Application of evolutionary optimisers in data-based calibration of Activated Sludge Models

Authors:
Jukka Keskitalo;Kauko Leiviskä
Affiliations:
Control Engineering Laboratory, University of Oulu, P.O. Box 4300, FIN-90014, Finland;Control Engineering Laboratory, University of Oulu, P.O. Box 4300, FIN-90014, Finland
Venue:
Expert Systems with Applications: An International Journal
Year:
2012

Citing 4
Cited 1

Differential Evolution: A Practical Approach to Global Optimization (Natural Computing Series)

Differential Evolution: A Practical Approach to Global Optimization (Natural Computing Series)
Data-based identifiability analysis of non-linear dynamical models

Bioinformatics
Systematic identifiability study based on the Fisher Information Matrix for reducing the number of parameters calibration of an activated sludge model

Environmental Modelling & Software
Original article: Subtraction analysis based on self-organizing maps for an industrial wastewater treatment process

Mathematics and Computers in Simulation

A software platform for process monitoring: Applications to water treatment

Expert Systems with Applications: An International Journal

Quantified Score

Hi-index	12.05

Visualization

Abstract

Modelling activated sludge systems has become an accepted practice in Wastewater Treatment Plant (WWTP) design, teaching and research, and Activated Sludge Models (ASM) are by far the most widely used models for activated sludge systems. In most ASM applications, calibration is based on more or less ad-hoc and trial and error approaches. Calibration of the ASMs remains the weakest link in the overall process of modelling biological wastewater treatment. In this paper, a calibration approach is proposed where the need for expert knowledge and modeller effort is significantly reduced. The calibration approach combines identifiability analysis and evolutionary optimisers to automate the ASM calibration. Identifiability analysis is used to deal with poor identifiability of the model structures and evolutionary optimisers are used to identify the model parameters. The applied evolutionary optimisers are Genetic Algorithms and Differential Evolution. Performance of the evolutionary optimisers is compared with a previously proposed calibration approach based on Monte Carlo simulations. All methods were capable of calibrating the model when given enough computation time. However, some of the evolutionary optimisation methods had an advantage in terms of computation time against the Monte Carlo method.