Measurement Integrity Modeling for Anomaly-Aware COP Estimation in HVAC Telemetry Systems

Abstract

Reliable evaluation of heating, ventilation, and air conditioning (HVAC) energy efficiency depends on the integrity of the measurement streams that are used to compute performance indicators. In long-duration building telemetry, measurement irregularities introduced by sensing, communication, and logging, including dropouts, counter resets, frozen readings, persistent constant-value segments, and nonphysical transients, propagate directly into derived quantities such as the coefficient of performance (COP) and can yield operationally inconsistent or distorted trends. This work develops a measurement-centered telemetry monitoring method on a full-scale 20 tons of refrigeration (20 TR) chiller testbed at IIT Jammu using synchronized thermal, electrical, and environmental measurements. Rule-based event detectors are constructed from physical operating constraints to identify persistent measurement anomalies and to localize event onsets, and the rule outputs are combined with unsupervised multivariate detectors through a consensus fusion strategy. The fused outputs form an anomaly-screening mask that is coupled to performance computation, producing an anomaly-aware efficiency time series based on intervals retained after anomaly screening. The results during extended operation show improved statistical stability, reduced nonphysical COP variability and improved reliability of performance comparison across cooling configurations, indicating the importance of explicit measurement-data screening before performance evaluation in large-capacity HVAC systems.