The objective of this research is to design and develop effective methods to prevent and detect conflicts and errors (CEs) through prognostics and diagnostics. Two main challenges in CEs prognostics and diagnostics are to: (1) enable prognostics and diagnostics with effective system modeling; (2) develop and identify optimal prognostics and diagnostics methods for different systems.

An agent-based approach is explored to model systems, and the Centralized and Decentralized Conflict and Error Prevention and Detection (CEPD) Logic is developed for prognostics and diagnostics over three types of real-world constraint networks, random network, scale-free network, and Bose-Einstein condensation network. The CEPD Logic is compared with the Tradition CEPD Algorithm in terms of four performance measures including total CEPD time (TT), CE coverage ability (CA), CE prognostics ability (PA), and total damage (TD). The performance of all three CEPD approaches has been analyzed over networks with large number of nodes.

Analytical and experiment results show that (1) compared to the Centralized Logic and Traditional Algorithm, the Decentralized Logic is preferred and requires the least TT, obtains the largest CA and PA, and minimizes TD; (2) if the Decentralized Logic cannot be applied for certain reasons, e.g., communication disruptions between distributed agents, the Network-Adaptive Centralized Logic should be used to obtain larger CA and PA and smaller TD compared to the Traditional Algorithm.

Prognostics and Diagnostics of Conflicts and Errors with Prevention and Detection Logic

Xin Chen

Ph.D. Candidate

Major Professor:  Shimon Y. Nof

ABSTRACT