Genetically Optimized Infrastructure Design Strategies in Degradable Transport Networks

Abstract

This chapter examines the problem of the resource allocation in degradable road transport networks within a stochastic evolutionary optimization framework. This framework expresses the stochastic equilibrium Network Design Problem (NDP) as a game-theoretic, combinatorial bi-level program. Both the discrete and continuous versions of the reliable NDP are considered in order to address different strategies of network infrastructure investment. The estimation procedure employs a Latin Hypercube sampling method for simulating degradation-inducing variations in users’ attributes and system characteristics, and hence evaluates the network travel time reliability which constrains the solution. This simulation-based risk assessment technique is combined with a genetic algorithm to handle the complex, non-convex nature of the NDP adequately. The test implementation of the proposed framework demonstrates the significant role of incorporating the stochasticity and reliability requirements in the design process to facilitate the selection of the optimal investment strategies in degradable road networks.