Trade-Off Analysis for Multiobjective Optimization in Transportation Asset Management by Generating Pareto Frontiers Using Extreme Points Nondominated Sorting Genetic Algorithm II

摘要

Investment decision making in transportation asset management is typically characterized by a wide diversity of asset types for purposes of optimization at overall system level. To enhance investment analysis and decision making for these multiobjective problem types, the analysis of trade-offs associated with different performance measures can be illuminating and informative. This paper provides techniques for efficient trade-off analysis as part of multiobjective-optimization for transportation asset management. The multiobjective-optimization problem is first formulated by establishing the objectives expressed in terms of network-level performance measures underlying the analysis of trade-offs. Then, the Extreme Points Nondominated Sorting Genetic Algorithm II (NSGA II) technique, an improvement over traditional NSGA II, is applied to generate Pareto frontiers that illustrate the trade-offs. Using candidate projects from a varied range of asset types as a case study, the paper successfully conducts the trade-offs between performance objectives and cost, and then among the performance objectives. The paper also shows that Extreme Points NSGA II has a faster convergence speed and yields a distribution that is superior to the traditional NSGA II.