Model-Based Condition Monitoring

Abstract

Abstract: One of the key challenges to model condition monitoring is; how tosuitable match the level of detail of the model description of thediagnostic requirements. To do so, a counter-current heat exchanger ofthe tire factory was investigated. Moreover, a two-stage grey-boxmodelling was deployed. The modelling is deployed in a discrete statespace, relying on a combination of prior knowledge, along with therecorded data. The nonlinear model structure of heat exchanger isderived considering the use of the prior knowledge; thereby, it isconverted into linearized form. The discretization of the linearizedmodel is applied. The optimization procedure based on recorded data isadopted to estimate the unknown physical parameters in state spacesystem matrices. The well-known N4SID subspace identification method isused to compare the obtained results. Show more