Outlier detection via multiclass deep autoencoding Gaussian mixture model for building chiller diagnosis

Deep learning in fault detection and diagnosis of building HVAC systems: A systematic review with meta analysis

“…Among these 47 articles, if one article does not report the confusion matrix or its dataset is not accessible, this article is included for qualitative analysis but excluded for meta analysis. As a result, 6 studies [23] [24] [25] [26] [27] [28], are identified as eligible for meta analysis.…”

“…The American society of heating, refrigerating and air-conditioning engineers research report 1043 (ASHRAE RP-1043) dataset [71] is used in all six studies eligible for meta analysis, which included seven types of faults, i.e., condenser fouling, excess oil, refrigerant leak, refrigerant overcharge, reduced condenser water flow, reduced evaporator water flow, non-condensable gas in refrigerant of four severity levels, among which, confusion matrix of all severity levels are reported in [20]. The severity level of the confusion matrix used in [21], [22] is unknown, the confusion matrix of severity level one is reported in [23], and that of severity level two is reported in [24]. Although the confusion matrix of all four levels is reported in [25], the severity level four confusion matrix is not correctly constructed due to one missing category.…”

“…RNNs are the most commonly used neural network (NN) architecture for time series data modelling in general. The feedback loops of the recurrent cells inherently capture the temporal patterns of the sequences [23]. Owing to this property, RNNs are effective in modelling highly time-dependent and strongly coupled events, and such characteristics are common within an HVAC system.…”

“…The second type involves combining a deep AE with 2D CNN [116], it takes advantage of using deep AE to generate high quality multi-dimensional residual that is the thermodynamic deviation and critical feature for HVAC FDD, and by reshaping the residual data into matrix, it can be effectively processed by 2D CNN. The third type ensembles multiple DL building blocks, e.g., ensemble deep MLPs [23] and 1D CNNs [49]. Because an ensemble model is more robust to variance, it can be used to remove noise data [23] before FDD.…”

Dynamic Segmentation of Sensor Events for Real-Time Human Activity Recognition in a Smart Home Context

“…In this work, we are interested in dealing with ambiguous outliers’ detection in both training and testing data and in the insufficiency of labeled data. Outliers could be detected using a multiclass deep auto-encoding Gaussian mixture model, for example [ 28 ]. This is a set of individual unsupervised Gaussian mixture models that helps the deep auto-encoding model to detect ambiguous outliers in both training and testing data.…”