给出了一种利用小波分解求取激光陀螺分形噪声参数的方法。采用dB10小波,分别对中精度和高精度激光捷联导航系统(LINs)100Hz导航数据做7次小波变换,估计出了两种陀螺中分形噪声的分形指数γ分别约为1和2;试验说明,在导航频率下,激光陀螺噪声表现出的不是随机游走,而是一种零偏不稳定性和角速率随机走,并对这一现象的成因进行了深入讨论。为了减少这种不稳定性,从激光陀螺设计制造角度,应提高光源输出功率和镀膜质量;从信号处理角度,应设计适合LINS的数字滤波算法。最后给出了利用FIR滤波器凹点作为陷波器的参数设计结果,实测数据证明,在阶数增加很少的情况下,实现了对3个陀螺抖动频率同时陷波,提高了激光陀螺数据输出精度。
A method of calculating fractal noise parameters by wavelet is developed. The signals of middle and high-precision laser inertial navigation system (LINS) are decomposed for 7 times by dB10 wavelet. The 100 Hz data sampling frequency is employed in both systems. The parameters of fractal noise are estimated. The fractal exponent 2' approximately equals 1 in high-precision laser gyro and about equals 2 in middle-precision laser gyro. The experiments further indicate that the noise in laser gyro signal is not angel random walk but bias instability and rate random walk. This conclusion is different from common paper. Two advices are presented in order to reduce the fractal noise. The laser output power and the technique for deposition should be improved during the laser gyro design and production. And good filter algorithm should be used in LINS when more laser gyros are applied. A new finite impulse response (FIR) filter example is introduced at last. It has three notches to decrease influence of three gyro jither at the same time.