为降低结构健康监测加速度信号中常见的白噪声以及脉冲噪声，提出了中值滤波与小波阈值降噪相结合的方法。分解层数对降噪效果有着重要影响，为取得更好的降噪效果，提出了一种分解层数自适应确定法，并给出了各层阈值的取值方法。数值模拟以及国家游泳中心健康监测系统实测数据分析结果表明，所提出的方法克服了传统阈值降噪法对脉冲噪声拟制效果不理想的缺陷，能同时有效降低白噪声以及脉冲噪声，为模态参数识别等提供更高信噪比的分析数据。

In order to reduce the white noise and pulse noise in the acceleration signal of civil engineering structural health monitoring, a combinative method of medial filter and wavelet threshold is proposed. Because of the decomposition level has a significant impact on the result of denoising, in order to get a better denoisng result, a self-adaptive method to select the decomposition level is proposed , the method to determine the value of threshold on each level is also been given in this paper. The analysis of numerical simulation and the on-site measured data from the structural health monitoring system of National Aquatics Center of Beijing 2008 Olympic venues shows that, the method proposed in this paper can overcome the defect of wave threshold method which doesn’t have obvious effect to reduce the pulse noise, it can reduce the white noise and pulse noise at the same time, provide data of higher signal to noise ratio for modal parameter identification and so on.