Predictive models using supervised neural network for pollutant removal efficiency in petrochemical wastewater treatment

• Main Authors: Mohan, Varun Geetha, Al-Fahim, Mubarak-Ali, Ameedeen, Mohamed Ariff, Vijayan, Bincy Lathakumary, Aminuddin, Afrig, Widayani, Wiwi
• Format: Conference or Workshop Item
• Language: English; English
• Published: IEEE 2022
• Subjects: QA76 Computer software; QD Chemistry; TA Engineering (General). Civil engineering (General); TP Chemical technology
• Online Access: http://umpir.ump.edu.my/id/eprint/35990/2/Predictive%20Models%20Using%20Supervised%20Neural_FULL.pdf; http://umpir.ump.edu.my/id/eprint/35990/3/Predictive%20models%20using%20supervised%20neural%20network%20for%20pollutant%20removal%20.pdf; http://umpir.ump.edu.my/id/eprint/35990/; https://doi.org/ 10.1109/ICOIACT55506.2022.9971929

The important process in wastewater treatment is the removal of pollutants, and the dataset having so many features may cause difficulty training the data and predicting key variables. This work aims to propose set parameters through normalization techniques, feature selection techniques, and AI techniques. The datasets have 36 features and a key parameter, and experimental datasets contain 628. Constant factor, Z-score, and Min-max normalization are the normalization techniques used to normalize the petrochemical wastewater dataset. SelectKBest, ExtraTreeClassifier, PCA, and RFE are the feature selection techniques for data mining. Then finally done with AI implementation with the help of a supervised neural network technique called backpropagation neural network (BPNN).