Intrinsically interpretable machine learning-based building energy load prediction method with high accuracy and strong interpretability

Energy and Built Environment, Journal Year: 2024, Volume and Issue: unknown

Published: Aug. 1, 2024

Black-box models have demonstrated remarkable accuracy in forecasting building energy loads. However, they usually lack interpretability and do not incorporate domain knowledge, making it difficult for users to trust their predictions practical applications. One important interesting question remains unanswered: is possible use intrinsically interpretable achieve comparable that of black-box models? With an aim answering this question, study proposes machine learning-based method forecast It creatively combines two learning algorithms: clustering decision trees adaptive multiple linear regression. Clustering automatically identify various operation conditions, allowing the training tailored each condition. can reduce complexity model data, leading higher accuracy. Adaptive regression improved algorithm load prediction. adaptively modify coefficients according operations, enhancing non-linear fitting capability The proposed evaluated utilizing operational data from office building. results indicate exhibits both random forests extreme gradient boosting. Furthermore, shows significantly superior accuracy, with average improvement 10.2 %, compared some popular algorithms such as artificial neural networks, support vector regression, classification trees. As interpretability, reveals historical cooling loads are most crucial predicting under conditions. Additionally, outdoor air temperature has a significant contribution prediction during daytime on weekdays summer transition seasons. In future, will be valuable explore integrating laws physics into further enhance its interpretability.

Language: Английский

---

Toward scalable prediction of indoor thermal dynamics: Neural-network-implanted state-space (NNiSS) model

Energy and Buildings, Journal Year: 2025, Volume and Issue: unknown, P. 115359 - 115359

Published: Jan. 1, 2025

Language: Английский

---

Physical model and multiple moth-flame optimization fusion temperature field prediction in large-space building fires

International Journal of Thermal Sciences, Journal Year: 2025, Volume and Issue: 214, P. 109892 - 109892

Published: March 25, 2025

Language: Английский

---

Off-grid PV systems modelling and optimisation for rural communities-leveraging understandability and interpretability of modelling tools

Energy, Journal Year: 2025, Volume and Issue: unknown, P. 135948 - 135948

Published: April 1, 2025

Language: Английский

---