据导出后验分布，得到了结构固有频率均值及方差的估计值。对一个4层钢筋混凝土框架结构模型进行了振动测试。试验结果表明：在不能获得大量样本信息时，充分利用先验信息，结合现场测试数据，对频率参数进行估计，能够有效地减小均值与方差的可信区间长度，具有工程实用参考价值。

In modal experiment of a structure, because of environment effect, instrument equipment, human error in operating, and the simplification in modal analysis, the uncertainty factors were involved into the measured modal parameters. In order to evaluate the statistical characteristics of physical parameter correctly, it is necessary to take the uncertainty of modal parameters measured in testing into account. Based on Bayesian evaluatiom, in this paper, a conjugate prior probability density function is seclected, the posterior probability density function is induced combining with test data, and then the mean and variance of the natural frequencies were obtained. A modal experiment was conducted on a four-story reinforced concrete frame structure. The test results show that when a large of samples is not available, the length of confident interval of mean and variance are shortened obviously by taking full use of prior information and test data in field to evaluate the uncertainties of natural frequencies, that is meaningful in practice.