This paper investigates the use of backpropagation neural network (BPNN) to predict humic substance (HS) UV absorbance experimental results. The studied experimental sets include HS and heavy metal agglomeration, HS coagulation using polyelectrolytes and HS and heavy metal coagulation using polyelectrolytes. BPNN simulation showed high prediction accuracy where regression coefficient (R) was > 0.95 for all simulations. Lower and higher than optimum training data input reduces BPNN reliability due to under training or over-fitting. The number of neurons study showed that a lower number of neurons led to under training, while a higher number of neurons resulted in the network memorizing the input dataset.
|الصفحات (من إلى)||153-157|
|المعرِّفات الرقمية للأشياء|
|حالة النشر||Published - أبريل 2010|
ASJC Scopus subject areas