Abstract
Recent technological advancements have facilitated the evolution of traditional distribution grids to smart grids. In a smart grid scenario, flexible devices are expected to aid the system in balancing the electric power in a technically and economically efficient way. To achieve this, the flexible devices’ consumption data are theoretically recorded, elaborated and their upcoming flexibility is bid to flexibility markets. However, there are many cases where explicit flexible device consumption data are absent. This paper presents a way to circumvent this problem and extract the potentially flexible load of a flexible device, namely a Heat Pump (HP), out of the aggregated energy consumption of a house. The main idea for accomplishing this, is a comparison of the flexible consumer with electrically similar non-flexible consumers. The methodology is based on machine learning techniques, probability theory and statistics. After presenting this methodology, the general trend of the HP consumption is estimated and an hour-ahead forecast is conducted by employing Seasonal Autoregressive Integrated Moving Average modeling. In this manner, the flexible consumption is predicted, establishing the basis for bidding flexibility in intra-day markets even in the absence of explicit device measurements.
Original language | English |
---|---|
Journal | I E E E Transactions on Smart Grid |
Volume | 6 |
Issue number | 4 |
Pages (from-to) | 1852 - 1864 |
Number of pages | 13 |
ISSN | 1949-3053 |
DOIs | |
Publication status | Published - Jul 2015 |
Keywords
- Heat Pump, Estimation, Prediction, Flexibility, Non-intrusive Load Identification