Publication:
LOAD MONITORING AND ACTIVITY RECOGNITION IN SMART HOMES

Simple item page
dc.contributor.advisorABDELHAMID, MOHAMED
dc.contributor.authorFRANCO TROYA, PATRICIA
dc.contributor.departmentUniversidad Técnica Federico Santa María. Departamento de Electrónicaes_CL
dc.contributor.otherMARTÍNEZ, . JOSÉ M.
dc.contributor.otherCREIXELL, WERNER
dc.contributor.otherBEGHELLI, ALEJANDRA
dc.coverage.spatialCasa Central Valparaísoes_CL
dc.date.accessioned2021-10-08T19:45:29Z
dc.date.available2021-10-08T19:45:29Z
dc.date.issued2021-06
dc.description.abstractAppliance load monitoring in smart homes has been gaining importance due to its sig nificant advantages in achieving an energy efficient smart grid. The methods to manage such processes can be classified into hardware-based methods, including intrusive load monitoring (ILM) and software-based methods referring to non-intrusive load monitoring (NILM). ILM is based on low-end meter devices attached to home appliances in opposition to NILM techniques, in which only a single point of sensing is needed. Although ILM solutions are relatively expensive, they provide higher efficiency and reliability rather than NILMs do. Moreover, future solutions are expected to be hybrid, combining the benefits of NILM along with individual power measurement by smart plugs and smart appliances. This thesis proposes a novel ILM approach for load monitoring that aims to develop an activity recognition system based on an IoT architecture. The proposed IoT architecture consists of an appliances layer, a perception layer, a communication network layer, a middleware layer, and an application layer. The application layer consists of an appliance recognition module and activities of daily living (ADL) classification algorithm. The main function of the appliance recognition module is to label sensor data and to allow the implementation of different home applications. Three different classifier models are tested using real data from the UK-DALE dataset: feed-forward neural network (FFNN), long short-term memory (LSTM), and support vector machine (SVM). The developed ADL algorithm maps each ADL to a set of criteria depending on the appliance used. The features are extracted according to the consumption in Watt-hours and the times where appliances are switched on. In the FFNN and the LSTM networks, the accuracy is above 0.9 while being around 0.8 for the SVM network. Other experiments are performed to evaluate the classifier model using a test set. A sensitivity analysis is also carried out to study the impact of the group size on the classifier accuracy. Once results were obtained, the proposed ADL classification system was enhanced in two frameworks: a training framework and an inference framework. This is to allow a practical implementation of the system. In this regard, several modifications were made in the appliance recognition module, including the use of new data, and therefore new appliances: an electric vehicle, an oven and a microwave, from the Dataport dataset. The frameworks include graphical interfaces that significantly facilitate its use. The dataset configuration, pre-processing and classification parameters can be easily selected and modified. In the feature extraction, inside a sliding window, statistical features of the power samples are computed. In this way, the same pre-processing can be applied in the two different datasets. A feature importance analysis can also be performed to analyze the contribution of the selected features in the models predictions. With this implementation, the real-time operation is directly related with the size of the window usedes_CL
dc.description.degreeMAGISTER EN CIENCIAS DE LA INGENIERIA ELECTRONICAes_CL
dc.description.programDEPARTAMENTO DE ELECTRÓNICA. MAGÍSTER EN CIENCIAS DE LA INGENIERÍA ELECTRÓNICA (MS)es_CL
dc.format.extent114 H.es_CL
dc.identifier.barcode26944021KUTFSMes_CL
dc.identifier.urihttps://hdl.handle.net/11673/50525
dc.subjectAPPLIANCE RECOGNITIONes_CL
dc.subjectACTIVITY CLASSIFICTIONes_CL
dc.subjectFRAMEWORKes_CL
dc.subjectINTERNET OF THINGSes_CL
dc.subjectINTRUSIVE LOAD MONITORINGes_CL
dc.subjectIOT PLATFORM, SMART GRIDes_CL
dc.subjectSMART HOMESes_CL
dc.titleLOAD MONITORING AND ACTIVITY RECOGNITION IN SMART HOMESes_CL
dc.typeTesis de Postgrado
dspace.entity.typePublication
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
m26944021-k.pdf
Size:
3.73 MB
Format:
Adobe Portable Document Format
Description:
Download
Collections
Arq_paso