航天飞机主发动机(SSME)故障的早期检测与诊断是很重要的,由此可以有充足的时间防止故障发生和采取纠正措施。因为象SSME这样一个复杂系统中的绝大多数故障发生很快,早期地、及时地检测和诊断这些故障对航天飞机的生存是至关重要的。我们设计了一个专家系统,可以自动学习、检测、识别、证实和纠正异常的推进器系统动作。本文详述了这种专家系统自动训练方面所采用的新颖的机器学习方法。 Early detection and diagnosis of a space shuttle main engine (SSME) failure is important so there is plenty of time to prevent failures and corrective actions. Because the vast majority of failures in a complex system such as SSME happen quickly and early, timely detection and diagnosis of these faults are crucial to the space shuttle’s survival. We designed an expert system that automatically learns, detects, identifies, verifies, and corrects unusual thruster system behavior. This article details the novel machine learning methods used in this type of expert system for automatic training.