ANN approach for the prediction of pyrolytic behaviour of pigeon pea stalk and insight into thermo-kinetics of pseudo components using RSM

Abstract

Artificial neural network (ANN) is the widely used technique for prediction and modeling of complex processes. In this study, ANN was applied for predicting the complex pyrolysis behaviour by employing critical process parameters like heating rate (°C/min) and temperature (°C). Two anticipated variables in input layer were used to predict the derivative weight loss (%/°C) by propagating in twelve neurons. The model's reliability and adequacy were validated by the best training performance data recorded at 36 training epochs with an MSE value of 2.34 × 10−5. In order to estimate the thermo-kinetics of the Pigeon pea stalks (PPS), thermogravimetric analysis was done at four heating rates (10–40 °C/min). The TG-spectra were resolved into their pseudo components using deconvolution technique. The activation energy and pre-exponential factor were estimated by Friedman isoconversional method. In continuation, response surface method (RSM) was employed to explore the effect of process parameter on thermodynamic parameters ΔG, ΔH, and ΔS. The recorded statistical values illustrated the adequacy and robustness of the fitted model with anticipated variables. Pyrolytic reaction order was also determined for each pseudo components through master plot method.