APPLICATION OF SLIME MOULD ALGORITHM SMA AND AAN MODEL TO PREDICT WATER QUALITY INDEX: THE YEŞİLIRMAK RİVER, TURKEY

INTERNATIONAL MEDITERRANEAN CONGRESS, Mersin, Türkiye, 16 Kasım 2022, ss.584-591, (Tam Metin Bildiri)

Prediction of the water quality index is crucial for planning a sustainable ecosystem and executing a good

management strategy to prevent environmental disasters. This study suggests a novel haybird model that

forecasts monthly water quality index based on historical data, which combines preprocessing data

techniques and artificial neural network (ANN) based on a metaheuristic Algorithm (Slime Mould Algorithm

SMA). The suggested methodology was developed and validated using eleven water quality parameters

data(BOD5, PH, DO, Ca+2, CL-1, EC, Mg+2, Na+1, SO4, TDS, NO2

−1 ) for the YeĢilırmak River in Çorum City,

turkey, spanning twenty years, from (1995 to 2014), along with the influence of climatic factors (rainfall).

Moreover, the assigned weight approach was adopted to calculate the water quality index (WQI). The raw

dataset was divided into three groups: training(70%), testing(15%), and validation(15%). The results show

that the preprocessing technique (the natural logarithm Ln, singular spectrum analysis SSA and tolerance)

enhances the quality of the original data and contributes by selecting the optimal prediction model's input.

Also, the haybird model (SMA-ANN) yields a good result in prediction water quality index depending on

different statistical criteria (determination coefficient (R²), mean absolute error (MAE), and root means

squared error (RMSE)) and was achieving to MAE=0.0186, RME=0.021 and R2= 0.921.

Keywords: Forecasting, Water Quality index WQI, Slime Mould Algorithm SMA, YeĢilırmak River.