Phytoremediation of wastewater using internet of things (IoT) and machine learning techniques

• Main Author: Hauwa Mohammed Mustafa, Dr.
• Format: text::Thesis
• Language: English
• Published: 2023
• Subjects: Phytoremediation of wastewater

In this research, three different scenarios were employed. Firstly, Pistia stratiotes, Salvinia molesta and Eichhornia crassipes plants were cultivated for the phytoremediation of secondary treated domestic wastewater. Physicochemical tests were conducted on the influent and effluent water samples. In the second scenario, water quality (WQ) indicators (total dissolved solids (TDS), temperature, oxidationreduction potential (ORP) and turbidity) of the S. molesta treatment system at a retention time of 24 hrs was measured using internet of things (IoT) device. The results obtained showed that the optimum conditions for the treatment of the influent water samples was observed at 24 hrs retention, with up to 6.56-7.02 (P. stratiotes), 6.57-6.96 (S. molesta) and 6.43-6.89 (E. crassipes) for pH, 76.7% (P. stratiotes), 91.4% (S. molesta) and 74% (E. crassipes) for colour, 91% (P. stratiotes), 94% (S. molesta) and 89.3% (E. crassipes) reduction for turbidity, 70.34% (P. stratiotes), 81.02% (S. molesta) and 67.2% (E. crassipes) for chemical oxygen demand (COD), 53.2% (P. stratiotes), 74.7% (S. molesta) and 58% (E. crassipes) for biochemical oxygen demand (BOD5), 81.2% (P. stratiotes), 82.7% (S. molesta) and 88.5% (E. crassipes) for phosphate reduction, 88.7% (P. stratiotes), 90.5% (S. molesta) and 89.1% (E. crassipes) for ammoniacal nitrogen reduction and 83.6% (P. stratiotes), 92.1% (S. molesta) and 93% (E. crassipes) for nitrate reduction. In addition, the results obtained from the modelling and prediction of the water quality parameters depicted that the machine learning (ML) models proved merit with high precision in both the training and testing phase compared to the linear model method. The nonlinear ensemble model has improved the accuracy of the ML models with significant precision. Hence, emerged promising in modelling the treated effluent water. Further studies should focus on employing circular economy (CE) concept in phytoremediation of domestic wastewater, conversion of the harvested plant biomass into bioenergy and application of other single computational techniques such as emotional neural network, extreme gradient boosting and Elman neural network.