MORE | Management of Real-time Energy Data - Deliverable: "D2.1 Data Models – Initial Version"
This deliverable presents how modelscan be used to compress large amounts of time series data efficiently. This is needed to be able to store and analysethe huge amounts of sensor data available from modern wind turbines and solar panels. The currently available MORE datafrom ENGIEis then investigatedin details. It is found thatmany parts of the time series can be modelled by a quadratic function.Based on this a new model type is proposedto be added to the ModelarDB model-based Time Series Management System (TSMS). Further, it is found that many of the time series can be derived from other timeseries. Based on this it is proposed how to avoid physical storage of time series that can be derived from othertime series.The ENGIE dataset is also compressed by use of models in ModelarDB and it is found that even with lossless compression, the required storage is reduced by 19% and with an error bound of 10%, the reduction is bigger than 76%, compared to the original data size.
31 May 2021