EURO 2024 Copenhagen
Abstract Submission

EURO-Online login

3917. Student Timetabling Optimisation for the Australian Defence Force

Invited abstract in session TA-20: Military, Defense, and International Security II, stream Military, Defense, and International Security.

Tuesday, 8:30-10:00
Room: 45 (building: 116)

Authors (first author is the speaker)

1. Katie Mortimer
Defence Science and Technology Group
2. Terry Caelli
Electrical and Electronic Engineeering, University of Melbourne

Abstract

Personnel in the Australian Defence Force must complete a number of training courses throughout their career. Currently, timetabling is performed manually across hundreds of courses and associated sessions, resulting in inefficient or incorrect timetables. To address this, we have created an optimisation algorithm to improve timetabling efficiency and administrative workloads. For this problem, there are a number of hard constraints, including: a maximum number of students per session, students cannot attend overlapping courses, and a number of fixed training pathways or course sequences. Soft constraints such as prioritisation of students is also important. In this domain efficiency is defined in terms of the total time delays in students being able to complete their training.

The algorithm uses a combination of Dynamic Programming (DP) and Genetic Algorithms (GA) to create optimal timetables where DP is initially used to find the optimal timetable for each student over course pathways and session options - which decrease as each timetable is generated over the student priority queue. The GA is then used to, where possible, improve upon these initial solutions by considering more efficient permutations and combinations of specific course and sessions over all the student DP solutions. The results showed a 45% increase in efficiency, reducing the average student delay by approximately half compared to the manual solution, and in a fraction of the time.

Keywords

Status: accepted

Back to the list of papers