Performance improvements at surface water treatment works using ANN-based automation schemes

Due to their ability to capture non-linear information very efficiently, ar tificial neural networks [ANNs] have found great popularity amongst the 'co ntrol community' and other disciplines. This paper discusses some recent ap plications of the ANNs at surface water treatment works. The range of appli cation is quite diverse and covers modelling, simulation, condition monitor ing, fault detection and control strategy design and implementation. Attempts to improve the performance of water treatment works through the ap plication of improved control and measurement have had variable success. Th e most quoted reason for this is that the individual dynamic operations def ining the treatment cycle are complex, highly non-linear and poorly underst ood. These problems are compounded by the use of faulty or badly maintained sensors. The efficient and robust operation of any industrial system is critically d ependent on the quality of the measurements made. Also, the structure of th e control policy and choice of the individual controller parameters are imp ortant decisions to the economic operation. Three examples are used to desc ribe how the introduction of ANNs has resulted in more reliable system meas urement and more efficient pH and coagulation control. A final example, sho ws an approach to the use of an ANN to provide 'assistance' to a convention al proportional-integral controller in the form of automatic on-line tuning of the controller parameters.