Sensor fault detection and isolation using artificial neural networks

In this paper, an approach for the sensor validation in the case of dynamical systems with time-delays is presented. It is based on the combination of feed forward neural networks and information fusion technique. A n architecture is proposed for sensor fault detection, isolation and accommodation using neural generalized observer scheme. Each sensor is dedicated with an observer, driven by the process inputs and the outputs of the process except the output to be supervised. The sensor values are compared with the observer outputs and validated. Multi component distillation column is employed for the investigation and the simulation results show good performance on this complex process.