Process monitoring methods have been studied widely in recent years, and several industrial applications have been published. Early detection and identification of abnormal and undesired process states and equipment failures are essential requirements for safe and reliable processes. This helps to reduce the amount of production losses during abnormal events. In this paper, statistical multivariate methods and neural networks applied in monitoring of an industrial dearomatisation process are compared. No appriori process knowledge for the methods were assumed. The data for the comparison were generated with a dynamic simulator model of the process. Special emphasis was put on a case of internal leak in a heat exchanger.

No abstract

No abstract

We propose a grid-like computational model of tubular reactors. The architecture is inspired by the computations performed by solvers of partial differential equations which describe the dynamics of the chemical process inside a tubular reactor. The proposed model may be entirely based on the known form of the partial differential equations or it may contain generic machine learning components such as multi-layer perceptrons. We show that the proposed model can be trained using limited amounts of data to describe the state of a fixed-bed catalytic reactor. The trained model can reconstruct unmeasured states such as the catalyst activity using the measurements of inlet concentrations and temperatures along the reactor.