amirkfard/enhancedga

Enhanced Genetic Algorithm Approach for Multi-Year Maintenance and Rehabilitation Optimization for Large-Scale Infrastructure Networks

Enhanced Genetic Algorithm Approach for Multi-Year Maintenance and Rehabilitation Optimization for Large-Scale Infrastructure Networks

This repository hosts the datasets, source code, and models associated with the research paper titled "Enhanced Genetic Algorithm Approach for Multi-Year Maintenance and Rehabilitation Optimization for Large-Scale Infrastructure Networks". This work tackles the significant and complex challenge of optimizing multi-year network maintenance and rehabilitation, a pivotal task in the management of infrastructure assets.

Traditional genetic algorithms (GAs), while commonly used in this domain, tend to falter as the scale of the network expands, often disrupting viable solutions due to limitations in standard crossover and mutation techniques. In response to this issue, our study proposes an advanced GA framework that introduces innovative methods for maintaining the integrity of solution structures and enhancing solution search efficiency, thereby significantly improving performance even for large-scale networks.

By employing a novel crossover method that swaps blocks of genes representing annual plans, and a unique mutation approach integrating linear programming to adjust plans according to varying budget scenarios, our enhanced GA method effectively addresses the complexities of infrastructure optimization. The repository provides a comprehensive package for applying this enhanced approach to real-world infrastructure networks, aiming to facilitate more effective and efficient planning processes.

We demonstrate the effectiveness of our hybrid LP-GA approach through two practical case studies: one focusing on a small-scale sewer network flushing program, and the other on a larger scale involving 13,610 pavement segments. Results from these case studies indicate that our proposed algorithm not only achieves rapid convergence but also maintains a 100% rate of generating feasible solutions, reaching optimal or near-optimal outcomes efficiently.

This work provides a sophisticated algorithmic tool for the field of infrastructure asset management, paving the way for further innovations in the sector.