A Special-Purpose Processor for Probabilistic Model-Based Diagnosis

One of the fundamental tasks in technical processes is online observation and diagnosis. Only if the supervising computer system constantly monitors the operating characteristics of the technical process, detects and localizes faults, can correctional actions be taken on time. Model-based diagnosis involves the internal structure and functionality of the technical process. Additionally probabilistic diagnosis deals with failure probabilities of components and facilitates sophisticated fault diagnosis techniques. The design of this new special purpose processor is based on a probabilistic model-based diagnosis algorithm. Our interest focuses on the architectural design (software/hardware migration) of a heterogeneous diagnosis agent, which will be implemented by multi-DSPs (digital signal processors) and FPGAs (field programmable gate arrays).