Bilinear model-based diagnosis of lock-in-place failures of variable-air-volume HVAC systems of multizone buildings

Abstract

Centralized heating, ventilation, and air-conditioning (HVAC) systems equipped with variable-air-volume (VAV) terminal boxes are widely used for energy-efficient control in multizone buildings. Any occurrence of faults in VAV boxes drastically hampers the control performance. In this paper, we present a novel bilinear observer-based fault diagnosis (FD) algorithm for detection and accurate estimation of the magnitude of the lock-in-place failure of VAV dampers, which plays a significant role in the supervisory health-aware control strategy. The key contribution in the proposed FD algorithm is that instead of using specific linearized models, the algorithm is inherited from the true bilinear model structure of the building thermal dynamics. As an immediate consequence, the algorithm is applicable over a wide range of unpredictable operating conditions, namely, weather dynamics, outside air temperature, zone occupancy profile. Moreover, a practical cost-effective feature of the derived underlying bilinear observer is the estimation of wall temperatures, which are generally essential but unmeasured state variables with regards to climate control in buildings. The effectiveness of the proposed algorithm demonstrating the successful detection and estimation of VAV multiple damper faults are validated through exhaustive simulations with SIMBAD (SIMulator of Building And Devices).