Design of online water quality monitoring system and prediction based on probabilistic neural network

Abstract

Online water quality monitoring systems have many significant benefits in fields such as source water protection, water treatment, and wastewater discharge, and in accidental or deliberate event of water contamination. It is also used to provide proofs and early warning for companies that discharge industrial wastewater. A typical online water quality monitoring system can collect data about specified factors automatically and continuously, analyse data, present data, store data and transfer data. This paper introduces the authors' application in BASF Shanghai Coatings Co., Ltd, a scalable online water quality monitoring system that focuses on rain drainage. The hardware platform of this system is based on an ARM processor and embedded Linux operating system; the software platform of this system is SOA Service-Oriented Architecture. Owing to the generic designed hardware and software platform structure, this system is also competent for other environment protection monitoring applications if the user modifies the software settings as he needs. Meanwhile, this paper suggests a method using a probabilistic neural network to predict COD Chemical Oxygen Demand value. pH Hydrogen Ion Concentration, temperature, water drainage traffic flow, humidity and light intensity were used as five input variables, while COD was used as the output variable.