Neural-network-based water quality monitoring for wastewater treatment processes

Wastewater treatment processes are typical complex dynamic processes. They are characterized by severe random disturbance, strong nonlinearity, time-variant properties and uncertainty. Sensors which have higher reliability and adaptability are needed in such systems. Sensors currently available for wastewater treatment processes are limited both in sorts and reliability, and most of the sensors are expensive. Software sensors can give estimation to unmeasured state variables according to the measured information provided by online measuring instruments available in the system. Soft measurement technology based on artificial neural network has an obvious advantage in solving high nonlinearity and uncertainty. BP neural network was used to construct a soft measurement approach to monitor the effluent COD and BOD in wastewater treatment processes in this paper. Simulation results show that the soft measurement approach based on neural network can estimate the state variables accurately and can be used in real-time monitoring of biochemical wastewater treatment processes.