Performance Prediction of the Full-Scale KNR Biological Nutrient Removal Process Using Multivariate Dynamic Process Model with the Grammar-Based Genetic Programming (GBGP) System

Abstract

Biological wastewater treatment systems are characterized by large temporal variability of inflow, variable concentrations of components in the incoming wastewater to the plant, and highly variable biological reactions within the process. This paper proposes an automatic process model induction system using an evolutionary computional intelligence, called grammar-based genetic programming (GBGP), that is specially designed to automatically discover multivariate dynamic process models that best fit observed process data. This automatic process model induction system combines an evolutionary self-organizing system of genetic programming paradigm with various mathematical functions for a multivariate nonlinear model evolution using a grammar system via the mechanism of genetics and natural selection. In the GBGP, a four-step modelling procedure was performed: (1) initialization: generatr an initial population of P models using contex-free grammar. generation K=0; (2) optimize each model (variables and constants) in the population; (3) execute and evaluate the fitness of each model in the population; (4) genetic loop: repeat until termination criterion is met (maximum generation Kmax). The results demonstrate that the multivariate dynamic process modelling technique presented using a grammar-based genetic programming (GBGP) provides a valuable tool for predicting the out puts with high levels of accuracy and identifying key operating variables for the full-scale KNR biological nutrient removal process. These modells are derived automatically in the form of understandable mathematical formulas that enable engineers to extract important knowledge hidden in the data and develop better operation and control strategies.