An Automated Cluster-Based Approach for Asset Rescheduling in Building Communities

The imbalance between load demand and power generation and high peak loads in the smart grid become a big concern for utility companies. Increasing the flexibility of the grid by using innovative demand response programs can improve the quality of the power grid and reduce the peak demand. In this paper, we propose a cluster-based approach for building assets rescheduling in a smart building community. This method takes advantage of interactions among the buildings and aims to minimize the total energy cost as well as the peak-to-average ratio (PAR). In the proposed model, several load actions for each household appliance are generated based on the class of the assets and consumers' preferences. A clustering approach is then presented to group similar load actions for assets via X-means clustering. In order to do a more accurate clustering, time-domain (TD) load actions are transformed to frequency-domain (FD) using the Fast Fourier Transform (FFT) method. The results of the clustering step are used as inputs of the optimization problem. Finally, a multi-objective mixed-integer linear programming (MOMILP) problem is proposed to reschedule building assets optimally. The results show that the energy cost for a small community could reduce by around 25% while the peak demand could reduce by around 7%.