OSTI OAI (U.S. Department of Energy Office of Scientific and Technical Information) · 1996 · 89 citations · 0 references
ComputingComputer ScienceIndustrial EngineeringAir ConditioningArtificial IntelligenceMachine LearningAir Handling UnitsDiagnosisDiagnosticsFault DetectionFault DiagnosisNeural NetworksAir Handling Unit DiagnosticsBackpropagationFault Detection Using Machine LearningMachine Learning-based Fault DiagnosisNeural Network DiagnosticsNeural Network-based DiagnosisResidual-based Fault DetectionSystem Variables
The objective of this study is to describe the application of artificial neural networks to the problem of fault diagnosis in an air-handling unit. Initially, residuals of system variables that can be used to quantify the dominant symptoms of fault modes of operation are selected. Idealized steady-state patterns of the residuals are then defined for each fault mode of operation. The steady-state relationship between the dominant symptoms and the faults is learned by an artificial neural network using the backpropagation algorithm. The trained neural network is applied to experimental data for various faults and successfully identifies each fault.