خوارزمية اختيار أفضل شبكة حافلات لمدينة كربلاء

Optimization Algorithm for Bus Network in Karbala

    Traffic congestion in many cities around the world causes an excessive increase in travel time, transportation cost, traffic accidents, and the level of environmental pollution. Reducing the volume of trips made by private cars, along with increasing the attractiveness of public transportation are vital actions that can partially alleviate these traffic issues. Persistent efforts being made by traffic authorities in many cities to increase transit network coverage, service frequency, and reliability. This is reflected in the vast number of studies in transportation engineering literature that sought to analyse, evaluate, enhance and design existing and new transit route networks.

    However, research on solving transit network design problem (TRNDP) is still emerging. Thus, this study aims to suggest a method to optimize the geometry of transit route network based on non-parametric techniques.

    Genetic Algorithm (GA) along with spatial analysis tools were used to find an optimal geometry for the bus network in Karbala city. This is formulated as an optimization problem that reduces the overall system cost and expresses in a function of passenger total travel time and bus operating cost. This study was processed in two stages: in the first stage, candidate route generation algorithm (CRGA) was used to generate a set of initial bus lines. In the second stage, a mathematical model was developed based on natural genetics to obtain the optimal solution from the initial candidate route set that developed in the first stage. The multi-objective nonlinear mixed-integer model was developed through MATLAB programming language using data of medium-sized network belongs to the city of Karbala, Iraq.

     The comparison of the optimal solution resulted from the second stage with the initial solutions in the first stage showed that the optimal solution offered the least user time, fleet size, bus kilometres and therefore the lowest total system cost. The total system cost of the optimal solution was (22, 9, 21, and 11) % less than the cost of the four initial solutions respectively. In this way, the developed model using natural genetics approach outperforms local search algorithm principles.