Optimization of activation process for water purification using one activation basin comprises feeding inflow and outflow rates to trained neural network which provides predicted values for the outflow rate and other sections of the plant

摘要

Optimization of the activation process for water purification using one activation basin comprises feeding the in- and outflow rates for this basin and optionally other data to a trained neural network which provides predicted values for the outflow rate, as in DE19827877. Preferred Features: The process is improved by the network also providing predicted values for other sections of the plant, e.g. post-clarification and filtration stages. Preferably stored data is used for post-training of the network while running the plant.