The increasing number and complexity of space vehicles and the cost reduction measures that call for fewer and less experienced satellite operators are driving the need for more automation in satellite control. Current methods for detecting and correcting anomalies onboard the spacecraft as well as on the ground are primarily manual and, therefore, slow and error-prone.
Today, it is generally recognized that space vehicle monitoring and fault diagnosis can be automated using advanced operator decision support systems such as rule-based, expert systems. Since, in most satellites, hundreds of health and status indicators are sent to ground stations every second, relieving operators of portions of the telemetry monitoring and anomaly diagnosis task is clearly desirable. Expert systems have proven to be untiring and accurate monitors of telemetry for several military programs including Titan, the Defense Meteorological Satellite Program (DMSP), and the Inertial Upper Stage (IUS). Further, testbed prototyping efforts have shown that it is possible to develop reusable components for certain functions of satellite command and control.
Prototyping efforts have shown that expert systems for satellite anomaly diagnosis can be written with a reusable core of rules that applies to more than one satellite program. These reusable expert systems reduce the amount of new code that must be written for each additional satellite anomaly diagnostic application. For example, a reusable expert system has been created to diagnose problems in the thermal subsystem of two satellite programs. A core of rules that was easily adapted to two applications required only a moderate additional amount of effort than would have been required to do one. In the domain of satellite anomaly diagnosis, and indeed in any domain where numerous similar expert systems are desired, there is a large savings in using this approach. The SSCS could deliver such a core of rules to each satellite program for tailoring to the specific program.
In addition, decision support technology is used in satellite commanding, anomaly diagnosis, failure detection and correction, mission planning, and scheduling. Rule-based systems and case-based reasoning systems can also be used to simulate mission scenarios for training and problem solving. Thus, integrating intelligent-system technology into open-architecture systems helps reduce operator training time, as well as the number of operators needed to support the ground systems.
The development of knowledge/rule bases requires both a knowledge engineer and a domain expert. An involved domain expert who becomes a champion of the project is desirable. Problems that are well defined are most amenable to expert systems for decision support.
Decision support within distributed open-architecture systems should be provided by using COTS products, such as those implemented in rule-based systems and case-based reasoning systems. A variety of decision support products are available today. These products include rule-based systems such as G2 from Gensym and RTWorks from Talerian, case-based systems such as CBR Express from Inference Corporation, and neural network development packages such as NeuralWorks from NeuralWare. These products are being applied in a variety of TT&C applications by Aerospace and various defense contractors.