Abstract: In view of the long lag, nonlinearity, multiple input factor, uncertainty, time-varying and fuzzy characteristics of the dosing process of tap water production, an automatic control system for coagulant dosage of waterworks is developed based on the self-adaption and self-learning of artificial neural network. Zongguan waterworks, the first largest waterworks in Wuhan, is taken as a case study. The influence of Elman neural network on dosage effect is researched, and the preprocessing and data storage and data reading for WinCC industrial control system are accomplished based on OLE-DB open data access standard. The system mainly consists of functional modules including dosing process, data query, curve generation, dosage query, alarm log and alarm statistics, drug consumption statistics, fluctuation assessment, and alarm settings. The system has been applied to Zongguan waterworks successfully. Online monitoring of operation parameters and full automation has been achieved, which provides safeguard for the plant’s safe production. The system also saved dosage consumption, and reduced labor intensity of operators.

Key words: Elman neural network, waterworks, coagulation dosage, WinCC, control system