Project description
The project consists of applying artificial intelligence technologies to energy consumption data in both the domestic and industrial sectors with the aim of finding consumption patterns and creating models that allow for improving efficiency and reducing energy consumption.
In a first phase, experimental data will be obtained and processed so that they can be subsequently analyzed. The first contributions are expected to be made in the field of signal processing and data pre-processing. Later, work will be done on “feature selection” algorithms to find better predictors of energy consumption and generation. Based on pattern selection, the different techniques for data modeling will be introduced, both in “machine learning” and “deep learning” (neural networks, convolutional neural networks, recurrent neural networks), regressions.
For the analysis of the operation and the diagnosis, techniques such as anomaly or novelty algorithms will be evaluated. Likewise, regression models will also be used as an indicator of malfunction.
Data clustering will be evaluated as a pre-processing technique and ensemble learning to obtain indicators that can be easily interpreted by non-expert users.
The objective is to obtain consumption patterns and disaggregation of consumption and electricity generation based on the application of these techniques, in order to detect the energy consumption habits of consumers and to be able to control and promote effective flexibility strategies.
The direct application of this entire study is energy savings. The objective will be to have aggregated and disaggregated models of electrical consumption, renewable generation and available state of charge in batteries that allow the optimization of flexibility services. Work will be based on the prediction of load availability to ultimately optimize flexibility services.
