一种应用于三轴稳定器中的陀螺仪漂移抑制方法

Gyroscope Drift Suppression Method Applied in Three-axis Stabilizer

针对三轴手持稳定器中陀螺仪随机漂移误差逐渐积累的问题,首先对陀螺仪误差分析进行误差建模,并使稳定器3个轴的角速度作为RBF神经网络算法的输入神经元值,通过神经网络算法自适应地调整3个轴的输出,之后再将输出值进行卡尔曼滤波降噪,最终解决三轴手持相机稳定器中微机械陀螺仪的随机误差漂移问题。仿真实验证明,此种方式可以达到三轴手持相机稳定器系统在锁定状态下的最大漂移角速度为0.1°/s,锁定模式下的航向漂移角度在±5°之间的目标精度,将神经网络方法与卡尔曼滤波结合来抑制陀螺仪的随机漂移是可行的。

Aiming at the problem that the random drift error of the gyroscope gradually accumulates in the three-axis handheld stabilizer.Firstly,the gyroscope error is modeled,then the angular velocity of the three-axis of the stabilizer is used as the input neuron value of an RBF neural network algorithm,and the output of the three-axis is adaptively adjusted by the neural network algorithm.After that,the output value is subjected to Kalman filter to reduce noise.The random error drift problem of the new generation of micro-machined gyroscope in the three-axis handheld camera stabilizer is solved.The simulation results show that this method can achieve the maximum drift angular velocity of the three-axis handheld camera stabilizer system in the locked state of 0.1°/s,and the heading drift angle in the locked mode is within the target accuracy of ± 5°,which proves that it is feasible to combine the neural network method with Kalman filtering to suppress the random drift of the gyroscope.