The study conducts an in-depth analysis of time-series historical data generated by buildings using machine learning techniques. A general model identification method was developed through an algorithm that optimizes competition based on black-box models. The final identification model was determined by optimizing competition among three machine learning methods: polynomial regression

artificial neural networks

and extreme gradient boosting. The study focuses on a near-zero energy office building in Beijing. Based on historical building data and TRNSYS heating system simulation data

load prediction and equipment energy consumption models were established using the developed model identification method. During deployment

the predicted R2 value and total energy consumption error were 0.87 and 5.18%

respectively. Results indicate that the prediction models established through this method possess high accuracy

providing a reliable basis for subsequent system energy consumption optimization.