International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395 -0056
Volume: 04 Issue: 03 | Mar -2017
p-ISSN: 2395-0072
www.irjet.net
Target response electrical usage profile clustering using Big data M.Thilagam1, Ms.J.Kalaivani2, Mrs.P.Hemalatha3 1 2Associate 3Asst
B.Tech (Information Technology), IFET College of Engineering, Villupuram, Professor, Dept.of Information Technology, IFET College of Engineering, Villupuram,
Professor, Dept. of Information Technology, IFET College of Engineering, Villupuram.
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Data streams are very large, quick-changing, and unable to calculate. Clustering is a prominent task in mining data task; it can group same kind of objects in a cluster. The aim of choosing a Re-Cluster subset group of good characteristics with respect to the goal concepts, feature subset selection is an effective way for reducing dimensionality, removing irrelevant data, accuracy learning, and improving outcome unambiguousness. While the effectiveness concerns the point in time necessary to find a recluster division of features, the efficiency is related to the value of the subset of features. In this, proposed clustering related to division selection algorithm works in two steps. In the first step, further are divided into clusters by using theoretic graph clustering methods. In the second step, the most representative feature that is strongly related to target classes is selected from each cluster to form a subset of features. To confirm the algorithm efficiency, we are working to use mRMR method with heuristic procedure. Heuristic algorithms used for solving a problem more quickly or for finding an approximate rearrange the cluster subset selection solution. Minimum Redundancy Maximum Relevance (mRMR) variety used to be more controlling than the extreme consequence selection. It will provide active way to expect the efficiency and success of the clustering based subgroup collection algorithm. Key Words: Cluster analysis, Load profiling, big data, Markov model, behavior dynamics, distributed clustering, demand response. 1. INTRODUCTION All over the world have some set of goals to implement the power system in monopolistic area mainly focused on demand side. Now days the load serving entities (LSEs) is used development of high values. To have a better understanding of electricity consumption patterns and power managements are effective ways to enhance the competitiveness of LSEs. It has been revolutionizing the electrical generation and consumption by a two-way flow of power data. Most important data source from the demand side, advanced metering infrastructure (AMI), has gained increasing popularity worldwide; AMI allows LSEs to obtain electricity consumption data at high frequency, e.g., Š 2017, IRJET
|
Impact Factor value: 5.181
|
minutes to hours Large volumes of electricity [16] consumption data reveal .By the Research Report, the determine that smart meters will surpass 1.1 billion by 2022 . AMI will collect the electricity usage data profile in the range among 1 hour; This will increase in the amount of usage of electricity will processed in the past years. It means that by 2022 the electric utility of power in industry will be increase the data annually from smart meters. The primary and secondary value embedded in the high density and same data sets from power distribution systems. Aggregated load has already been successfully modeled using top-down methods. Singh et model distribution system load and Valverde et al. model load for load flow analysis with Gaussian mixture models to capture the probability density functions. However, autocorrelation found in electricity request of households was never combined. Bottom-up methodologies have in general good results because of the incorporation of a performance model. Top-down approaches have a lot of potential because of the lower modeling intensity: there is no need to model every appliance individually, which lowers the intensity of modeling significantly. The detection of behavior is in general done by pattern analysis. Techniques have been developed to find similarities within load profiles as between profiles within different domains such as clustering or classification of profiles forecasting selecting scenarios for load-wind combinations and selecting demand response policies a new short-term load forecasting framework based on big data technologies is proposed in this paper. In Section II, the framework and relevant techniques of the short-term load analysis and forecasting method are presented in detail. Section III introduces a technical framework of the proposed method using big data technologies. Section IV provides case study results. Section V concludes this paper. In general, short-term forecasting methods perform direct forecasting of the total system load using historical load data and weather data as inputs. However, since the grid consists of thousands of individual users and many time varying characteristics, a single forecasting method, such as those mentioned earlier, cannot adequately forecast individual loads, as well as the accompanying factors that influence the
ISO 9001:2008 Certified Journal
|
Page 2332