Charging Pile Recommendation Method For Idle Electric Taxis Based On Recurrent Neural Network

With the increasing global energy crisis and the depletion of oil resources,electric taxi technology has been developed rapidly,more and more electric taxis have appeared.Although the electric taxis are of great significance to satisfy the traffic demand,reduce the oil consumption and pollution emissions,the electric taxis are limited by battery power,and they need to be recharged by some charging piles before their batteries have been exhausted.Generally,before the power is exhausted,the electric taxi must drive to the charging point near the given route to charge,and it is expected that the extra movement is minimized.Idle electric taxis do not have specific destinations,and thus it is very hard to recommend the optimal charging piles for idle electric taxis.To this end,this thesis proposes a Charging Pile Recommendation Method for Idle Electric Taxis(CPRM-IET)based on recursive neural network to recommend the optimal charging piles for idle electric taxis.Typically,the movement of each idle electric taxi depends on the subconscious movement tendency and driving habits of the driver.Therefore,it is necessary to predict the future movement based on its historical trajectories,so as to find the charging pile with the least extra movement.In CPRM-IET,a dual-stage attention-based recurrent neural network(DA-RNN)model is provided to predict the future trajectories of idle electric taxis.DA-RNN model includes two types of attention mechanisms: input attention mechanism and temporal attention mechanism.The input attention mechanism assigns weights to the input driving sequence at each time slot,and the temporal attention mechanism assigns weights to the hidden states of the encoder.According to the predicted trajectory,the proper charging pile is recommended for the electric taxi.In the process of charging pile recommendation,the extra movement and waiting time are considered,that is,the charging pile is recommended for the electric taxi in the spatio-temporal dimension,so that the electric taxi will pay the least expense if it chooses the recommended charging pile.the extra movement refers to the increased path length that the electric taxi needs to travel to the charging pile.Waiting time refers to the time required for an electric taxi to wait for the charge after arriving at the charging position and the time required for the electric taxi to complete the charging process.The simulation results show that CPRM-IET can achieve preferable results in terms of extra movement and root mean square error,which reflects that CPRM-IET can accurately predict the future trajectories of idle electric taxis and recommend optimal charging piles for these electric taxis.This thesis develops a prototype system of charging pile recommendation based on android system to implement DA-RNN algorithm and CPRM-IET algorithm.This thesis mainly introduces the development environment,functional modules and prototype implementation,and analyzes the prototype test results.The results show that the prototype system has achieved good results in predicting the running time of the future trajectory sequence of the electric taxi,the prediction accuracy,and the recommendation accuracy of the charging pile for the electric taxi,which shows that the prototype system has well realized the DA-RNN algorithm and CPRM-IET algorithm.In this thesis,the minimum cost of electric taxi is calculated by using the extra movement and waiting time of charging.The combination method can not reflect the weight of time factor and space factor.Therefore,in the future work,we will also seek a better compromise way to recommend charging piles to electric taxis,which can balance the factors of time and space.