I am working on Bus schedule optimization. For which, I have read some research papers. Almost all of them have considered passenger waiting time, fleet size, frequency, etc as parameters for genetic Algorithm. I want some information on how these parameters are decided.
I was wondered if you use GAs for Bus scheduling problems, then in addition to problem specific parameters like waiting time, fleet size, frequency, why you did not consider GA's own parameters like crossover and mutation rates, selection pool size. Without having these parameters you can not control the exploration and exploitation mechanism of GAs. I would suggest you to go through production scheduling problems like single machine, job shop, flow shop or open shop scheduling problems where GAs have been designed and implemented to solve the problem.
Have you tried to use mixed-integer linear programming? This method has the advantage that it does not only produce a solution, but also a proof of global optimality, if the instance is not too big (depending on the other structure of your problem, up to 1000 buses).
Please try MILP or Particle Swarm Optimization is ok since bus has velocity.
Or another choice is you write simulation using Arena.
You can take a look on my paper which is related to fleet management.
- Badges
- Science topic
- Similar topics
- Geoscience
- Asia
- India