摘要
遥测数据的变化能够反映航天器性能的状态和改变,根据这些改变,能够对航天器关键部件的性能和趋势进行预测。首先利用基于x-11的数据分解方法,将遥测数据分为趋势项、季节项和随机项,之后针对这3种变化类型的数据分别利用BP神经网络、多项式拟合、AR模型、非参数回归等方法进行预测,形成融合后的预测结果,并对算法的应用效果进行了分析,此外还计算了衰减因子。针对某航天器温度参数的实验结果表明,提出的预测方法平均相对误差小于10%,能够有效的用于航天器遥测数据的性能趋势预测,具有较强的航天器工程应用价值。
The change of the telemetry parameter can affect the status and changes of the spacecraft performance. According to these sta- tus and changes, we can predict the performance and tendency of the spacecraft key components. This paper firstly analyzes the spacecraft te- lemetry data change rule. Secondly, the telemetry data is decomposed into 3 types, that is the trendency item, seasonal item and random i- tem, through the method of x--ll. After that, the different prediction methods, such as the AR linear regression method, the BP neural network method, the nonparametric regression method and so on, are taken respectively to predict the long--term performance trend of the above 3 types of the telemetry data. Finally, the predicted data using the above prediction method are fused to form the final prediction re- sult, and the attenuation coefficient between the predicted data and the original data is computed. Furthermore, the experiment result shows the proposed prediction method can be effectively applied to the predicting of the performance trend of the spacecraft telemetry data, and has strong practical significance in the field of the spacecraft Droiect.
出处
《计算机测量与控制》
北大核心
2013年第7期1792-1796,共5页
Computer Measurement &Control
基金
国防基础科研资助项目
