Over the past decade there has been a shift in emphasis from design and construction of wastewater treatment facilities to operation. Poor plant performance, high costs and damage to the environment have resulted from operational problems. Wastewater treatment consists of a complex sequence of inter-dependent biological, physical and chemical processes subject to time-varying hydraulic and organic load conditions. Wastewater treatment process operation and control is a knowledge intensive task. Research on improving operation and control has centred on identifying important mechanisms responsible for observed behaviour and modelling both the process and optimum ways of operating the process. These models have served as useful tools for improving operation and control. Many different approaches have been used, including deterministic modelling, stochastic modelling and, more recently, linguistic modelling. Complex mathematical models of wastewater treatment processes consisting of large numbers of non-linear differential equations can be constructed using tools such as the General Purpose Simulator (GPS) and, given appropriate data, model parameters can be evaluated and updated using existing optimization routines. Object oriented programming (OOP) and a model based reasoning (MBR) approach provides a useful framework for development of deep-knowledge expert systems (ES). Data-driven modelling methods, including both time series analysis and artificial neural network (ANN) techniques, can also be employed to make maximum use of information contained in process data. Each of these model types is a necessary component of a computer system for operational control of wastewater treatment but, in isolation, none are sufficient for making the system robust. An integrated environment for combining these techniques has been developed for this purpose and the basis for its development is described.

No abstract

No abstract

No abstract

No abstract