INTELLIGENT DISCRIMINATION MODEL TO IDENTIFY INFLUENTIAL PARAMETERS DURING CRYSTALLISATION FOULING

The introduction of redundant independent variables into any function approximation model, or the neglect of important variables, may result in a correlation with poor prediction and reduced reliability. This paper demonstrates that a novel integrated model of neural networks and genetic algorithms can deal with this problem robustly with good accuracy, while be far less time-consuming compared to lengthy conventional models Furthermore, a redundant variable input was imposed to the model to discern if the approach could identify it among other important variables. Genetic algorithms were exploited as a powerful optimisation tool for the selection of best set of inputs with the help of process “prior knowledge ” rules. A comprehensive databank of crystallisation fouling under subcooled flow boiling was used. The resulting model was capable of handling the data and successfully discriminated among several independent inputs if there is any redundant input. The technique may be regarded as a robust method to prevent data over-fitting as well as processes where a large number of inputs are involved such as crude oil fouling