Genetic Algorithm Based Sustainable Traffic Management in Smart Cities: A Case Study for Istanbul

The increasing population density in cities necessitates the improvement of bus schedules to meet passengers' travel requirements. In metropolitan cities like Istanbul, minimizing waiting times and optimizing departure times are crucial for enhancing the travel experience. There is a trade-off between the number of bus stops and the waiting time of passengers. Short waiting times reduce inconvenience and contribute to a more efficient journey. Therefore, this paper proposes a novel optimization model for bus schedules in smart cities, comprising data selection, prediction, and optimization layers. Using a dataset from Istanbul, we analyzed the hourly distribution of passengers, clustered similar lines, and selected the Avcdar-Zincirlikuyu line to determine minimum waiting times and optimal departure times. Employing the N-BEATS deep learning model, we predicted new passenger flow distribution with high accuracy. Based on this distribution, we analytically modeled a Bus Scheduling Optimization Model to minimize trips and travel costs, and proposed Greedy and Genetic Algorithm-based solutions for the bus scheduling problem, comparing their performance. Results show the genetic algorithm reduces total waiting time by 32.8% compared to the greedy approach, and the total number of bus trips decreased from 110 to 89. These algorithms, based on the N-BEATS model with 98% R-Square success, demonstrate the effectiveness of the proposed method over conventional travel management. Our results indicate that the genetic algorithm-based bus scheduling approach significantly enhances passenger travel experience and reduces waiting times at bus stops, offering a promising method for revolutionizing travel management systems in smart cities.