A framework for unsupervised fault detection and diagnosis based on clustering assisted kriging observer

This paper presents an unsupervised data-driven method for Fault Detection and Diagnosis (FDD) of nonlinear dynamic processes. The proposed approach is based on the combination of automatic and non-automatic clustering techniques with a data-driven observer based on Multivariate Dynamic Kriging (MDK) metamodels. The proposed framework is studied via its application to a well-known benchmark simulation case study based on the control of a three-tanks system, showing promising performance in terms of accuracy, robustness and simplicity of applications.