Cloud-Based Massive Electricity Data Mining and Consumption Pattern Discovery
With the development of the power systems in China, there is large volume of basic electricity consumption data accumulated. Mining these data to discover possible consumption patterns and group the users in a more fine-grained way can help the State Grid
- PDF / 2,322,111 Bytes
- 15 Pages / 439.363 x 666.131 pts Page_size
- 106 Downloads / 248 Views
2
Shandong University of Science and Technology, Qingdao, Shandong, China State Grid Corporation of China, Qingdao power supply company, Qingdao, Shandong, China [email protected], [email protected], [email protected]
Abstract. With the development of the power systems in China, there is large volume of basic electricity consumption data accumulated. Mining these data to discover possible consumption patterns and group the users in a more finegrained way can help the State Grid Corporation to understand users’ personalized and differentiated requirements. In this work, an algorithm called TMeans is proposed to mine the electricity consumption patterns. TMeans improves the classical K-Means algorithm by presenting a set of static and dynamical rules which can dynamically adjust the clustering process according to the statistical features of the clusters, making the process more flexible and practical. Then a MapReduce-based implementation of TMeans is proposed to make itself capable of handling large volume of data efficiently. Through experiment, we first demonstrate that the consumption patterns can be effectively discovered and can be refined to very small granularity through TMeans, and then we show that the MapReduce-based implementation of TMeans can efficiently speed up the clustering process. Keywords: electricity consumption pattern, cluster, cloud, map-reduce, massive data.
1
Introduction
Smart Grid is a modernized electrical grid that uses information and communication technology to gather and act on information, such as information about the behaviors of suppliers and consumers, in an automated fashion to improve the efficiency, reliability, economy, and sustainability of the production and distribution of electricity [1]. Intelligent electrical information collection and analysis is one of its key tasks. With the development of the national power systems in China, there is large volume of basic electrical data accumulated, like voltage, load, current, etc. In recent years, the data volume takes a sharp increase owning to the refined metering facilities in Smart Grid. Qingdao, a city in China with a 7 million populations, can generate the electrical data up to 3.1G in a day. These data are not only massive, highly-concurrent and disperse, but also have potential correlations among them, such as users’ electricity consumption habits. Mining these data to discover possible consumption patterns and group the users in a more fine-grained way can help the State Grid Corporation to Z. Huang et al. (Eds.): WISE 2013 Workshops 2013, LNCS 8182, pp. 213–227, 2014. © Springer-Verlag Berlin Heidelberg 2014
214
M. Chen, M. Cao, and Y. Wen
understand users’ personalized and differentiated requirements, and to provide a grounded foundation for the electricity delivery patterns and the price policies [2]. To this end, this work tries to mine consumption patterns from massive electrical data. It has the following challenges: 1) How to develop a mining algorithm that can discover electricity consumption patterns effe
Data Loading...
