Method of noise filtering in gyroscopic sensor signal based on adaptive alpha-beta Kalman filter

To ensure flight stabilization of a multirotor UAV, it is necessary to process data from micromechanical sensors installed on board and convert them into control signals. During operation, the UAV is affected by many external and internal factors, which leads to the occurrence of noise in these types of sensors. Noise has a negative impact on the operation of the device control system. The gyroscope signals, which are used to calculate the angular velocities and tilt angles of the vehicle, are particularly susceptible to noise. To solve this problem, the paper presents a method for cleaning the signal from a gyroscopic sensor for the differential component of the PID controller. The developed method is implemented as an interconnected adaptive system of two filters and allows for cleaning the original signal from the gyroscopic sensor from high-frequency electrical noise, vibrations caused by motors and resonant oscillations of the UAV frame. During the conducted field experiments, the signal/noise metric for the processed data reached a maximum value of 8,40 dB, which is significantly better compared to the 2 dB figure found in other scientific papers.