Download presentation
Presentation is loading. Please wait.
1
An indirect genetic algorithm for a nurse scheduling problem Ya-Tzu, Chiang
2
Introduction To create schedules by assigning one of a number of possible shift patterns to each nurse. These schedules have to satisfy working contracts and meet the demand for a given number of nurses of different grades on each shift, while being seen to be fair.
3
Introduction The higher qualified nurses can substitute less qualified nurses but not vice versa. It is extremely difficult for any local search algorithm as finding feasible solutions.
4
The problem
5
Object function : Subject to :
6
Infeasible solution - GA Penalty functions try to avoid infeasible solutions by steering the search away from them. Repair functions try to fix such solutions so that they become feasible.
7
Approach Combination of GA with a separate heuristic decoder function. The decoder is a constraint handler. Decoder : - Cover - Contribution - Combined
8
Decoder - Cover Determine type of nurse Find shifts with corresponding largest amount of undercover Assign nurse to shift pattern that covers them Nurses’ requests cannot be taken into account by the decoder.
9
Decoder - Contribution Take account of the nurses’ preferences Cycle through all shift patterns of a nurse Assign each one a score based on covering uncovered shifts and preference cost Choose the shift pattern with the highest score
10
Decoder - Contribution
11
Decoder - Combined Combines the feasibility of the ‘Cover’ and features of the ‘Contribution’ Cycle through all shift patterns of a nurse Assign each one a score proportional to its contribution to uncovered shifts and preference cost Choose the shift pattern with the highest score
12
Feasibility cannot be guaranteed Object function
13
Experiments
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.