Building Energy Doctors: An SPC and Kalman Filter-Based Method for System-Level Fault Detection in HVAC Systems

Fault detection and diagnosis for building cooling system with a tree-structured learning method

“…Principal Component Analysis (PCA) [16,17,18], Statistical Process Control (SPC) [19,20,21], Multivariate Regression Models [22], Bayes Classifier [23,24,25], Neural Networks (NN) [26,27,28], Fisher Discriminant Analysis 40 (FDA) [29], Gaussion Mixture Model [30], Support Vector Data Description (SVDD) [31,32], and Support Vector Machines (SVM) [33, 34,35,36,37]. A c c e p t e d M a n u s c r i p t [13,38,39].…”

Fault detection of cooling coils based on unscented Kalman filters and statistical process control

“…By selecting parameters which are related to faults as fault indicators, physical models can be used for fault detection [6]- [7]. These papers were reviewed in our recent paper [3].…”

“…Therefore, four faults 1) blocking of tubes; 2) deposits in tubes; 3) damage of fins; and 4) dust accumulation on the outside surface area of fins are considered in this paper. In our recent work, a method of integrating Kalman filter and Statistical Process Control (SPC) based on a gray-box model is used to detect faults of HVAC devices [3]. The key idea of the method is to view the model parameters as slow-varying states.…”

“…The gray-box model, therefore, is used for fault detection in this paper. Using the method developed in [3] for fault detection of cooling coils is challenging because 1) no gray-box models of cooling coils were developed and validated in the literature; and 2) the gray-box model developed is nonlinear thus the standard Kalman filter method is not applicable.…”

“…Therefore, the method integrating Kalman filtering and Statistical Process Control (SPC) for addressing the linear problem is not applicable [3]. To address this problem, in Section 4, the method is extended to the nonlinear case by using UKF to estimate fault indicators and provide SPC control limits for fault detection.…”