Advanced hybrid fuzzy-neural controller for industrial wastewater treatment

Many uncertain factors affect the operation of wastewater treatment plants. These include the physical and chemical properties of wastewater streams a s well as the degradation mechanisms exhibited by biological processes. Bec ause of the rising concerns about environmental and economic impacts, impro ved process control algorithms, using artificial intelligence technologies, have received wide attention. Recent advances in control engineering sugge st that hybrid control strategies, integrating some ideas and paradigms exi sting in different soft computing techniques, such as fuzzy logic, genetic algorithms, and neural networks, may provide improved control of effluent q uality. The methodology proposed in this study employs a three-stage analys is that integrates three soft computing approaches for generating a represe ntative state function, searching a set of multiobjective control strategie s, and autotuning the fuzzy control rule base used for controlling a treatm ent plant. The case study, using an industrial wastewater treatment plant i n Taiwan as an example, demonstrates the applicability of the approach. The findings from this research suggest that a genetic-algorithm-based hybrid fuzzy-neural controller can produce better plant performance than does a si mple fuzzy logic controller, in terms of both environmental and economic ob jectives. This methodology can be extended to control many other types of w astewater treatment processes, as well, by making only minor modifications.