Rare Event Analysis of High Dimensional Building Operational Data Using Data Mining Techniques

摘要: Today’s building automation systems (BASs) are becoming increasingly complex. A typical BAS usually stores hundreds of sensor measurements and control signals at each time step, which produces massive high dimensional data sets. Traditional analysis methods for only focus on a small subset the data, resulting in huge information loss. Data mining techniques more effective knowledge extraction data. This study develops holistic methodology analyzing using advanced techniques, with aim identifying rare events operation. Rare event helps to identify atypical operating patterns, detect diagnose faults, eventually improve operational performance. Two main challenges exist performing i.e. dimensionality complexity system The former results that conventional analytics, such as distance-based measures, lose their effectiveness, later negatively influences robustness reliability identification events. proposed method is specially designed tackle these by integrating power techniques. It consists four steps, i.e., preparation, detection, diagnosis, post-mining. adopted analyze tallest Hong Kong. successfully detected diagnosed, providing clues enhance