Scholarlink

This paper presents an innovative approach to hybrid parameter tuning using Gaussian Process Regression (GPR) to forecast energy consumption patterns in smart homes. This approach combines Bayesian optimization with domain-specific feature engineering to improve prediction accuracy on data derived from multiple heterogeneous smart home data sources such as weather variables and appliance-level consumption logs. The resulting probabilistic forecasts facilitate anticipatory demand-response strategies, reduce peak loads, and optimize renewable energy usage. This paper presents the details of the hybrid parameter tuning method using GPR for a smart home dataset, as case study. We use a subset of parameters from the Rye microgrid dataset to develop supervised machine learning (ML) models that achieve high predictive accuracy for renewable energy production, consumption, and related behavioral dynamics of smart homes.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Smart Home Energy Prediction Using Bayesian Hyper-Parameter Optimization of Gaussian Process Regression

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Ahmed Alzimami,
  • Saptarshi Das,
  • Markus Mueller

摘要

This paper presents an innovative approach to hybrid parameter tuning using Gaussian Process Regression (GPR) to forecast energy consumption patterns in smart homes. This approach combines Bayesian optimization with domain-specific feature engineering to improve prediction accuracy on data derived from multiple heterogeneous smart home data sources such as weather variables and appliance-level consumption logs. The resulting probabilistic forecasts facilitate anticipatory demand-response strategies, reduce peak loads, and optimize renewable energy usage. This paper presents the details of the hybrid parameter tuning method using GPR for a smart home dataset, as case study. We use a subset of parameters from the Rye microgrid dataset to develop supervised machine learning (ML) models that achieve high predictive accuracy for renewable energy production, consumption, and related behavioral dynamics of smart homes.