Gyroscopes are instruments capable of sensing changes in their inertial orientation. A gyroscope has one or more sensitive axis and can measure a rotation around them. They are usually mechanical devices that employ a rapid spinning mass to detect the inertial variations. Satellites usually have gyroscopes in the Attitude and Orbit Control System as attitude (orientation) sensors that measure angular displacements. They are critical components, which are constantly monitored during the duration of the mission. In face of serious failures, proper corrective actions are possible, but they depend on the time available for the spacecraft controllers to react. Currently used gyroscope monitoring methods (low-level hardware checks and validation of operational parameters) are adequate for detecting failures that can happen in the short-term but fail to predict long-term failures. To detect these long-term degradation trends, spacecraft controllers perform detailed studies of the power spectral density properties of the gyroscope output. In a real situation, the decision made by the control team to declare a gyroscope faulty depends on qualitative reasoning based on experience. This paper describes a fuzzy expert system for gyroscope fault detection that formalizes and captures the knowledge and experience gathered in previous missions. This expert system provides the spacecraft controller's team, responsible for satellite operations, with a new type of gyroscope health diagnostic tool. This diagnostic tool generates alarms with different degrees of criticality and a severity level of the alarm itself, instead of a simple presence/absence of the alarm. It also supplies an explanation of the alarm. The model presented here is now embedded in a decision-support system, the ENVISAT Gyroscope Monitor, specifically designed for monitoring and diagnosing ENVISAT satellite gyroscopes. The ENVISAT Gyroscope Monitor was jointly developed by GTD(+) and UNINOVA* for the European Space Agency(+) project "Fuzzy Logic for Mission Control Processes".
