rbf_mems

  为了提高使用精度，研究了某型号MEMS陀螺仪的随机漂移模型。采用游程检验法分析了 该陀螺仪随机漂移数据的平稳性，并根据该漂移为均值非平稳、方差平稳的随机过程的结论， 采用梯度径向基（RBF）神经网络对漂移数据进行了建模。实验结果表明：相比经典RBF网络模 型而言，这种方法建立的模型能更好地描述MEMS陀螺仪的漂移特；相对于季节时间序列模型而 言，其补偿效果提高了大约15％。(In order to improve accuracy, to study a particular model of the MEMS gyroscope random drift model. Using run-length analysis of the test gyro random drift data stationarity, and in accordance with the drift for the average non-stationary, the variance of the random process a smooth conclusion, the use of gradient radial basis (RBF) neural network drift data to build mode. The experimental results show that: compared to the classical RBF network model, this method of establishing a model to better describe the MEMS gyroscope drift special compared with the seasonal time series model, the effect of their compensation increased by approximately 15.)

梯度RBF神经网络在MEMS陀螺仪随机漂移建模中的应用.pdf