Fault Identification in Non-Stationary Systems Based on Sliding Mode Observers
Year: 2021
Type: article
Abstract: The paper is devoted to the problem of fault identification in technical systems described by non-stationary nonlineardynamicequationsunderunmatcheddisturbances.Tosolvetheproblem,slidingmodeobserversareused.The suggested ap-proachis based on the model of the original system of minimal dimension having different sensitivity to the faults and distur-bancesin contrast to the traditional approaches to sliding observer design which are based on the original system. Additionallyitis assumed that matrices describing such a model have the canonical form and are constant. The main purpose of usingsucha model is possibility to take into account the non-stationary feature of the systems. As a result, the model has stationarydynamicandnon-stationaryadditionalterm that allows to promote sliding mode design. Besides, the new approach to designslidingmodeobservers is suggested. The peculiarity of this approach is that it does not require that original systems should beminimumphaseand detectable. According to the traditional approaches stability of the observer is provided by minimum phaseanddetectabilityproperties.In our approach, stability of the observer is achieved due to the canonical form of the matricesdescribingthemodel.In addition, the matching condition is not necessary. This allows to extend a class of systems for whichslidingmodeobservers can be designed. Theoretical results are illustrated by practical example of electric servoactuator.
Language: English
Subfield: Control and Systems Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4200139792