Search

In this paper, we consider the task of clustering a set of individual time series while modeling each cluster, that is, model-based time series clustering. The task requires a parametric model with sufficient flexibility to describe the dynamics in various time series. To address this problem, we propose a novel model-based time series clustering method with mixtures of linear Gaussian state space models, which have high flexibility. The proposed method uses a new expectation-maximization algorithm for the mixture model to estimate the model parameters, and determines the number of clusters using the Bayesian information criterion. Experiments on a simulated dataset demonstrate the effectiveness of the method in clustering, parameter estimation, and model selection. The method is applied to real datasets commonly used to evaluate time series clustering methods. Results showed that the proposed method produces clustering results that are as accurate or more accurate than those obtained using previous methods.

The unit commitment problem (UCP) for an electric power system is used to determine the schedules of power units that minimize the total production cost over a planning horizon while satisfying the load demand, spinning reserve, and operating constraints of individual units. When the number of units is large and the planning horizon is long, the UCP is a large-scale problem, for which an exact optimal solution is difficult to obtain within a reasonable computation time. The Lagrangian relaxation (LR) method is known to be useful for large-scale UCPs. The LR method first solves the dual problem of the UCP, and then constructs a feasible solution from the dual solution by using some heuristics. In this paper, we propose new local search (LS) methods for improving the feasible solution obtained by the LR method. We define the neighborhood of the local search as the feasible set in which the on-off states of all but one or two units are fixed. The neighborhood search can then be executed by solving the one unit or two UCPs, which are efficiently solved by dynamic programming if no ramp-rate limit constraint exists. Numerical results show that the proposed LS methods can find feasible schedules for which the costs are lower than those obtained by the existing methods. The applicability of the proposed methods to long-term UCPs is also demonstrated.

While many studies have estimated the energy saving potential of office equipment power management, there is no recent study about the energy saving potential of shortening the power management delay time. In this paper, we estimated the energy saving potential of the complete saturation of power management, and also the additional energy saving potential of shortening the power management delay time for PCs, displays, copiers, and laser printers in Japanese offices. To obtain key data, we conducted the surveys and field measurements of usage in business hours and the turn-off rate at night. We found that office equipment power management can save as much as 3.5 TWh per year, which is nearly equal to 2% of commercial electricity consumption in Japan. We also found that the energy use of office equipment in Japanese offices with no use of power management is even lower than that in US offices with maximum use of power management. This is primarily because the portable computer is much more popular in Japan, and also because the manual-off rate at night is more than twice that in the US.