Fault prognosis of HVAC air handling unit and its components using hidden-semi Markov model and statistical process control

Air handling units are key sub-systems of heating, ventilation and air conditioning systems, which are used to condition air to satisfy human comfort requirements. Fault prognosis allows maintenance crews to identify the Remaining Useful Life (RUL) of a system, thus unexpected breakdowns are avoided, leading to a decrease in maintenance costs. To estimate RULs, a Hidden Semi-Markov Model (HSMM)-based method is proposed. To estimate states of HSMMs accurately, a revised scaled method is developed to guarantee that state estimates do not approximate to infinity. Additionally, a new discrete statistical process control method is developed to filter out false state estimates of HSMMs. To estimate RULs of components and systems accurately and effectively, a backward recursive method is developed to integrate HSMMs' parameters of time-duration distributions for multiple failure modes to generate those of components and systems directly, thus low computational effort is achieved. Experimental results illustrate that the RULs of components/systems can be predicted by our method accurately in an efficient way. (c) 2021 Elsevier B.V. All rights reserved.

Automated summary

Air handling units play a crucial role in HVAC systems, and fault prognosis using Hidden Semi-Markov Models can accurately estimate Remaining Useful Life to prevent unexpected breakdowns and reduce maintenance costs. This study introduces a revised scaled method and a new discrete statistical process control method to improve state estimation accuracy, along with a backward recursive method for efficient RUL estimation. The experimental results demonstrate the effectiveness and accuracy of the proposed approach in predicting RULs of components and systems.