Loading...

Control of the Activated Sludge System Using Neural Network Model Predictive Control

Hejazi, Hessam | 2011

632 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 42359 (06)
  4. University: Sharif University of Technology
  5. Department: Chemical and Petroleum Engineering
  6. Advisor(s): Shaygan Salek, Jalaloddin
  7. Abstract:
  8. Activated sludge systems are widespread biological wastewater treatment systems that have a very complex and nonlinear dynamics with a wide range of time constants and, as a consequence, are difficult to model and control. On the other hand, using neural networks as function approximators has provided a reliable tool for modeling complex dynamic systems like activated sludge. In this study a multi-input multi-output neural network model predictive controller (NNMPC) is developed and tested based on the basic control strategy of a benchmark simulation model (called BSM1) suggested by european co-operation in the field of science and technical research (COST) actions 682/624. The controller uses oxygen mass transfer coefficient and internal recycle flow rate as manipulated variables to control dissolved oxygen and nitrate-nitrogen concentration in the proposed activated sludge system. Before implementing the controller, the tuning of the controller is done by trial and error method. The controller is then tested using 14 days dynamic input disturbance data based on dry-weather condition and for evaluation purposes, different performance indices are evaluated. The results of the simulation revealed that NNMPC, if the neural network is well-trained, has an excellent performance for multivariable control of activated sludge systems
  9. Keywords:
  10. Activated Sludge ; Neural Network Model Predictive Control (NNMPC) ; ASM1 Model ; Benchmark Simulation Model (BSM1)

 Digital Object List

  • محتواي پايان نامه
  •   view

 Bookmark

No TOC