DISCUSSION OF MULTIOBJECTIVE OPTIMIZATION IN DECISION MAKING IN
DISCUSSION OF MULTI-OBJECTIVE OPTIMIZATION IN DECISION MAKING IN TRANSPORT INFRASTRUCTURE ASSEST MANAGEMENT Lin CHEN Chang’an University, Xi’an, China July 12, 2017
About me Chang’an University China University of Auckland New Zealand
About me University of Pretoria South Africa
Content 1. Road Development in China 2. Infrastructure Asset Management 3. Decision Making 4. Multi-Objective Optimization
1. Road Development in China History: 1949: Management department Laws Lower classes of highways 1980 s: Medium and high classes of highways
1. Road Development in China Expressway Development in China
1. Road Development in China 4. 69 million km Main Road Networks in China
2. Transport Infrastructure Asset Management What & When?
2. Transport Infrastructure Asset Management efficiently manage a road network in order to achieve goals and requirements What treatments should be used? When should I apply the treatment? How to implement those treatment? What are the outcomes of my management?
3. Decision Making (DM) attempts to identify the proper strategies for an infrastructure asset network Strategies Analysis period Strategy Index 2017 1 Treatment 1 ●●● 2 Treatment 2 ●●● 3 2018 2019 Treatment 2 4 ●●● Treatment 3 ● ●●● ●●●
3. Decision Making (DM) attempts to identify the proper strategies for an infrastructure asset network
3. Decision Making (DM) attempts to identify the proper strategies for an infrastructure asset network Cost
3. Decision Making (DM) attempts to identify the proper strategies for an infrastructure asset network Condition
3. Decision Making (DM) attempts to identify the proper strategies for an infrastructure asset network ? Condition Cost
3. Decision Making (DM) • A wide range of considerations Various variables • A large number of segments and strategies Powerful & cheap computing power • Conflicting goals Multiple objectives • A number of requirements • Subjectivities Multiple constraints No expert experience needed Multi-Objective Optimisation (MOO)
4. Multi-Objective Optimization Multi-Objective Optimisation (MOO) describes Multi-Objective Decision Making more practically and rational, especially when objectives cannot be reasonably aggregated into a single objective.
4. Multi-Objective Optimization Multi-Objective Optimisation (MOO) - Abilities Pareto solutions: Best possible solutions
4. Multi-Objective Optimization Multi-Objective Optimisation (MOO) - Abilities Best achievable outcomes
4. Multi-Objective Optimization Multi-Objective Optimisation (MOO) - Abilities Outcome relationship Outcome trade-offs
4. Multi-Objective Optimization Multi-Objective Optimisation (MOO) - Abilities Outcome relationship Outcome trade-offs
4. Multi-Objective Optimization MOO has the potentials to help with decision making in TIAM What are the applicable methods? Which one is suitable for my problem? How to implement it?
4. Multi-Objective Optimization Weighted Sum Method (WSM) Dichotomic Approach (DA) Epsilon Constraint method (ECM) Exact method Revised Normal Boundary Intersection (RNBI) Genetic Algorithm (GA) Case Study Nondomianted Sorting Genetic Algorithm II (NSGA II) Simulated Annealing (SA) Tabu Search (TS) Ant Colony Optimization (ACO) Particle Swarm Optimization (PSO) Heuristic
4. Multi-Objective Optimization Comparison of MOO methods Criterion WSM DA ECM RNBI GA NSGA II SA TS ACO PSO Type Objectives Constraints exact method √ 2 heuristic √ √ type Large size hardness √ Solutions distribution quality & distribution Speed depends controllable & slow Implementation theory & math framework
4. Multi-Objective Optimization Solution Quality
4. Multi-Objective Optimization Solution Distribution Good distribution Poor distribution
4. Multi-Objective Optimization Comparison of MOO methods Criterion WSM DA ECM RNBI GA NSGA II SA TS ACO PSO Type Objectives Constraints exact method √ 2 heuristic √ √ type Large size hardness √ Solutions distribution quality and distribution Speed depends controllable & slow Implementation theory & math framework
Thank you!
- Slides: 27