商业建筑的测量冷冻水（CHW）、加热热水（HHW）和电力（ELE）公用设施消耗对建筑经理和调试工程师有兴趣，以获得节能机会。这三种类型的能源消耗通常取决于室外天气条件、一天中的时间、运行、占用和其他因素。从数据驱动建模的角度来看，确定峰值或基本消耗发生的时间对于监控建筑能耗和规划多建筑校园的集中工厂分布非常有价值。所提出的识别峰值和基本能耗的方法嵌入到非参数成对统计比较测试中。该测试比较了一天中的所有时间对，并根据测量的每小时历史数据，对建筑在每个公用设施中处于峰值或基本消耗水平的时间进行分类。除了一小时的平均小时消耗值最高或最低外，测试结果还提供了消耗量保持在统计相似水平的时间段。 当一栋建筑的高峰小时数很短，基准小时数很长（高度倾斜的分布）时，该方法具有鲁棒性，并且当建筑的每小时消耗水平没有显著变化时，该方法能够标记情况。引文:2018年冬季会议，伊利诺伊州芝加哥，会议论文

Measured chilled water (CHW), heating hot water (HHW), and electricity (ELE) utility consumption of commercial buildings are of interest to building managers and commissioning engineers for energy savings opportunities. These three types of energy consumption typically depend on outdoor weather conditions, time-of-day, operation, occupancy, and other factors. From the perspective of data-driven modelling, identifying the hours when peak or base consumption occurs is valuable for monitoring building energy consumption and planning central plant distribution for multi-building campuses. The proposed method for identifying peak and base energy consumption is embedded in a non-parametric pairwise statistical comparison test. The test compares all pairs of the hours of day and categorizes the hours at which the building is at a peak or base consumption level in each utility according to measured hourly historic data. Beyond one hour of day which appears to have the highest or lowest average hourly consumption value, the test results provide time periods during which consumption stays at a statistically similar level. This method shows robustness when a building has very short peak hours and very long base hours (highly skewed distribution), and is capable of flagging cases when the hourly consumption level of the building does not significantly vary.