Power and energy requirements for size reduction of wheat are affected by t
he physical and mechanical characteristics of wheat, and the operational pa
rameters of the roller mill. Wheat milling tests were conducted using 204 s
amples of six classes and various varieties collected from around the Unite
d States. Backpropagation neural network models were designed, trained and
validated for the prediction of power and energy requirements of wheat mill
ing using a roller mill: fast roll power (P-f), slow roll power (P-s), net
power (P-n), energy per unit mass (E-m), and specific energy (E-a). Nine va
riables including physical properties of wheat samples and operational para
meters of the roller mill were used as inputs of the networks. Each of the
networks had only one layer of hidden neurons. Sensibility studies also wer
e carried out to investigate effects of input variables on the output varia
bles. The developed network models performed well during validation. Compar
ed to the experimental data, the values of the root mean square error (RMS)
and relative error (RE) of the predicted values were small, ranging from 0
.34 to 0.73 for the RE values. The r(2) values were higher than 0.98 for al
l five networks, The prediction accuracies of the neural network models wer
e significantly improved compared to statistical models.