Use of adaptive model-based reasoning for embedded diagnostics and redundancy management for fault tolerant systems

Abstract

Safety, sustainability and mission criticality considerations often predicate the requirement for built-in fault tolerance in aerospace systems. Existing approaches to accomplishing fault tolerance typically focus on "brute-force" hardware redundancy and extensive, complex control logic developed as a "point solution" to effect reconfiguration actions. This paper describes the principal concepts and design implementation of an innovative approach for embedding an adaptive model-based diagnostic reasoning capability into a Fault Tolerant Remote Power Controller (FTRPC) to provide rapid fault diagnostics and reconfiguration of powerflow to critical users. A key aspect of this approach is that a systems engineering process was used to develop the reasoning capability that could be embedded in the system to accomplish fault detection, isolation, reconfiguration and recovery. The system engineering process, applied through an automated tool set, is generic in nature and can be applied to any system, as opposed to a "point solution" developed by intensive engineering efforts. The extensibility and applicability of the overall approach is a key technology accomplishment of the program. This paper describes the underlying concepts and implementation of embedding Diagnostician-on-a-Chip technology into a state-of-the-art remote power controller. This design was recently implemented in an Integrated Product Development environment under a NASA Phase II SBIR Program conducted under the auspices of Marshall Space Flight Center (MSFC). This new approach can revolutionize the implementation of health management for fault tolerant systems by developing a deterministic model-based diagnostic capability that is adaptive to a vast number of dynamic reconfiguration states.