Load estimation for load monitoring at distribution substations

This paper addresses the issues related to the real-time monitoring of loads at distribution substations. A method is proposed for estimating the measurements that become unavailable due to metering problems. To address the low redundancy associated with the load monitoring scheme, the method uses a regression-based model and makes use of the strong correlation between the loads that are geographically close to each other. The method was tested with actual field data. The results indicate that the method has the acceptable performance for measurement loss of up to a week. The average estimation error varies 2-7.5%, depending on which measurement is lost and how up-to-date the historical data is.