Prediction of Paddy Straw Mechanical Properties under Varying Moisture Content and Loading Rate using ANN

In this work, the mechanical properties of paddy straw were evaluated using force-deformation curves and neural network approaches. The mechanical strength of the paddy straw was estimated at Moisture Contents (MC) of 10.8, 13.5, and 18.4% (w. b.), on distinguished internode positions, that is, N1, N2, and N3, and with applied Loading Rates (LR), that is, 25, 30, and 35 mm/min, respectively. Results shows that the values of Bending Strength (BS), Shear Strength (SS), and Young's Modulus (YM) were increasing from 5.35 to 17.34 MPa, 4.99 to 7.35 MPa, 0.43 to 1.39 GPa, respectively, through the node position from N3 to N1 with a decrease in MC and an increase in load through internode N1 to N3. Thus the MC and LR at different internode positions significantly (P<0.05) affected the BS, SS, and YM of paddy straw. The Levenberg-Marquardt’s backpropagation based neural network was used to model the mechanical properties of paddy straw. The lower value of RMSE (0.69) and higher value of R2 (0.9998) indicates a best fit for the developed model. The coefficients of determination for BS, SS, and YM were 0.9994, 0.9942, and 0.998, respectively, indicating that the neural network model for estimating the mechanical properties of paddy straw can be used as an excellent alternative under selective experimental conditions.