Study On The Taxi Scheduling And Optimization Based On Artificial Fish-swarm Algorithm

Taxi, as a supplement to urban public transport vehicles, provides a lot of conveniencefor people, and plays an increasingly important role in urban transportation. However, theextensive taxi management model and the backward way people hail a taxi at the roadsidehave led to that the information obtained by the taxi drivers and passengers is asymmetrical,which makes people hard to take a taxi and causes the high rate of taxi empty crossing.Meanwhile, it has increased traffic congestion, environmental pollution and other problems,which have seriously affected the quality of life.The development and application at home and abroad of the taxi dispatch technology arefull researched in this thesis. According the management characteristics and technicalrequirements of the taxi industry, a system framework for taxi scheduling and management isdesigned based on Beidou positioning technology (BD), the third generation mobilecommunication technology (3G), and Google Maps API technology. The system includes sixparts: vehicle terminal, communications platform, call center, messaging platform, schedulingplatform and management platform. The system is desighed as B/S mode. Windows Server2003operating system, SQL Server2005database, Microsoft Visual Studio2008integrateddevelopment environment, ASP.NET&C#development language and Google Maps API areused in this system.The focus of this thesis is the research and improvement of the shortest path algorithmfor taxi dispatch system. To solve basic artificial fish-swarm algorithm(AFSA)’s drawbacks oflow convergence rate in the latter stage, a large amount of computation and easiness oftrapping in local optimal solution, which are all caused by the constant vision of the artificialfish, an improved artificial fish-swarm algorithm based on adaptive vision(AVAFSA) isproposed. In the improved algorithm, only the vision of the preying behavior of artificial fish is adjusted, inorder to make the vision gradually decrease with the increase of the number ofiterations of the algorithm. When the value of the vision becomes less than half the initialvalue, it makes the value be equal to half the initial value. The proposed improved artificialfish swarm algorithm is applied to the static shortest path problem based on road network toprovide customers with the best path. Simulation results depict the improved algorithm has ahigher convergence rate and a smaller amount of calculation, and is more accurate and stablethan the basic AFSA and ant colony optimization (ACO).