Name: JANARIA CANDEIAS DE OLIVEIRA CARMINATI
Type: MSc dissertation
Publication date: 28/03/2018
Advisor:
Name |
Role |
|---|---|
| DANIEL JOSÉ CUSTODIO COURA | Co-advisor * |
| WANDERLEY CARDOSO CELESTE | Advisor * |
Examining board:
| Name |
Role |
|---|---|
| ANIBAL COTRINA ATENCIO | Internal Examiner * |
| DANIEL JOSÉ CUSTODIO COURA | Co advisor * |
| WANDERLEY CARDOSO CELESTE | Advisor * |
Summary: Non-intrusive electrical load monitoring systems (NILMs) have varied applications and enable better and more efficient demand side load management. This work aims to present a new approach for the extraction of characteristics that allow the identification of loads using the NILM approach. This new approach is here termed NILM active. In the identification through active NILM, an excitation signal is injected into the system, so that the system response to such excitation signal may have characteristics that are specific to each charge, or at least assist in composing a charge signature, along with other features extracted from other methods, in order to enable the accurate and robust identification of electric charges of a smart grid. This work is based on the study of reflectometry for the active identification of loads. The reflectometry is based on the comparative analysis performed in relation to an excitation signal inserted in the transmission line and its reflection. From the possible methods present in the reflectometry theory, the TDR (Time Domain Reflectometry) method is used, based on a generic model approach for complex structures with n branches, which can be subdivided into more groups simple, such as, for example, two loads and one excitation point, giving rise to a Y-format. The models used in the literature for the use of complex network TDR are used, from which the impedance electric charges in a Y-network, being applied in 4 senarios: lossless network; lossy network; and lossy using the lossless network method; and network with losses with distance variation. Among the proposed characteristics are the estimated loads, their respective physical locations in the electric line and the branch point of the electric line. These characteristics can be submitted to systems dedicated to the identification of cargo, in order to improve the performance of such systems, especially in relation to the identification of loads with a high degree of similarity. The method can estimate the load values, and for a lossy line it presents an error of 0.05% for a purely resistive load and 4.5% for a short-circuit load. With these results, it is possible to conclude that the TDR can be a feature extraction tool that allows the identification of loads by the NILM method. However, for this purpose, it is necessary to know the value of the propagation constant of the transmission line. The work contributes to the use of an active process in a NILM system, in order to aggregate more data to the set of characteristics that determine the electric signature of a load, which can enable a better success rate of identification systems when, for For example, there are loads with a high degree of similarity.
