Enhanced H-GASA Algorithm for Efficient Path Optimisation in Online Ride-Hailing Carpooling

Online ride-hailing carpooling services often need help with bottlenecks, such as delayed response times and low computing efficiency, which negatively impact user experience and platform operation. Current path optimisation algorithms also need help managing real-time dynamic requests and large-scale computing challenges. In this respect, this paper proposes a bi-directional path-based online taxi carpooling optimisation model that considers road network conditions and time window constraints. It minimises operating and passenger travel costs under multiple constraints. The fitness assessment and acceptance criteria are optimised based on a genetic algorithm, combined with the temperature regulation mechanism of simulated annealing, and a hybrid genetic-simulated annealing algorithm (H-GASA) is proposed. In addition, this paper brings the parallel repair mechanism and accelerates the solution repair process using modern multi-core processors and parallel computing framework, significantly improving the solution efficiency. The experimental results show that the H-GASA algorithm substantially reduces the passenger travelling time and vehicle operating cost under multiple time windows, which is better than the existing algorithms and effectively solves the common premature convergence problem of genetic algorithms. The study verifies the efficiency and reliability of the algorithm in practical applications and provides strong technical support for optimising online carpooling services.