Neural Network Based Modeling for Polycaprolactone Synthesis by Bio-Polymerization of ε-caprolactone

Extensive study of ring-opening polymerizationε-caprolactone (ε-CL) using lipase Novozym 435 (immobilizedform of lipase B from Candida antarctica) as biocatalyst usingring-opening polymerization (ROP) of ε-caprolactone wascarried out at impeller speeds of 250, 500, 750, 1000 rpm andtemperature of the reactor of 60°C, 70°C and 80°C. Themaximum molecular weight out of all the experiments carriedout is 310000 Kilo Daltons (weight average molecular weight,Mw) which was obtained at a temperature of 70°C and 3 hoursfor an impeller speed of 500 rpm. In order to develop apredictive model a multilayer feed-forward neural network(FANN) trained with an error back-propagation algorithm wasincorporated. The results showed that a 3-7-1 for FANN1 withthe inclusion of the Reactor impeller speed and 2-6-1 FANN2arrangement with exclusion of the Reactor impeller gave thebest performance.