Design and Evaluation of a High Performance Distributed Expert System (HPDES) for Aerospace Ground Verification System

Abstract

Ground verification system is significant in aerospace engineering because it can verify the correctness and robustness of the entire system before spacecrafts are launched. Since expert system can detect errors or exceptions automatically, it is the core component of a verification system. As the data throughput in aerospace engineering is huge, and verification processing is complex, conventional serial expert system cannot satisfy the requirement in real-time processing. Therefore, in this paper, a High performance Distributed Expert System (HPDES) is proposed. Firstly, HPDES proposes a novel language, called CASES, for rule configuration, and then each rule is compiled into an intermediate code. This method allows rules to be added, deleted or modified dynamically. Secondly, in order to improve the performance to meet the requirement of real-time processing, Virtual Executable Environment (VEE) is proposed. VEE allows our expert system to run in parallel or distributed fashion. Thus, real-time processing capability is achieved by assigning different pattern detection tasks to different VEEs dynamically, which utilizes the power of modern parallel and distributed computer systems efficiently. Experiments in a real ground verification system shows that HPDES can detect errors and exceptions in the data stream from certain spacecraft efficiently.