Detecting Anomalous Load Data

An automatic detector for a particular pattern of anomolous load research data, called a quiet period, is described. Quiet periods exhibit a reduction in electricity usage due to a change in the underlying load-causing processes, most likely the absence of household members. The detector uses pattern recognition methods to identify and flag quiet period data. Examples of quiet periods are presented and an annual distribution of quiet periods for a set of 298 homes is developed. The quiet period detector allows refinement of the standard practice of forecasting holiday periods.