基于有色噪声Kalman滤波的桥梁自振频率提取模型研究

RESEARCH ON EXTRACTING MODEL OF BRIDGE AUTOOSCILATION FREQUENCY BASED ON KALMAN FILTING WITH COLORED NOISE

在分析有色噪声卡尔曼滤波的前提下,针对GPS动态测量的时间序列,分析误差序列的时间延迟,利用高斯-马尔科夫模型对其做有色量测噪声处理,从而得到去除相关噪声的时间序列值。将滤波残差序列分为两部分,每部分的长度均为5分钟,基于FFT方法对滤波残差信号进行频谱分析,获得误差信号的频率谱,对比两部分误差序列的频率谱,获取桥梁的自振频率,并通过FFT分析加速度数据,得到加速度的频率谱,进而验证基于GPS动态变形数据提取的桥梁低频动态特征。

Under the premise of analysis of Kalman fihering with colored noise, aiming to the time series of GPS kinematic surveying, the time delay of the error sequence is analyzed and the Gauss-Markov model is introduced to process it with colored measurement noise, thus the time series without correlated noise was obtained. The time series of the residuals is divided into two sessions with each session period lasting for 5 minutes and then the spectrum analysis of the residual signal is made with FFT. Finally, the frequency spectrum of the error signal is obtained, and then the two frequency spectrum are compared with each other to obtain the natural frequency of the bridge, and the acceleration data is analyzed with FFT to gain the frequency spectrum to confirm the low-frequency dynamic features extracted with GPS dynamic data.