Artificial neural network modeling of a multiphase photodegradation system

• Published in: Journal of photochemistry and photobiology. A, Chemistry. Vol. 149; no. 1; pp. 139 - 146
• Main Authors: Pareek, V.K, Brungs, M.P, Adesina, A.A, Sharma, Raj
• Format: Journal Article
• Language: English
• Published: Lausanne Elsevier B.V 28.06.2002; Elsevier Science
• Subjects: Artificial neural networks (ANN); Backpropagation; Bayer liquor; Chemistry; Exact sciences and technology; General and physical chemistry; Photochemistry; Photodegradation; Physical chemistry of induced reactions (with radiations, particles and ultrasonics); Photodegradation; Backpropagation; Artificial neural networks (ANN); Bayer liquor; Multiphase system; Backpropagation algorithm; Feedforward neural nets; Algorithm performance; Spent liquor; Theoretical study; Photochemical reactor; Chemical degradation; Metallurgy; Viscous solution
• ISSN: 1010-6030; 1873-2666
• DOI: 10.1016/S1010-6030(01)00640-2

Photodegradation of spent Bayer liquor was carried out in an 18 l pilot scale photoreactor. The experimental data indicated that the average reaction rate was a complex nonlinear function of various process variables, such as lamp power, catalyst loading, initial solution pH, liquid batch time, and total organic carbon (TOC) concentration. The experimental data were modeled using feed forward artificial neural networks (ANN). The networks were trained with 350 sets of input–output patterns using backpropagation algorithm. Out of several configurations, a three-layered network with eight-neurons in its hidden layer yielded optimal results with respect to data validation. The optimal model gave excellent predictions with a correlation coefficient of 0.9955.