Using neural networks for fault detection in a distillation column

Several methods of fault detection have been put to testing with the purpose of securing the installations and reducing the risks of accidents. This paper presents a new approach of fault detection based on the realisation of a Bayesian neural separate at radial basis functions. In this paper, our contribution consists of demonstrating the way this kind of network can be used as faults separate, applied to a continuous distillation column containing a binary mixture of toluene/methylcyclohexane. The latter is carried out through the use of test base containing two operating modes: normal and abnormal.