Abstract:
Redundancy management of a technical system involves a monitoring procedure (control of the current state of its components) to reconfigure the system as needed. Consisting of four parts, this survey presents modern and newly developed technical condition monitoring methods for redundancy management. Part I was devoted to a general description of built-in control tools, voting schemes, and fidelity rules; control codes and program execution control methods were briefly covered. Part II of the survey considers diagnosis methods based on the classical modeling of the system diagnosed in the discrete time and frequency domains. The Chow–Willsky fault detection scheme, as well as the definition of a residual and its generation procedures in the diagnosis problem, are presented. The main model-based diagnosis methods using equation errors, observers, parity equations, and redundant variables are described. In conclusion, the robustness problem of diagnosis methods of the corresponding type is briefly discussed.
Keywords:technical condition monitoring, redundancy management, diagnosis, residual, analytical models of systems, equation error method, observers, Beard filter, fault detection via parity equations, redundant variable method, robustness of methods.
Fulltext:
PDF file (1460 kB)