AN OPTIMISATION MODEL OF ROAD NETWORK MAINTENANCE PROGRAMME CONSIDERING FLOODING RISK USING HEURISTIC- AND MACHINE-LEARNING-BASED PROCEDURES
Flood is a natural disaster caused by excessive water volume that cannot be accommodated and inundates an area. Floods directly impact the victims such as material losses in the form of money and goods. In addition, the indirect impact felt by the community is the damage to public facilities such as...
Saved in:
| Main Author: | |
|---|---|
| Format: | Theses |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/80058 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | Flood is a natural disaster caused by excessive water volume that cannot be accommodated and inundates an area. Floods directly impact the victims such as material losses in the form of money and goods. In addition, the indirect impact felt by the community is the damage to public facilities such as roads. High rainfall poses a risk of flooding to pavement infrastructure. In fact, dependence on roads is very dominant because it is the backbone of the movement of human activities in Indonesia. Floods are generally unpredictable. Stages of pavement maintenance by related parties are only routine maintenance that has been planned. This causes unpreparedness in terms of costs and funds from the person in charge of road facilities for pavement damage due to floods. So, the purpose of this research is to identify the impact of floods on pavement and then optimize efforts to maintain the road network due to budget constraints in a region. Through optimization analysis with 3 (three) methods namely Greedy Heuristic, Reinforcement Learning, and Naïve Bayes Classifier, each method has its suitability oriented to the expected benefits. Greedy Heuristic is suitable for those that have a tendency of more damaged roads with high flow, making the section have a large fitness value. Reinforcement Learning is suitable for those that have a tendency to maximize the value of benefits with evenly distributed handling. Meanwhile, Naive Bayes Classifier is suitable for those who have a tendency to make decisions with a lot of data cases. However, with the research parameters set, the Greedy Heuristic method provides the highest benefit to cost ratio (BCR) value of 0.016 in the non-flooding scenario and 0.018 in all scenarios with a funding limit of Rp 80M. While the flooding scenario without funding restrictions, the Naïve Bayes Classifier method provides the highest BCR of 0.017.
|
|---|