Identification of Similar Loads for Electric Power Management in Smart Grid

Abstract

This paper shows the result of application of techniques of characterization to define load’s signatures in a Smart Grid. The difference is those loads has the same brand and data sheet making the identification process more difficult and describing a condition challenging: highly similar loads. In this work is used a platform with four technical identic compact fluorescent lamps allowing sixteen possible operation configurations. For each one configuration are measure and record 150 cycles of total circuit current and voltage with 25 kHz sampling frequency. Then are tests two ways to define load’s signatures: with fourteen simple characteristics for each configuration and other way based on Shannon and Renyi entropy totally 5, 15, 45 or 85 characteristics for each one of the
sixteen possible operation configurations. These characteristics represents the load signature. The next step is to test the load signature into the classification’s system to find the accuracy of each method of characterization. The best accuracy has been about 74 percent and is close to the results finds in literature when the identification of different loads, proving the efficiency of present methods.