基于动态神经元网络的激光陀螺输出误差模型

惯性敏感器误差补偿技术对提高武器装备的性能具有重要的意义，而误差补偿的关键在于误差模型的辨识。由于动态神经元网络是在前馈网络的节点引入前馈和反馈环节，理论上已证明其具有很强的动态逼近能力，可用来描述任意的非线性动态系统。根据惯性敏感器误差的动态特性，本探讨将动态神经元网络引入到激光陀螺误差建模中去，详细介绍了网络结构和对应的动态梯度算法。通过仿真算例说明，动态神经元网络在激光陀螺输出误差建模时具有一定的优点：网络收敛速度快、较好的跟踪性能、稳定性好。

Error Model of Laser Gyro Output Based on Dynamic Neural Unit Networks

It's important to improve the performance of weapon for compensating the errors of inertial sensors.Identification of error model is the key to compensating errors.It's theoretically proved that the dynamic neural unit networks have good power of approximating random nonlinear dynamic system,because of using feedforward and feed back loop in the nodes of forward networks.According to the characteristics of errors of inertial sensor,this paper studies to adapt dynamic neural units networks to model errors of laser gyro.Framework and dynamic gradient arithmetic are detailedly presented in the paper.The result of simulation example shows that dynamic neural units networks has some advantage for modelling errors of laser gyro output:rapid convergence,good performance of tracking and stabilization.