Leakage Detection Via Model Based Method

This paper presents a model based approach for real time leakage detection in a nonlinear multi tank flow rig. A state space model of the process was derived based on the expansion of the nonlinear function into a Taylor's series about the operating point and the retention of only the linear term. The model based approach performs consistency checks of the process against the state space model of the process. Using fault detection technique, leakage indicating signals were generated. False detection due to uncertainty as a result of using the linearised reference mathematical model and sensor noise can be avoided by referring to real time measurement data recorded when system is operating normally. This method has been implemented in real time and results show promising potentials.