Recommendation Algorithm And Its Application In Power Marketing

After the purchase and sale of electricity are permitted in the electricity market,the purchaser and the seller can realize the energy transaction through bilateral consultation.However,it is a subject to be studied for purchasers,that is,how large electric power users can find suitable trading objects from a large number of power sellers(generators).In order to promote the development of the power market,the grid marketing staff is responsible for the active matchmaking between the purchase and sale of electricity.However,the user's clear purchase preferences are unknown and a sufficient number of historical data on the purchase and sale of electricity are unable to be obtained.To do this matchmaking work,a large amount of other relevant information must be taken full use of.This paper studies the key technologies of recommendation algorithm,the recommendation algorithm based on user collaborative filtering and its application focusing on the problem of power purchase recommendation under the condition of large user direct purchase of information overload.The main work is as follows:Firstly,the three major types of large-scale online purchase organization of large-scale users in China are introduced in detail.The construction and evaluation methods of the recommendation system are analyzed.The study lays a theoretical foundation for the further research.The user behavior data commonly used in the recommendation system is qualitatively analyzed.Various recommendation algorithms including their application scenarios,advantages and disadvantages are compared in detail.The analysis shows that there are few users with independent transactions in the current power purchase scenario,and the user preferences are not obvious.The user implicit feedback data set and the corresponding recommendation algorithm should be adopted.Then,a power purchase recommendation algorithm based on user collaborative filtering is proposed to the problem of recommendation of the transaction object of large consumers direct-purchasing.The in-depth analysis of the user implicit feedback data set is carried out,and the user behavior data suitable for the user-based collaborative filtering recommendation algorithm is determined.For the variable data in the dataset,the common similarity measurement algorithm is compared and analyzed.The user-based recommendation algorithm is constructed by using the Mahalanobis distance as the core measurement method.At the same time,corresponding handle methods are proposed for other invariant data.In order to consider the impact of industry categories on power purchase behavior,the pyramid model is used for industry processing.Aiming at the direct purchase of electricity transactions by large user users in Shandong Province,the user behavior data is analyzed in detail,and the user behavior data is divided into four dimensions.On this basis,a power user model that can be flexibly applied to different algorithms is established.In order to verify the user-based collaborative filtering power purchase recommendation algorithm proposed in this paper,the publicly available data is used to simplify the user behavior data,and the simulation results are carried out.The results verify the feasibility of the algorithm.The algorithm is applied to engineering practice,and the recommendation system is designed and implemented.Through the analysis of system requirements,four corresponding schemes are proposed,and the application effect of power purchase recommendation system based on user consumption behavior is demonstrated.