A Distributed Reinforcement Learning Solution With Knowledge Transfer Capability for A Bike Rebalancing Problem