压电振动陀螺温漂性能粒子群算法优化  

POS algorithm optimizing for temperature drift performance of piezoelectric vibratory gyroscope

作　　者：刘宇[1] 陈利娟[1] 王仕亮[1] 何晓艳[1] 刘聪[1] 

机构地区：[1]重庆邮电大学光纤通信重点实验室,重庆400065

出　　处：《重庆邮电大学学报（自然科学版）》2013年第4期527-531,共5页Journal of Chongqing University of Posts and Telecommunications(Natural Science Edition)

基　　金：国家自然科学基金(51175535);重庆市科委自然科学基金(CSTC2012jjB40003);重庆市科技项目基金(KJ110507)~~

摘　　要：针对压电振动陀螺的温漂问题,采用基于粒子群的BP神经网络算法对压电振动陀螺的温度漂移现象进行建模。该算法借助粒子群算法帮助BP神经网络越过局部最小解,并通过加入高斯噪声的方式,模拟自然人脑。仿真实验表明,相对于传统的单BP神经网络算法,含有噪声的粒子群-BP神经网络算法,在精度方面提高了至少42.6%,所构建的温漂模型具有更好的非线性描述能力,从而能为压电振动陀螺提供了更高精度的零电位误差补偿;同时,在收敛速度方面快了5.2倍。The paper uses particle swarm optimization of BP neural network to establish mathematic model of zero position of piezoelectric vibratory gyro in the algorithm.Particle swarm optimization help BP neural network to get across the local minimum solution and by adding Gaussion noise to simulate condition of natural human.The simulation experiment indicates that comparing to the traditional BP neural network algorithm,the novel particle swarm optimization of BP neural network which has noise in it has 42.3% improvement in precision at least.The temperature-drift model has better nonlinear description ability,therefore it can provide zero potential error compensation of higher precision for piezoelectric vibratory gyro.In addition,in the aspect of convergence speed,the novel improves nearly 5.2 times.

关 键 词：压电陀螺 算法优化 粒子群算法 高斯噪声 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] V241.5[自动化与计算机技术—控制科学与工程]