Abstract
This paper applies a genetic algorithm to static and dynamic models of a case study water treatment works to find near optimal designs. The mechanisms of these models, their calibration and accuracy are described. The models were used with stochastic data representative of conditions observed at the works and the NSGAII genetic algorithm was applied to minimise the size of the works and the failure likelihood. The dynamic model was found to predict more conservative designs than the static model. The genetic algorithm was found to require greater calibration to identify near-optimal solutions efficiently.
Original language | English |
---|---|
Title of host publication | Machine Learning in Water Systems - AISB Convention 2013 |
Pages | 20-24 |
Number of pages | 5 |
Publication status | Published - 1 Dec 2013 |
Event | Machine Learning in Water Systems, Held at the AISB Convention 2013 - Exeter, United Kingdom Duration: 3 Apr 2013 → 5 Apr 2013 |
Conference
Conference | Machine Learning in Water Systems, Held at the AISB Convention 2013 |
---|---|
Country/Territory | United Kingdom |
City | Exeter |
Period | 3/04/13 → 5/04/13 |
ASJC Scopus subject areas
- Human-Computer Interaction