Research On Differential Privacy Clustering Algorithm For Electricity Consumption Of Consumers

With the increasingly prominent global energy problem,many countries in the world urgently seek a green and sustainable development path.The smart grid came into being at the historic moment.Subsequently,many countries have carried out extensive research on smart grid.The Smart grid is a real time power system that includes power generation,transmission,power transmission,distribution,electricity use and dispatch power.Electricity use is the end of the smart grid,which is directly felt by the users,so the quality of its service is very important.Improve the quality of service phase of the stage is to obtain acceptance of one of the most direct and effective way.In recent years,the rapid development of intelligent meter makes it easy to collect users' electricity data.The smart meter realizes the real-time collection of users' electricity information.The massive fine grained power consumption data makes it possible to accurately analyze the user's electricity consumption behavior.Accurate electrical behavior analysis is essential for improving power service,forecasting regional load,rational distribution and reducing grid loss.Clustering analysis is one of the most commonly used technologies for analyzing power consumption data.In this paper,several clustering analysis algorithms have been selected and improved to make them more suitable for electricity data analysis to meet our requirement.However,if the clustering algorithm is not protected during the analysis,it is easy to leak user information.Therefore,in this thesis privacy protection is added to cluster analysis,the differential privacy protection technology is selected for privacy protection of clustering algorithm.In this thesis,firstly CFSFDF and CHAMELEON are improved,and then them are merged to complement each other to form a more suitable clustering algorithm.Secondly,the DPCFSFDP algorithm is added to the CFSFDP algorithm by adding differential privacy protection.This algorithm is suitable for privacy clustering of small data.Finally,this paper presents a method of clustering analysis of consumer electricity using differential privacy protection.This method uses two-stage privacy protection clustering to solve the contradiction between precision analysis and privacy protection.Two phase clustering adopting distributed computation idea includes local clustering and global clustering.Local clustering uses differential privacy adaptive k-means algorithm to complete first cluster electricity consumption data collected by smart meter.In global clustering,a new clustering algorithm basing on density and hierarchy is designed to optimize the result from local clustering.Relevant experimental results show that this method can achieve the purpose of privacy protection and accurate analysis of power data simultaneously.