Success in the development and application of a model requires, as a rule, high-quality field data. In general, studies in controlling the dynamics of wastewater treatment processes have been poorly served in their access to such data. The Environmental Process Control Laboratory has been developed in order to rectify this limitation. The Laboratory is a mobile facility housing instrumentation for on-line respirometry and sensors for real-time monitoring of sludge blanket level and the concentrations of dissolved oxygen, suspended solids, ammonium-N, nitrite-N, total oxidised nitrogen, total organic carbon and orthophosphate-P concentrations. The Laboratory has been designed for deployment in a variety of contexts, but principally in the study of municipal and industrial wastewater treatment, protection of surface water quality, aquaculture, and groundwater contamination. Its purpose is to support the development of process models and, where appropriate, procedures of decision support and automatic control for these systems. Preliminary results from commissioning trials with the Laboratory at the Athens, Georgia, Water Pollution Control Facility Number 2 are reported. These expose some critical issues of signal pre-processing and the need to re-think a strategy for developing models in order to interpret the very large volumes of data generated by the Laboratory.
Autocalibration of Environmental Process Models Using a PAC Learning Hypothesis
