This repository implements a stack based simulator incorporating a heuristic to solve the Dynamic Stochastic Bicycle Rebalancing and Charging Problem (DSBRCP). The simulator is event triggered. An event can be triggered by a customer arrival at a station, a bicycle trigger, or a service vehicle arrival at a station, a vehicle trigger. The test instance data is collected from Oslo City Bike. A valid Google Cloud user with access to UIP's data is required.
The model is run from main.py where the following input-parameters should be set:
- start_hour = the hour of the day to start the simulation
- no_stations = the number of stations to include in the subset
- branching = the branching factor in the first stage of the branching algorithm
- subproblem_scenarios = the number of subproblem scenarios to include in each subproblem
- simulation_time = the length of simulation in minutes
When running the simulator, a menu gives the user 7 possible analyses that can be conducted:
- w: weight analysis. Compares the performance of different weight sets.
- c: strategy comparison analysis. Compares the performance of 3 different strategies to solve the DSBRCP.
- r: runtime analysis. Compares the computational time with different input parameter configurations.
- fs: first step analysis. Compares the first step solutions using different input parameter configurations.
- v: vehicle analysis. Compares the performance when using different number of vehicles.
- charge: charging station analysis. Compares the effect of additional charging stations.
- vf: fleet analysis. Compares the effect of a non-homogeneous fleet.
The simulation output is saved in associated excel-files in the Output directory. A visualization of the simulation can be seen by running map_view.html after completing the simulation.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent