Abstract: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 in practical applications. One important and interesting question remains unanswered: is it possible to use intrinsically interpretable models to achieve accuracy comparable to that of black-box models? With an aim of answering this question, this study proposes an intrinsically interpretable machine learning-based method to forecast building energy loads. It creatively combines two intrinsically interpretable machine learning algorithms: clustering decision trees and adaptive multiple linear regression. Clustering decision trees aim to automatically identify various building operation conditions, allowing for the training of multiple models tailored to each condition. It can reduce the complexity of model training data, leading to higher accuracy. Adaptive multiple linear regression is an improved regression algorithm tailored to building energy load prediction. It can adaptively modify regression coefficients according to building operations, enhancing the non-linear fitting capability of multiple linear regression. The proposed method is evaluated utilizing the operational data from an office building. The results indicate that the proposed method exhibits comparable accuracy to both random forests and extreme gradient boosting. Furthermore, it shows significantly superior accuracy, with an average improvement of 10.2%, compared with some popular black-box algorithms such as artificial neural networks, support vector regression, and classification and regression trees. As for model interpretability, the proposed method reveals that historical cooling loads are the most crucial for predicting building cooling loads under most conditions. Additionally, outdoor air temperature has a significant contribution to building cooling load prediction during the daytime on weekdays in summer and transition seasons. In the future, it will be valuable to explore integrating the laws of physics into the proposed method to further enhance its interpretability.

A novel prediction model for integrated district energy system based on secondary decomposition and artificial rabbits optimization

Predicting daily heating energy consumption in residential buildings through integration of random forest model and meta-heuristic algorithms

A novel intelligent modeling and prediction of heat energy consumption in smart buildings

A Period-Based Neural Network Algorithm for Predicting Building Energy Consumption of District Heating

A novel hybrid model based on modal decomposition and error correction for building energy consumption prediction

Short-term building energy consumption prediction strategy based on modal decomposition and reconstruction algorithm

A novel combined model for heat load prediction in district heating systems

Forecasting District-Scale Energy Dynamics Through Integrating Building Network and Long Short-Term Memory Learning Algorithm

A novel Transformer-based network forecasting method for building cooling loads

Optimising Building Energy and Comfort Predictions with Intelligent Computational Model

An integrated building energy performance evaluation method: From parametric modeling to GA-NN based energy consumption prediction modeling

A hybrid deep learning-based method for short-term building energy load prediction combined with an interpretation process

Early energy performance analysis of smart buildings by consolidated artificial neural network paradigms

A hybrid machine learning approach for the load prediction in the sustainable transition of district heating networks

Intrinsically Interpretable Machine Learning-Based Building Energy Load Prediction Method with High Accuracy and Strong Interpretability

Urban building energy prediction at neighborhood scale

Novel approach to energy consumption estimation in smart homes: application of data mining and optimization techniques

Auto-tuning Method for Data-Driven Models in Building Energy Consumption Prediction: a Case of Cooling Load Prediction

Machine learning-based digital district heating/cooling with renewable integrations and advanced low-carbon transition

Proposing hybrid prediction approaches with the integration of machine learning models and metaheuristic algorithms to forecast the cooling and heating load of buildings

A novel improved model for building energy consumption prediction based on model integration