Scholarlink

<p>Accurate energy consumption forecasts are crucial for better energy management and resource efficiency in residential and commercial buildings. This research improves forecasting accuracy by combining traditional predictors with advanced metaheuristic optimization techniques. We enhance Logistic Regression Classification (LRC) and Stochastic Gradient Descent (SGD) using Tasmanian Devil Optimization (TDO) and Chaos Game Optimization (CGO), evaluating metrics such as Accuracy, Precision, Recall, F1-Score, and Matthews Correlation Coefficient (MCC) across training and testing phases. Sensitivity analyses considering building size, operational schedules, and load history verify the robustness of results. On testing, the optimized LRC variants improved from 0.970 (MCC 0.955) to 0.980 with TDO and to 0.995 with CGO (MCC 0.993), establishing LRGG as the best performer. For SGD, the baseline model’s accuracy dropped to 0.900 (MCC 0.854) on unseen data, but optimization restored performance—SGTD and SGGG reached 0.958 and 0.955 accuracy with MCCs of 0.937 and 0.932. The CGO-optimized SGD achieved 95.7% accuracy with R² = 0.942. Overall, these findings demonstrate that combining TDO/CGO with LRC/SGD significantly improves accuracy and generalization, providing reliable forecasts that support renewable integration and dynamic energy strategies in smart building and grid management. The novelty of this work lies in the systematic coupling of optimizers and models, direct comparisons, robustness checks, and pathways for practical implementation.</p> Graphical abstract <p></p>

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

Improving energy management strategies in residential and commercial buildings through predictive models and optimization algorithms

  • 期刊论文

  • 发表日期：

    2026年2月16日

查看全文
  • Yanfen Gong,
  • Yaxiao Wang

摘要

Accurate energy consumption forecasts are crucial for better energy management and resource efficiency in residential and commercial buildings. This research improves forecasting accuracy by combining traditional predictors with advanced metaheuristic optimization techniques. We enhance Logistic Regression Classification (LRC) and Stochastic Gradient Descent (SGD) using Tasmanian Devil Optimization (TDO) and Chaos Game Optimization (CGO), evaluating metrics such as Accuracy, Precision, Recall, F1-Score, and Matthews Correlation Coefficient (MCC) across training and testing phases. Sensitivity analyses considering building size, operational schedules, and load history verify the robustness of results. On testing, the optimized LRC variants improved from 0.970 (MCC 0.955) to 0.980 with TDO and to 0.995 with CGO (MCC 0.993), establishing LRGG as the best performer. For SGD, the baseline model’s accuracy dropped to 0.900 (MCC 0.854) on unseen data, but optimization restored performance—SGTD and SGGG reached 0.958 and 0.955 accuracy with MCCs of 0.937 and 0.932. The CGO-optimized SGD achieved 95.7% accuracy with R² = 0.942. Overall, these findings demonstrate that combining TDO/CGO with LRC/SGD significantly improves accuracy and generalization, providing reliable forecasts that support renewable integration and dynamic energy strategies in smart building and grid management. The novelty of this work lies in the systematic coupling of optimizers and models, direct comparisons, robustness checks, and pathways for practical implementation.

Graphical abstract