Research On Recommendation Algorithm Of Chain Charging Pile Based On Conflict Resolution

With the vigorous development of the electric vehicle industry,the number of electric vehicles and charging piles continues to grow,and the surge in information makes it difficult for users to make a choice when choosing a charging pile.In this complex many-to-many charging selection mode,the recommended service for charging piles began to appear.Compared with traditional energy supply method in gas station,the charging services provided by charging piles have the characteristics of long service time,unstable prices,limited capacity or service capabilities,and parameter matching,which brings great challenges to the recommendation research for charging piles.Through the research on domestic and foreign research and the existing charging pile recommendation application services on the market,it is found that most of the existing recommendation algorithms are oriented to a single demand,which is likely to cause problems such as recommendation conflicts,duplication,and superposition.Based on the drawbacks of existing research and actual needs,taking user’s needs of distance,time,price and the utilization balance of charging pile into account,this thesis proposes a chained recommendation method for charging piles based on conflict resolution.The main work is as follows:(a)Carry out detailed analysis and precise quantification of the charging needs of users and operators.On the one hand,this thesis analyzes the charging demand of users from the aspects of distance,time and price;on the other hand,for the utilization of charging piles,the variance of the service density of multiple charging stations is used to indicate the equilibrium degree of the utilization of charging piles.(b)Design the chained recommended model for charging piles based on conflict resolution.According to the urgency of users’ charging needs,the chained recommendation mechanism for charging piles is designed to provide personalized,fair and integrated charging recommendation services for users of different warning levels.To solve the problem of recommendation conflict,the dynamic charging area mechanism is introduced into the model,which can solve the problem of recommendation conflicts effectively.In particular,when constructing the recommendation model,the fuzzy programming method is used to convert the multi-objective recommendation task into the single-objective task,and the collaborative filtering recommendation algorithm based on fusion features is introduced to obtain the final recommendation result.(c)Validate the recommendation model and implement the system.The effectiveness of the algorithm is verified by comparing the performance between our method and several benchmark methods;and the rationality is verified by comparing the utility of the recommendation results.The result shows that the recommendation model proposed in this thesis can effectively resolve recommendation conflicts on premise of meeting the multi-level charging needs of users,and can balance the utilization of charging piles to a certain extent.Finally,based on the research results of the laboratory and the Apache Mahout framework,the system is implemented,and modules as the system of demand response,the management database and the chained recommendation engine for charging pile based on conflict resolution are designed to generate recommendation lists based on requests.