Nonparametric kernel smoothing and FIR filtering for model-free fault symptom generation

This paper describes some new theoretical and practical developments to a recently proposed approach to the generation of fault symptoms in nonlinear systems. The method is model-free, in the sense that no analytical model of the plant is needed. The kernel-smoother makes it possible to detect changes in the plant dynamics, possibly due to some malfunction. After recalling several basic definitions and results, a novel interpretation in the context of FIR filters is presented, and some practical implementation issues are addressed. Experimental results on the three-tank benchmark are finally reported, showing the effectiveness of the proposed methodology.