A neural networks based approach for fault detection and diagnosis: application to a real process

This paper proposes a new fault detection and diagnosis (FDD) method based on the online parameter estimation using the frequency contents of the signals and backpropagation neural networks. When a fault occurs the parameters in a nonlinear mathematical model of the process change. A method for detecting and tracking the different values of the parameters is proposed, which tries to be robust with respect to low frequency disturbances. The new FDD method together with a classical fault detection method are applied to a wastewater treatment plant, placed in Manresa, Spain. A set of real experiments are presented in order to compare and validate the methods in industrial applications.