A Neural Network Model to Estimate the Clogging of Ash Mixed Sand Filter

Infiltration filters of LID facilities are frequently confronted with clogging pores by contaminants. To avoid malfunctioning facilities, should be maintained according to the degree clogging. In this study, was characterized curves representing correlation between infiltration rate (<i>Q</i>) and quantity (<i>IR</i>). estimate clogging, a regression neural network model developed label (infiltration quantity) corresponding selected features (ash mix ratio, rate, variation rate). According results obtained from evaluation accuracy applicability final trained over whole training dataset, following conclusions were drawn: 1) verified correct mean absolute errors 0.00445 0.00491 for test datasets, respectively; 2) could applied multiple, multivariate arbitrary ash mixing ratios range time series in terms rate; 3) given that not curve future steps which trained, CNN or LSTM-RNN suggested as an alternative.