Sensor fusion noise suppression method based on finite impulse response complementary filters

The noise of an inertial sensor is a major issue that limits low-frequency measurement in active control systems. In order to extend low-frequency measurement bandwidth, an optimal finite impulse response (FIR) complementary filter design method is proposed to fuse a displacement sensor and an inertial sensor in this paper. The equivalent relationship between the system noise and the super-sensor noise is analyzed. The mechanism of complementary filters affecting them is clarified. Then, a value function is first constructed as the optimization objective to minimize the cumulative power spectrum (CPS) of the super sensor noise. Based on this and filter amplitude constraints, a convex optimization problem is constructed for solving the optimal filter coefficients. Finally, the experimental results show that the super sensor based on the proposed filters has a wider low-frequency bandwidth. The proposed filters are significantly superior to conventional IIR complementary filters.