Research on building energy consumption anomaly detection based on data mining technology

To enhance the accuracy of anomaly detection in building energy consumption management, this study develops a hybrid detection model that integrates the Isolated Forest and Self-Encoder within a data mining framework. By analyzing the time periodicity and fluctuation characteristics of building energy consumption, key indicators are extracted and multidimensional input features are designed, and unsupervised learning algorithms are combined for modeling and training. The experiments are carried out in multiple time periods and scenarios, and the results show that the model is better than the traditional methods in terms of accuracy, F1 value and detection delay, and has strong practicality and stability, providing an effective technical path for intelligent operation and maintenance of buildings.