Fault detection in semi-batch reactor using the EKF and statistical method

In this paper, we present a new fault detection method in semi-batch and batch reactors using model-based technique. This method is based on the use of statistical local approach and discrete Extended Kalman Filter (EKF) with parameter estimation. A statistical test for fault detection is developed from predicted mixture temperature error and the estimation sensitivity of mixture temperature to overall heat transfer coefficient estimation. The efficiency and robustness of the proposed method are demonstrated through simulations and experimental data. For the experimental validation we have used a semi-batch pilot reactor under the adiabatic conditions with a very exothermic reaction (oxidation of sodium thiosulfate by hydrogen peroxide).