Open AccessIntegrating Fuzzy System and Meta-Heuristic Algorithms to Predict Influent Parameters for a Sewage Treatment
Author(s) -
Mozafar Ansari,
Faridah Othman,
El-Shafie
Publication year - 2020
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/498/1/012076
Subject(s) - ammoniacal nitrogen , particle swarm optimization , chemical oxygen demand , biochemical oxygen demand , algorithm , genetic algorithm , sewage treatment , environmental engineering , total suspended solids , fuzzy inference system , heuristic , fuzzy logic , sewage , environmental science , adaptive neuro fuzzy inference system , mathematics , mathematical optimization , computer science , wastewater , fuzzy control system , artificial intelligence
Sewage treatment plants (STPs) are built to reduce the concentration of sewage parameters to a safe level that reduced their impact on the environment. To have an optimal operation of STPs, it is essential to estimate influent parameters precisely. In this research, four influent physicochemical characteristics, i.e. biochemical oxygen demand (BOD), chemical oxygen demand (COD), ammoniacal nitrogen (NH3-N), and suspended solids (SS), of a sewage treatment plant (STP) were analysed and predicted by integrated genetic algorithm Sugeno fuzzy inference system (GA-FIS) and particle swarm optimisation FIS (PSO-FIS). The GA-FIS and PSO-FIS were applied on 10 time-series scenarios, and the results of each scenario were compared to find the best algorithm as well as the scenario for each parameter. Based on the results, both GA-FIS and PSO-FIS algorithms provided very good results, and the differences of error in predicting influent parameters is very less. However, to select the best algorithm for predicting the missing values of each parameter, GA-FIS predicted BOD and SS more accurately than PSO-FIS algorithm, COD and ammoniacal nitrogen had more accurate results when they were predicated by PSO-FIS algorithm.