Hierarchically coordinated ultra-short term load forecasting for load aggregator

The load aggregator is a collector, acting as a hub, to link the system control center and multiple electricity users, which integrates multiple end electricity users to participate in the market competition or the system dispatch. The load aggregator is an effective measure to give play to demand response resources. This presents a new demand of ultra-short term (15min) load forecast for the aggregator and the affiliated users. Large scale Advanced Metering Infrastructure (AMI) installation introduces huge real time electricity consumption information which makes it possible to forecast load in a small spatial and temporary scale. To single user or an aggregator with small electricity consumption but relatively large fluctuation, this paper firstly proposes a method to choose the historic similarity day based on fuzzy C-mean clustering. Then wavelet decomposition is adopted to resolve the similarity day load curve, low frequency series and high frequency series is modeled and used to forecast by time series method and neutral network method, respectively. Furthermore, a hierarchical coordination approach is presented to tune the aggregator load forecast and the affiliated individual user load forecast to ensure the whole forecast accuracy and feasibility. The forecasting result for a load aggregator with three users shows the proposed approach is effective and more accurate.