Chemical System Dynamic Identification with Application to Sensor Fault Detection

This work proposes a method for input-output sensor fault detection and isolation of an industrial processes using fuzzy process models. The presented technique concerns the identification of a piecewise affine fuzzy system based on Takagi-Sugeno models. The process under investigation may, in fact, be represented as a composition of several Takagi-Sugeno models selected according to the process operating conditions. This work also addresses a method for the identification of the local Takagi-Sugeno models from a sequence of noisy measurements acquired from the real process. The fault detection scheme adopted to generate residuals uses the Takagi-Sugeno fuzzy model. The developed technique was applied to fault diagnosis of input-output sensors of a sugar cane crushing mill.