| Summary: | 碩士 === 國立中興大學 === 資訊科學與工程學系 === 104 === Recently, there are tremendous researches about Non-Intrusive Load Monitoring, and the vast majority of them used costly equipment in the market, such as watts up, to integrate self-established analysis system. However, the difficulty of it is that, installation is often not been set up on the power meter. In addition, the analysis of data collection is complicated because it needs to process several stages.
In this paper, we developed a hardware system which is cost-effective, easy to install, simple to collect data and can capture various features. It is more practical and feasible for typical families as well as businesses, and it can promote green energy information industry development as well. Furthermore, because the electricity information can be collected easily, advanced energy analysis and tracking carbon footprint can be realized by algorithms.
In terms of hardware contribution, we use Arduino (MCU), which is more approachable, equipped with ADI''s ADE7878 voltage electric current detection(ASIC), to integrate with SD Card and Wi-Fi module, and the sensed data can be stored in the memory card, or be uploaded to the cloud server(MySQL database).
In the back-end analysis systems, we use SVM classification algorithms to identify status of appliances (on or off). The predictive model can be trained using a small dataset. Besides, this model can be re-trained by the user’s feedback when they confirm the result of appliance recognition through user interface we provide, consequently, this will increase the recognition accuracy.
|
|---|
