| Developing efficient techniques for energy optimization and conservation requires a thorough understanding of the patterns of energy usage among different home appliances. This study looks into the energy usage patterns of several household appliances and assesses how well machine learning methods classify these patterns. Appliances are categorized into three groups: constantly on devices, program-based on-demand devices, and non-program-based on-demand devices. Key statistical features such as periodograms, mean, and standard deviation are extracted for machine learning classification, employing DenseNet 1D, XGBoost, LightGBM, SVM, KNN, and Random Forest. The findings show DenseNet and LightGBM performing exceptionally well, with nearly 98% accuracy in classifying constantly-on and program-based devices, indicating their potential in optimizing energy usage. |
*** Title, author list and abstract as seen in the Camera-Ready version of the paper that was provided to Conference Committee. Small changes that may have occurred during processing by Springer may not appear in this window.