On-line load estimation for distribution automation using AMI data

This paper presents a method to estimate the loads on a distribution feeder in real-time with limited real-time customer load data from AMI. The method consists of two modules: a load clustering module and a load modeling module. Load clustering helps to identify customer loads with similar profiles thereby helping to determine the subset of customers that need to be monitored in real-time. The load modelling module constructs statistical models for each load on the feeder. The module uses both the historical AMI data and the real-time AMI data from the selected customers in order to improve the performance of the model. Test results based on actual AMI data show the effectiveness of the method.