基于RBF神经网络的天线姿态测量系统设计  

Design for antenna attitude measurement based on RBF neural network

作　　者：赵琳[1] 刘付强[1] 吴鹏 王文晶[1] 

机构地区：[1]哈尔滨工程大学自动化学院,哈尔滨150001

出　　处：《中国惯性技术学报》2008年第2期132-135,共4页Journal of Chinese Inertial Technology

基　　金：国家自然科学基金(60474046)

摘　　要：基于MEMS陀螺仪的测姿系统体积小、成本低,但较低的陀螺精度无法保证系统长期工作。采用Kalman滤波技术将陀螺仪和加速度计、磁强计信息相融合,可保证系统的长期精度。船舶机动行驶引入的干扰加速度会造成Kalman滤波精度降低,为了克服这一不足,利用神经网络的自学习能力,采用径向基神经网络设计船用天线的微型捷联测姿系统,以提高船舶机动航行时的系统精度。阐述了组合式捷联姿态系统算法原理,并通过仿真试验结果证明:训练后的网络能够克服干扰加速度的影响,保证系统稳定工作。The attitude measurement unit based on MEMS gyros use Kalman filter to integrate the information of gyros, accelerometers and magnetometers to improve its long-term precision, but the filter precision may be weakened by the disturbance acceleration generated by ship maneuver. In order to overcome this shortage, the self-study ability of radial basis function（RBF） neural network was used to design a strapdown attitude measurement system for ship antenna. The principle of the integrate strapdown attitude measurement arithmetic was expounded, and the simulation experiment was also made to test the efficiency of the proposed method. The results show that the trained neural network can effectively overcome the influence of disturbance acceleration, and ensure the system work stably.

关 键 词：MEMS陀螺仪 姿态测量系统 干扰加速度 径向基神经网络 

分 类 号：U666.1[交通运输工程—船舶及航道工程]