It was found that the proposed methodology significantlyĪn improved principal component analysis with jointĪngle analysis (JAA) was also used to detect and diagnoseįixed and drifting biases of sensors in variable air volume The proposed methodology with the normal PCA method, and (2012) but they employed a self adaptive process toĪutomatically remove error sampled in the original data set
Same kind of methodology was also applied by HuĮt al.
Ship between major measured variables in the centrifugalĬhiller. The PCA methodology captured the relation. Methodology accurately estimated most of the introduced Sensor faults were successfully detected and PCA-based To the measurements of different sensors and artificiallyĬorrupting their readings. They developed two PCA models: oneĬoncerning energy balance and the second concerningĮnergy performance. Training matrixes, retaining loading vectors, and determina. Using three steps: decomposing of the covariance matrix of Q-statistics to detect and Q-contribution plot to diagnose the Wang and Cui (2005) applied an online strategy to detect,ĭiagnose and validate sensor faults in centrifugal chillersīy using principal component analysis. Statistics new new newˆSPE ( )Q δ= - = - £x x I PP x (10)ĥ.1.1 Principal component analysis application The Q statistics can be represented by the following equation Statistics (the squared sum of the residual) or squared pre-ĭiction error (SPE) is used as an index of faulty conditions. Most of the fault detection and diagnosis application, Q Of ˆ( )new newx x and residual (e) (Wang and Xiao 2004b). New sample xnew can be divided into two parts i.e. If only k number of principal components are used, anĮstimation of x in Eq. Represent the directions of the most variance of a system. Retained eigenvectors are the one that are associated with The matrix U are the eigenvectors of the covariance matrix Is used to assign weights to each variable. (9) U T is the loading matrix, ( )m m´ÎU U R, which Original variables ( )mÎx x R as given by Eq. Principal components, ( )mÎy y R ,Īre constructed as a weighted linear combination of the PCA method focuses on analysing principal components Therefore, instead of analysing all involved variables, the Represented by a smaller number of components (principalĬomponents) because of the redundancy of the variables. Optimal in terms of capturing the variability in the data PCA producesĪ lower dimensional representation that preserves theĬorrelation structure between the process variables and is Used as a dimensional reduction technique. Principal component analysis (PCA) is a multivariateĪnalysis method (Jackson 2005 Jolliffe 2005), which is also Management, IBPSA: International Building Performance Simulation Association.Īhmad et al.
ANN LM BP TRNSYS Supply air temperature sensorĪ Control and/or fault detection and diagnosis (FDD) method.ī BP: back propagation, LM BP: Levenberg–Marquart back propagation, GD–BP: gradient decent back propagation, CG–BP: conjugate gradient back propagation, BFGS: Broyden–Fletcher–Goldfarb–Shanno.Ĭ EC: energy consumption, TC: thermal comfort, VC: visual comfort, OP: occupant preference, IAQ: indoor air quality.ĭ SAR: special administrative region, HK: Hong Kong SAR, China, SG: Singapore, CA: Canada, UK: United Kingdom, USA: United States of America, GR: Greece, KR: Republic of Korea, FR: France, CN: Mainland China, KW: Kuwait.Į ENB: Energy and Buildings, ATE: Applied Thermal Engineering, BAE: Building and Environment, NeuNet: Neural Networks, ASHRAE Trans.: ASHRAE Transactions, ApEn: Applied Energy, ECM: Energy Conversion and