基于TSEOPM的在轨航天器故障预报方法研究

针对在轨航天器非线性系统的故障预报,提出一种基于时间序列事件征兆模式挖掘的在轨航天器故障预报方法,该方法以在轨航天器遥测数据建立状态监测时间序列,根据事件特征函数利用层次聚类算法挖掘出故障事件征兆模式,然后利用故障事件征兆模式对航天器的状态监测时间序列数据进行分析,判断是否为故障征兆点,从而实现故障预报;实验结果表明,该方法能有效地预测在轨航天器状态监测时间序列数据中的故障事件。

Research on Method of Fault Prediction for Onboard Spacecrafts Based on Time Series Event Omen pattern Mining

A new fault prediction method for onboard spacecrafts based on time series event omen pattern mining is proposed. The fault omen patterns are mined based on character function by application of the hierarchical clustering algorithm in the condition monitoring time series that are formed by the telemetry data of onboard spacecrafts. Then, the fault omen patterns are used to analyze the condition monitoring time series data to find the fault omen dots. If the dots are detected, the fault event is predicted. Experimental results indicate the proposed method is effective in fault event prediction of onboard spacecrafts.