Diagnosis of interconnected systems via well tuned model-based coupling algorithms

This paper presents a heuristic tuning method for a recently proposed model-based coupling algorithm. The purpose of the coupling algorithm is to mitigate effects caused by non ideal coupling of different subsystems where coupling imperfections like time delays, data losses or noisy measurements degrade the overall system performance. If the coupling algorithm is well tuned, a consistency criterion can be used for online closed-loop diagnosis. The simplicity and effectiveness of the proposed tuning and diagnosis approach is demonstrated via a laboratory experiment.