Gray Box Time Variant Clogging behaviour and Pressure Drop Prediction of the Air Filter in the HVAC System

Identification and prediction of clogging behavior in heating, ventilation, air conditioning (HVAC) filters is crucial to avoid issues such as system overheating, energy waste, lower indoor quality, etc. Researchers are focusing more on the particle loading characteristics a filter medium laboratory environment under steady-state conditions, fixed concentrations, area porosity, dust feed volumetric flow rate. However, recent research still shows uncertainties modeling well implementation problems constructing HVAC test bench equipment. In addition, subjects non-uniform deposition depreciation condition various type mechanical fibrous, fabric, granular, membrane or electrostatic which typically used systems perform some assumptions need research. The studies become even difficult acquiring large number time-varying noisy signals. Another approach among data-driven knowing that Building Automation System (BAS) not equipped with appropriate sensor measuring clogging, it needed drive mathematical model from pressure drop signal. This paper bridges gap between particle-size study black box has received much attention authors. proposed method assumes result two functions; f(t), represents dynamics and, g(t), refers remained terms. exponential polynomial second order functions express behavior. software package based Particle Swarm Optimization Artificial Bee Colony (PSOABC) algorithm, developed implemented estimate coefficients smallest RMSE, high coefficient correlation acceptable tracking. Five Air Handling Unit (AHUs) selected for practical verification results show applied can successfully predict behaviour filters.