Research And Application Of The Behaviors Of Users Demanding For Large Amounts Of Electricity Based On The Clustering Algorithm

With the development of computer technology and network technology in power system and enterprise, the data of power enterprise is increasing, which reflects the situation of power supply enterprises in a certain extent for a long time. How to apply the technologies of big data and cloud computing to exploring the behaviors and characteristics of users demanding for large amounts of electricity, and provide customized power service for users demanding for large amounts of electricity has become the focus of competition in the electricity market. At the same time, it can also bring new directions for the development and progress of the electricity industries.In this paper, firstly we introduces the clustering algorithms, the domestic and foreign research situations and research results of the behaviors of users demanding for large amounts of electricity, and analyzes the characteristics of the behaviors of users demanding for large amounts of electricity, and then studies the methods of using the K-clustering algorithm and big data technologies to mine and analyze data on electricity; Secondly, we introduces the common clustering algorithms and similarity measures, and elaborates the platforms of Spark and the resource management YARN used in this paper.Then this paper analyzes the K-means algorithm. For the shortcomings of the K-means algorithm, an improved algorithm based on the optimization of the initial clustering centers is proposed. In order to verify the clustering validity and speedup ratio, the UCI data sets and electric power big data sets on electricity are used to implement the improved algorithm and the applications are submitted to the cluster of Spark for the parallelization of the improved algorithm. The comparison of the experimental results display that the improved algorithm has better clustering effects and speedup ratios than the existing K-means algorithm.Finally, in order to further study the behaviors and characteristics of users demanding for large amounts of electricity, this paper analyzes that from the aspects of user segmentation, distribution transformer overload warning, tips of electricity optimization, electric power outage plan management respectively. Using software engineering methods and basic procedures from the aspects of system requirements analysis, development environment, structural design, module using cases and system tests, the system of behavior analysis of users demanding for large amounts of electricity based on the Spark platform is designed and developed depending on the analysis results. By big data technologies and visualization technologies, the behaviors of users demanding for large amounts of electricity are demonstrated vividly. The scheme applying the big data technologies to power industries has been realized in the system.