Forecast of Bicycle Sharing Demand Using BP Neural Network based on Improved Brain Storm Optimization

Public bicycle sharing systems can effectively improve the overall service level of urban public transport. The quality of bicycle sharing system depends on the accurate prediction of the public bicycle demand for each rental point. BP neural network (BPNN) could forecast public bicycle sharing demand. Based on the discussion that BPNN is sensitive to its initial weights and it is easy to fall into local minima, brain storm optimization algorithm is adopted to find optimum initial weights and thresholds It can be found that the algorithm has no fusion operation with the cluster center and the random individuals, which will easy to be premature. A novel improved BSO algorithm with 2 new regeneration modes is put forward in order to solve this problem,. The results show that the BPNN based on the improved brain storm optimization outperforms general BPNN in public bicycle sharing demand prediction.