石英振梁加速度计的小波神经网络温度补偿研究

捷联惯导系统中使用的石英振梁加速度计随温度输出漂移比较显著，通过理论分析和试验研究了其静态定点温度特性，提出了在处理受温度影响的加速度计数据时建立基于梯度下降学习算法的小波神经网络模型，并对其进行补偿。通过惯导系统初始对准实验结果表明，该方法与传统的最小二乘法相比，小波神经网络的非线性逼近能力更强，曲线拟合精度更高，能有效补偿加速度计的温度漂移，降低惯导系统初始对准后的姿态解算误差。

Temperature Compensation of Quartz vibrating Beam Accelerometer Based on Wavelet Neural Network

Quartz vibrating beam accelerometer used in the strapdown inertial navigation system drifts significantly with the temperature,through theoretical analysis and experimental research on its static point temperature characteristic,we put forward to establish a model of wavelet neural network based on gradient descent algorithm in the treatment of the affected by the temperature of the beam accelerometer data,and to carry on the compensation. Initial alignment by inertial navigation system experiments show that wavelet neural network ’s nonlinear approximation ability is stronger,and its curve fitting accuracy is higher, which can effectively compensate the temperature drift of accelerometer and reduce error of attitude algorithm after the initial alignment of inertial navigation system.