基于数据驱动的航天器故障检测技术研究

Study on Spacecraft Fault Diagnosis based onthe Data Driven

随着航天器设计的日益复杂,积累了庞大测试数据,故障检测、测试数据应用的有效性问题凸显.本文基于航天器性能测试中的测试数据,利用数据特征提取和识别技术,提出了一种基于数据驱动的故障检测技术.本文提出的技术克服了建立传统专家知识库需要庞大数据做支撑的弊端,同时避免了基于规则的数据判读的局限性,进而解决现阶段航天器故障检测、诊断研究过程中的不足.

Along with the growing complexity of the spacecraft design, huge volume of data has been generated. The issue of effectivelydiagnose the spacecraft fault has been increasingly crucial. Based on the spacecraft test data, this paper proposes a fault diagnosismethod using the feature extraction and pattern recognition technology. This method can overcome the need for huge data supportto build traditional expert level database and avoid the limit of data interpretation based on the scope regulation, and thus toimprove the spacecraft fault diagnosis.