基于信息矩阵和神经网络的振动陀螺动态建模

Dynamic Modeling of Vibratory Gyro Based on Information Matrix and Neural Network

为了分析半球谐振陀螺的动态响应过程、得到陀螺的动态性能参数,推进半球陀螺的实用化水平,首先分析了反馈式半球谐振陀螺的工作原理。然后构造了由输入输出量组成的信息矩阵,给出了模型系数估计的最优目标函数,对信息矩阵进行Household变换求解最优目标函数并得到了动态响应模型系数;使用神经网络搜索最优目标的思想,以函数拟合度为评判标准确定了动态模型阶次,最终得到了三阶的传递函数作为该陀螺的动态响应函数。函数可以很好地反映陀螺的动态响应过程。

In order to analyze dynamic response process, acquire dynamic performance parameter and advance the practical level of HRG, operational principle of the gyro is analyzed firstly. Then information matrix consisting of input-output is constructed and the optimal objective function used to evaluate model coefficient is given. Household transform approach is employed to solve the optimal objective function and at the same time model coefficients are acquired. The thought of neutral network searching the optimal objective is used to confirm the order of the model and function goodness-of-fit is regarded as evaluation standard. At last, a transfer function with three order is confirmed as dynamic response function, and this function can greatly reflect dynamic response of the gyro.