摘要
为了实现自动导向车(automatic guided vehicle,AGV)的高精度导引控制,设计其双磁传感器测量系统,提出结合自适应卡尔曼滤波定位和滑模变结构控制的路径跟随控制器。利用扩展卡尔曼滤波融合里程计和磁传感器的测量数据,提高了系统的定位精确度,采用单神经元网络在线估计测量系统的噪声协方差,使得该方差接近实际噪声统计特性,增强了扩展卡尔曼滤波的性能和适应性。滑模控制不但保证控制器精确度,还有效地消除了系统未知控制输入扰动的影响,其设计过程基于Lyapunov方法,保证系统的稳定和渐近收敛。仿真结果表明了该控制器的有效性。
In this paper, the measurement system with double magnetic sensors is designed and a path following controller is proposed for high-precision inductive guidance of the automatic guided vehicle （AGV）. This controller with location based on extended kalman filtering （EKF） integrated with odometer readings and magnetic sensor data for accurate location of the AGV, and the noise covariance of mea- surement system was estimated online by using single-layer neural network to be close to the actual noise statistical properties, and therefore the performance and EKF adaptability was enhanced. Moreover, the s- liding mode controller was employed to meet the requirement of control accuracy, and eliminate unknown control input disturbance. Furthermore, the Lyapunov theorem was implemented to ensure the stability and asymptotic convergence of the system. The validity of the controller is verified by simulation results.
出处
《电机与控制学报》
EI
CSCD
北大核心
2013年第9期111-118,共8页
Electric Machines and Control
基金
国家高技术研究发展计划(863计划)资助项目(2012AA040909)
广东省产学研结合专项资金资助项目(2011A090200054)
广东省战略性新兴产业核心技术攻关资助项目(2012A010702004)
关键词
自动导向车
鲁棒导引控制
自适应扩展卡尔曼滤波
定位
单神经元网络
磁传感器
automatic guided vehicle
robust guidance control
self-adapted extended kalman filtering
localization
single-layer neural network
magnetic sensor
