Abstract: During state monitoring for in-orbit spacecraft using multi-sensor data, uncertainty in multi-source information can cause inconsistent decisions among channels. The common engineering technical method to this problem is further analysed according to artificial expert expertise, which can add subjectivity to the decision result and cause low efficiency of monitoring. To realize automatic monitoring, a fusion method based on evidence theory is proposed for the monitoring problem. Six kinds of confidence function for anomaly state monitoring are designed. By comparison, the radical function is suitable. Improvement on subsection is made according to the data characteristics. Considering the engineering
requirements, principles setting adjustable parameters are introduced. This method is applied to deal with the downloaded multi-sensor data from in-orbit spacecraft. Monitoring result is consistent with the artificial conclusion. The effectiveness of this fusion monitoring method is verified.

Key words: Evidence theory, Confidence function, Information fusion, In-orbit spacecraft, Anomaly state monitoring