Fault detection in waste water treatment plant using statistical analysis & machine learning

Abstract

A suitable model needs to be developed for detecting fault of wastewater treatment plant in order to monitor, predict plant performance and for reducing environment pollutions. Main objective of this study is to introduce time and cost effective data science & machine learning technique to monitor WWTP’s performance and detect plant’s fault instead of manual, laboratory based time consuming, costly, difficult methods. One year of unsupervised data of WWTP collected and convert into supervised data in order to visualized plant’s fault using python. Moreover four model has been created based on water quality standard parameters(Ph, BOD, COD, suspended solid) and we applied different machine learning algorithm’s to take decision by machine itself after identifying normal or faulty data. Machine learning technique in case of finding fault, taking decision gives satisfactory result but different algorithms shows best accuracy for different model. However, machine-learning method will be accurate automatic solution for detecting fault of wastewater treatment plant and reducing environment pollution.