基于PSO-BP神经网络的激光陀螺温度补偿方法  

Temperature compensation method of laser gyroscope based on PSO-BP neural network

作　　者：张文 王庭军 王雷 陶陶 ZHANG Wen;WANG Tingjun;WANG Lei;TAO Tao(Xi'an Aerospace Precision Mechatronics Institute,Xi’an 710000,China)

机构地区：[1]西安航天精密机电研究所,西安710000

出　　处：《中国惯性技术学报》2022年第5期652-657,共6页Journal of Chinese Inertial Technology

基　　金：装备重大基础研究项目(514010207-307-2)。

摘　　要：激光捷联惯导系统上电启动时,陀螺受温度影响其零偏会经历快速变化到逐渐稳定的过程,影响惯导系统应用精度。因此,提出了一种基于粒子群-反向传播神经网络(PSO-BP)的激光陀螺温度补偿方法,利用粒子群算法寻找神经网络模型的最优权值与阈值,以温度和温度梯度作为自变量,建立陀螺零偏输出的补偿模型。激光惯导系统工作温度范围内的温度试验结果表明:与传统反向传播神经网络算法相比,所提出的PSO-BP神经网络模型的速度提高了4倍,模型拟合精度更高,且避免了反向传播算法易陷入局部最优解的问题。经过粒子群-反向传播算法补偿后,陀螺零偏稳定性相比温补前提高了60%,进一步验证了模型的有效性。After the laser strapdown inertial navigation system(SINS)is powered on and started,the bias of the laser gyro will undergo a process of dynamic change to stability with the temperature change,which will affect the application accuracy of the SINS.Therefore,a temperature compensation method for laser gyro based on particle swarm optimization-back propagation(PSO-BP)neural network is proposed.PSO is used to find the optimal weights and thresholds of the neural network model.Temperature and temperature gradient are used as independent variables to establish a network of gyro zero-bias compensation model.Temperature test results within the working temperature range of the laser SINS show that compared with the traditional BP neural network algorithm,the training speed of the proposed PSO-BP neural network model is improved by 4 times.The model fitting accuracy is higher.The problem that the BP algorithm is easy to fall into the local optimal solution is avoided.After compensation by PSO-BP algorithm,the gyro bias stability is improved by 60%,which further verifies the validity of the model.

关 键 词：激光陀螺仪 粒子群算法 神经网络算法 零偏温度误差模型 零偏稳定性 

分 类 号：U666.1[交通运输工程—船舶及航道工程]