Abstract: In a method for failure, detection, operational data from a plurality of components in a system are received. A bank of submodels is created based on the operational data. The bank of submodels corresponds to a normal mode and one or more faulty modes of the system and is valid in different operating regimes of the system. Each of the banks of submodels has a respective weight (validity) and a respective suboutput. An output of the system is a weighted sum of the suboutputs of the submodels. The operational data is therefore processed to generate a validity profile through a constrained Kalman Filter (KCF) based multimodel fault detection and diagnosis (FDD). Subsequently, the validity profile is output. The validity profile is indicative of an operation state of the system at a given time, and the operation state includes a normal state and a fault state.

Abstract: In a method for failure detection, operational data from a plurality of components in a system are received. A bank of submodels is created based on the operational data. The bank of submodels corresponds to a normal mode and one or more faulty modes of the system and is valid in different operating regimes of the system. Each of the banks of submodels has a respective weight (validity) and a respective suboutput. An output of the system is a weighted sum of the suboutputs of the submodels. The operational data is therefore processed to generate a validity profile through a constrained Kalman Filter (KCF) based multimodel fault detection and diagnosis (FDD). Subsequently, the validity profile is output. The validity profile is indicative of an operation state of the system at a given time, and the operation state includes a normal state and a fault state.