航天器小样本遥测数据智能判读规则框架构建与应用OA

Spacecraft telemetry data is the core information for characterizing operational status and enabling health management and fault diagnosis of spacecraft.These data are typified by small samples:valid samples are limited in total and scarce fault anomaly samples are scarce,which imposes substantial challenges to telemetry data interpretation.Conventional interpreta-tion methods used in current engineering applications rely heavily on fixed criteria formulated from expert experience,limited by simplistic criterion structures and insufficient generalization in anomaly detection under small-sample scenarios,making it difficult to meet the requirements of intelligent spacecraft health management.To resolve these issues,this paper proposes an intelli-gent interpretation rule framework for small-sample spacecraft telemetry data,and develops a complete workflow consisting of system analysis,data preparation,rule generation,and intelli-gent interpretation.Positive sample augmentation and feature representation are implemented u-sing convolutional expansion and autoencoders.For abrupt faults and gradual faults,interpreta-tion rules are generated by change-point detection and dynamic threshold algorithms,respective-ly.Validation on spacecraft engine pressure signals shows that the proposed framework accurate-ly identifies anomalies in small-sample telemetry data and improves the intelligence of interpreta-tion,offering key support for engineering application of related technologies.