论文部分内容阅读
为了对微小型飞行器上的MIMU(微惯性测量单元)的随机漂移进行补偿,在比较了Mallat算法与à trous算法之后,基于小波变换与多尺度分析方法,提出了多尺度时间序列建模方法,它充分利用了à trous算法的快速性与时间平移不变性,将MEMS陀螺仪随机漂移进行多尺度分解。对各尺度上分解得到的信号进行重建,并对重建得到的各个信号进行时间序列建模。将各尺度时间序列模型的预测输出的和作为陀螺仪的随机噪声估计,对陀螺仪的随机漂移进行补偿。最后的实际数据建模表明该建模方法运算量小、建模速度快、精度高、模型适用性强,有很强的实际应用价值。
In order to compensate for the random drift of MIMU (Micro Inertial Measurement Unit) on a micro-aircraft, a multi-scale time series modeling method based on wavelet transform and multi-scale analysis is proposed after Mallat algorithm and à trous algorithm are compared. It makes full use of the fastness and time shift invariance of à trous algorithm, and randomizes the MEMS gyroscope for multi-scale decomposition. The signal decomposed on each scale is reconstructed, and each signal reconstructed is modeled in time series. The sum of the predicted output of each scale time series model is used as a random noise estimation of the gyroscope to compensate for the random drift of the gyroscope. The final actual data modeling shows that the modeling method has the advantages of low computational complexity, fast modeling speed, high precision, strong applicability of the model and strong practical value.