Accurate Estimation Method of Vibration Signal based on Master-Slave Filtering with Double Colored Noise

Traditional active vibration isolation technology is constrained by sensor noise and limited measurement accuracy, particularly in the suppression of ultra-low frequency vibrations. To address this limitation, this paper integrates the active vibration isolation system model and develops a dual-colored noise master-slave filtering algorithm to achieve more accurate estimation of vibration signals within the studied frequency range. The main filter is built upon Kalman filtering, incorporating a 15th-order autoregressive (AR) process to model colored noise, resulting in an augmented Kalman filter capable of accurate velocity and displacement estimation. The first-level auxiliary filtering algorithm employs a complementary filter, integrating the backward difference model with the velocity sensor to suppress both high- and low-frequency noise, thereby enabling accurate estimation of the base frame's absolute velocity; the second-level auxiliary filtering algorithm is grounded in the principle of minimizing error variation. By deriving an error allocation criterion during the absolute position iteration process, it enables accurate estimation of the absolute positions of both the base and metrology frames. This method significantly enhances the estimation accuracy of vibration signals and effectively improves the performance of the active vibration isolation system.