摘要
对船摇数据建立了一个时间序列自回归实时处理模型，在该模型的基础上，对船摇数据进行实时滤波及预报。经过大量实测数据及仿真数据的计算，从均方根差、均方根偏差、残差相关性、预报误差等多个方面研究了该方法的处理效果。计算结果表明该方法具有计算速度快、滤波及预报精度高等优点；将该方法与指数递推滤波＋极大似然预报法进行比较，前者明显优于后者。
In this paper,a time -series autoregressive real-time processing model for ship posture data is built and formulae for both real-time-filtering and real-time prediction are given. The processing effect of themethod form many aspects such as mean-square root error, mean - square root deviation, residuals correlation and predictive error is studied through a large amount of actual testing data and simulation data calculation. The calculation result indicates that this method is with such advantages: quick calculation speed, high accuracy for filtering and prediction. By comparing this method with index - recurrence filtering + maximum like lihood prediction method, the former is obviously better than the latter one.
出处
《航天控制》
CSCD
北大核心
1996年第4期35-39,共5页
Aerospace Control
