Linkage Clustering for Aggregated Baseline Loads Estimation Methods Considering Incentive-Based Demand Response

In the incentive-based demand response (IBDR) mechanism, the aggregated baseline load (ABL) is constructed by the load aggregator through the customer baseline load (CBL), which provides an accounting basis for the operator to implement the economic compensation mechanism. Most of the current methods adopt the step of combining clustering and estimation, but due to the lack of synergy between clustering and estimation objectives, there is still room for significant improvement in estimation accuracy. This paper proposes an ABL modeling framework based on the EM to determine the optimal number of clusters linked with estimation, and the core idea is to enhance the ABL estimation accuracy by optimizing user clustering. Therefore, a two-stage CBL estimation framework is adopted: Firstly, initial clustering is implemented to initially classify the load of the optimal number of clusters for all users in order to generate the input dataset for second-stage clustering. Secondly, second-stage clustering is implemented and the clustering is adjusted according to the performance of estimation results on the validation set, which in turn leads to the optimal clustering results. Simulation experiments based on real data show that the method exhibits obvious advantages over the traditional ABL estimation method.