Quality control of topping plant by neural networks models

Inferential models of product quality are very important for chemical plant operation without an on line sensor. In this case, it is necessary to develop inferential models using measurable process variables, e.g. temperature, pressure, and flow rate. we build an inferential model for estimating the quality of petroleum products, light gas oil 90% recovered temperature, and kerosene 95% recovered temperature by using a backpropagation neural network and apply it to quality control for a topping plant in this study. It is shown that quality control can be efficiently performed by use of a neural networks model.