Mitigating Effect of Communication Link Failure in Smart Meter-Based Load Forecasting, NCCS-2018
Published in Springer, Singapore, 2018
Abstract: With the ever-increasing number of smart meter installations, an enormous amount of power consumption data is collected by these meters in real-time. The availability of this large amount of power consumption data has changed the way power system analyses were done traditionally; one such area being load forecasting. The load forecasting is now largely data-driven and hence failure to receive data from the smart meters leads to forecasting errors. The present paper targets to solve this problem by introducing a novel classification-based load forecasting methodology that enables day-ahead load prediction in case of missing data due to communication link failure. In this method, the loads are classified or clustered in sub-classes based on the amount of consumption and then day-ahead forecasting is done using this clustered load data. The proposed methodology is demonstrated using the data collected in a practical smart system installed at the NIT Patna campus.