Location Optimization And Recommendation Algorithm For Electric Taxi Charging Station Based On Charging Behavior

The widespread promotion of electric taxis on a large scale is an important way to achieve the rapid development of the new energy automobile industry and the national "dual carbon" goal.Electric taxis not only need to continuously improve new energy technologies and increase their cruising range,but also need to provide sufficient public charging facilities in a reasonable operating area.Based on the big data analysis of taxi users,effective charging station layout and charging strategy optimization are of great significance to meet the timely charging needs of electric taxis and promote the development of the electric rental industry.However,there are still many difficulties in how to analyze and characterize the charging behavior of taxis to support the optimization of the layout of charging stations.For example,how to mine and predict the actual charging demand of electric taxis? How to incorporate the charging behavior into the traditional charging station layout optimization model? Will differentiated charging points of interest(POI)affect the charging behavior of users? In response to these problems,this thesis will use the big data analysis of electric taxis in Shanghai to deeply describe the charging behavior of electric taxis,revealing the key elements that affect charging behavior and charging demand;then,a two-stage mixed integer programming model is used to conduct charging stations.Layout optimization.At the same time,this paper further proposes a charging recommendation algorithm based on the implicit semantic model to improve the intergroup effect,and designs a user charging strategy based on user collaborative filtering similarity,a user charging strategy based on K-means,and a final user charging recommendation strategy.Thus,it provides support to improve the efficient use of electric taxis and low-carbon travel in cities.Specifically,this paper has carried out the following specific research work.(1)Research on user behavior characteristics and charging station operating characteristics based on charging order dataFirst,according to related research such as literature,establish the descriptive attribute indicators for the charging behavior of electric taxi drivers and the operation of charging stations;then,taking Shanghai as an example,use the order data of charging stations to generate indicator data for drivers and charging stations;finally,Use clustering algorithm to classify different taxi drivers and charging stations,and analyze the charging behavior characteristics of different types of electric taxi drivers and the operating characteristics of different charging stations to provide support for subsequent charging station location optimization and recommendation algorithm design.(2)Research on charging station service demand theme mining based on user charging evaluation dataIn the process of using the platform for charging,electric taxi drivers generate and accumulate a large amount of service evaluation information.This information is mainly for complaints or complaints about service defects at charging stations.This paper will use natural language processing technology to standardize the evaluation text,and use the LDA topic algorithm to extract and mine the service defect topic words in the review content to provide a basis for subsequent research.(3)Research on the location and layout of charging stations considering user needsFor the overall layout and optimization of charging stations,in addition to traditional path and location planning,full consideration must be given to the actual needs of users and the operating status of existing charging stations.Based on the data analysis of the first two chapters,this paper fully considers the actual charging needs of taxi drivers,and optimizes the layout of existing charging stations based on the operating status of existing charging stations and related factors such as routes and locations.(4)Research on user charging recommendation based on implicit semantic model to improve between-group effectsThis thesis first collects charging station data and user data.The former includes charging station ID,charging station geographic location,charging speed and other information,while the latter includes user ID,user consumption level,user model,user average charging time and other data.Furthermore,this paper proposes a recommendation algorithm based on a hidden semantic model,which can effectively predict the user's preference for different charging stations,making the recommendation more reasonable and effective.(5)User charging recommendation strategy based on scoring mechanismThe interest rating of electric taxi drivers on the surrounding environment of the charging station can affect the overall rating of the charging station to a certain extent.Thus,the overall impression of other users of the charging station is changed.When the charging station in the camp is in normal operation,the rating of the charging station will become the main influencing factor for attracting electric taxi drivers to charge.Then,we further proposes a user charging strategy based on user collaborative filtering similarity,a charging strategy based on K-means,and a final user charging recommendation strategy.The main innovations of this thesis are:(1)An optimization model for charging station location based on two-stage integer programming is proposed.The goal of the first phase is to satisfy the travel needs of all electric taxi users to the greatest extent while meeting the limit on the number of charging stations.The goal of the second stage is to meet the charging needs of all users in the area and minimize operating costs on the basis of comprehensive consideration of location,price,environment,and service experience restrictions.Based on this,this thesis uses Python language to build a new algorithm for the first stage model in the environment of Cplex and Doclpex.The algorithm is a heuristic method,combined with many actual situations,choose to use Matlab to call the intlinprog function for analysis.Actual case analysis was carried out to verify the effectiveness and applicability of the algorithm.(2)Propose a charging recommendation algorithm based on implicit semantic model to improve the between-group effect.This paper improves the original implicit semantic model,and proposes an inter-group effect model that incorporates users' historical charging behavior,user type,and driving speed.Using this model,the user's preference value for different charging stations can be predicted more accurately,and the candidate charging station can be obtained by matching the user's current location and remaining power data,thereby giving a more reasonable and effective charging station recommendation result.A more accurate prediction of user preferences can be achieved,making the recommendation more reasonable and effective.(3)A user charging recommendation strategy based on the scoring mechanism is proposed.This paper considers taxi drivers' comments on charging stations and proposes a matrix decomposition collaborative filtering recommendation algorithm based on POI weighting to solve the problem of confusion and difficulty in selecting current charging station user ratings.Due to the large difference in the value of the weighted POI scores around the charging station,this paper corrects the average value and obtains the final predicted score,so as to recommend the charging station users.The Foursquare data set verifies the effectiveness of the algorithm by comparing the two statistical indicators of precision and recall.And considering the impact of user age on charging behavior,based on the K-means algorithm,users of different age groups are evaluated,and the evaluation results are used as a standard to accurately recommend users of the same age group.