A comparative study between mathematical models and the ANN data mining technique in draft force prediction of disk plow implement in clay loam soil

Seyed Mojtaba Shafaei, Mohammad Loghavi, Saadat Kamgar

Abstract


This paper communicates the prediction of required draft force of disk plow implement during tillage operations. The well-known mathematical model proposed by American Society of Agricultural and Biological Engineers (ASABE), multiple linear regression (MLR) and data mining model, based on artificial neural network (ANN), were employed for this purpose. The input variables of the models were considered as forward speed of 2-6 (km/h) and plowing depth of 10-30 (cm). The development details of the models are documented in the paper. On account of statistical performance criteria, the best ANN model with coefficient of determination of 0.971, root mean square error of 0.762 (kN), mean absolute percentage error of 1.886 (%) and mean value of absolute prediction residual errors of 0.968 (kN) was better performed than ASABE and MLR models for prediction of required draft force. The ANN modeling results also showed that the simultaneous or individual increment of forward speed and plowing depth caused nonlinear increment of draft force. The well-developed ANN model is considered operational to predict draft force as an essential step toward proper selection of combination of tractor and disk plow implement.


Full Text:

PDF