摘要
针对单探测器复合轴粗跟瞄预测过程中数据利用率不高的问题,提出了基于卡尔曼滤波和最小二乘法的实时预测方法,并通过实测振动数据进行仿真验证。结果显示:随着步长的增加,数据点之间的关联性降低,导致预测准确性和稳定性下降;数据利用率的提高又会增加数据点关联性,提升预测准确性和稳定性。在预测步长为25 ms即数据利用率为75%的情况下,达到预测误差最小和稳定性最优的平衡点。
To address the problem of low data utilization efficiency in the coarse-aiming prediction of a single-detector composite axis,a real-time prediction method based on Kalman filtering and the least-squares method is proposed and validated via simulations using actual vibration data.The results show that as the step size increases,the correlation between data points decreases,thus reducing the prediction accuracy and stability.However,increasing data utilization enhances the correlation between data points,thereby improving the prediction accuracy and stability.An optimal balance between minimal prediction error and optimal stability is achieved at a prediction step length of 25 ms,which corresponds to a data utilization rate of 75%.
作者
陈加文
孙凝
刘建国
CHEN Jiawen;SUN Ning;LIU Jianguo(Laboratory of Nano-Optoelectronics,Institute of Semiconductors,Chinese Academy of Sciences,Beijing 100083,CHN;College of Materials Science and Opto-Electronic Technol,University of Chinese Academy of Sciences,Beijing 100049,CHN)
出处
《半导体光电》
CAS
北大核心
2024年第4期658-661,共4页
Semiconductor Optoelectronics
基金
国家自然科学基金项目(12374397)。
