Indian Journal of Science and Technology
DOI: 10.17485/ijst/2011/v4i5.17
Year: 2011, Volume: 4, Issue: 5, Pages: 542-546
Original Article
Shilpa Tripathi1* and J. K. Shrivastava2
1 Chemical Engineering Department, Institute of Engineering & Science, IPS Academy, Indore (M.P.), India
2 Chemical Engineering Department, Ujjain Engineering college, Ujjain (M.P.), India
[email protected]
This paper focuses on the effect of process constraints on success of the artificial neural network (ANN) program in predicting the compost process. Agricultural waste (sugarcane bagasse, soya husk, and wood straw) was composted aerobically and data on experimentally determined carbon dioxide evolved was used for simulation of composting process. A computer program was developed using ANN technique for simulation purpose. The ANN based model was tested for fitting a combination of exponential functions on experimental data, for two different rates of reaction, corresponding to easily and moderately hydrolysable waste. Initial trials failed to give satisfactory fit, but introduction of certain constraints like relative magnitude of the two reaction rates and constraint on total carbon helped in convergence of the solution. This study reveals the important role of constraints on process parameters in modeling of aerobic composting of agricultural waste. ANN program required initial estimates of process parameters to start the convergence process. Effect of these estimates is also discussed here.
Keywords: Agricultural waste, compost, artificial neural network, carbon.
Subscribe now for latest articles and news.