TY - JOUR
T1 - Generating heuristics to mimic experts in water distribution network optimisation
AU - Walker, DJ
AU - Johns, MB
AU - Keedwell, E
AU - Savić, D
PY - 2018/1/1
Y1 - 2018/1/1
N2 - Evolutionary algorithms (EAs) are well known for their ability to optimise water distribution network designs, however the approach of optimising with respect to mathematically defined objective functions can lead to solutions that, while mathematically optimal, are unsuitable for implementation in the real world. By incorporating human experience into the optimisation process, the result is solutions that are both mathematically optimal and "engineering feasible". This work proposes a method for capturing this expertise in an automated heuristic that can be used to generate new solutions to WDN design problems and is competitive with a well-known heuristic from the literature. The method, in combination with a well-known multi-objective evolutionary algorithm is demonstrated on the Hanoi network.
AB - Evolutionary algorithms (EAs) are well known for their ability to optimise water distribution network designs, however the approach of optimising with respect to mathematically defined objective functions can lead to solutions that, while mathematically optimal, are unsuitable for implementation in the real world. By incorporating human experience into the optimisation process, the result is solutions that are both mathematically optimal and "engineering feasible". This work proposes a method for capturing this expertise in an automated heuristic that can be used to generate new solutions to WDN design problems and is competitive with a well-known heuristic from the literature. The method, in combination with a well-known multi-objective evolutionary algorithm is demonstrated on the Hanoi network.
M3 - Conference proceedings published in a journal
VL - 0
JO - 1st International WDSA / CCWI 2018 Joint Conference
JF - 1st International WDSA / CCWI 2018 Joint Conference
IS - 0
ER -