An improved decision tree-based fault diagnosis method for practical variable refrigerant flow system using virtual sensor-based fault indicators

Review on Fault Detection and Diagnosis Feature Engineering in Building Heating, Ventilation, Air Conditioning and Refrigeration Systems

“…To eliminate this effect, PIs should be transformed into dimensionless forms -PI residuals -which denotes the deviation between the field measured PI and the regression model predicted PI based on input data from normal operation [140,[160][161][162][163]. Virtual sensors [51,52,[182][183][184] are also a good method for generating dimensionless PIs and have been widely applied in many HVAC&Rs including ASHPs [9,[187][188][189], chillers [71,[147][148][149][150] and VRFs [88,89,190]. These new FFs enrich the candidate FF pool.…”

“…The embedded method chooses FFs with larger weight coefficients contributing to higher model accuracy using feature-ranking techniques. Some tree-based algorithms [89,190,235,248] inherently take advantage of their own FS and data classification processes simultaneously. Yan et al [235] developed a decision tree (DT) based AHU FDD model.…”

Literature Review for Sensor Impact Evaluation and Verification Use Cases - Building Controls and Fault Detection and Diagnosis (FDD)

“…Lee and Yik [173] Sensor fault Energy cost impacts of sensor faults 19 Guo et al [174] Feature selection from available sensors Hybrid feature selection, VRF 20 Guo et al [175] Sensor fault, sensor data analysis/mining Senor fault detection, VRF 21 Shi et al [176] Sensor data analysis/mining VRF 22 Li et al [177] Virtual sensor Virtual sensor-based fault indicators for a VRF 23 Verhelst et al [178] Sensor fault Economic impact of persistent sensor and actuator faults 24 Yoon and Yu [179] Sensor calibration Virtual in-situ sensor calibration 25 Shi et al [180] Feature selection from available sensors VRF 26 Yu et al [181] Sensor calibration Indirect virtual calibration method for a supply air temperature in an RTU 27 Kim [182] Virtual sensor Review of virtual sensor, evaluation of sensors, development and assessment of alternative virtual sensors in an RTU 28 Katipamula et al [183] Additional senor, sensor layout/location, sensor fault…”

“…Li et al [177] proposed an improved decision tree-based fault diagnosis method for a practical VRF system. The proposed method was a three-stage method combining the decision tree model with virtual sensor-based fault indicators.…”

Review: Sensor Impact on Building Controls and Automatic Fault Detection and Diagnosis (AFDD)

“…The virtual sensor is also an important topic in FDD research [50,52,59]. Virtual sensing systems use the information available from other sensors and process parameters to calculate an estimate of the quantity of interest.…”