Details of Research Outputs

In this paper, we propose a novel algorithm for tracking the Changes of Patterns in Load Profile (CPLP) data of factories. CPLP consists of two stages. The first stage is to cluster the load profiles in each time window and use the clusters to model the power consumption patterns. We propose a new ensemble clustering method to cluster the load profiles in consecutive time windows. It uses a hierarchical binary k-means algorithm to generate component clusterings and a new objective function to ensemble them to produce the final clustering. The second stage is to track the changes of patterns along the time windows. We propose a new method to detect the change of clusters from one window to the next one by using the distribution models of two related clusters in two neighboring windows. By using this method, we can link the clusters in the sequence of time windows to track the patterns. Experiments on synthetic and real-world load profile data have shown that the proposed algorithm was able to track the changes of power consumption patterns of different factory groups and identify the period of significant change, which are very useful for the smart grid applications.