Fault Detection Of Electromechanical Actuation System Based On Unknown Input Observer And SVM

With the development trend of multi-electric aircraft,the Electro-Mechanical Actuation System(EMA)is gradually being widely used.Compared with the traditional hydraulic actuation system,a large number of hydraulic circuits are eliminated,which not only reduces maintenance costs,but also greatly reduces the weight of the aircraft.However,due to its new scope,small application range and short application time,EMA lacks fault detection experience and accumulation of fault data,especially due to its unique mechatronic design that introduces new failure modes and fault performance similarity.A series of potential problems,hindered the massive use of EMA on civil aircraft.In order to further explore the above problems,First establishes the EMA mathematical model.On this basis,the Unknown Input Observers(UIOs)are derived.Through the unknown input observer,the interference is completely decoupled from the residuals,reducing the external and improves the robustness of fault detection.At the same time,considering the high false alarm rate and missed detection rate caused by fixed threshold,it is proposed to use adaptive threshold instead of fixed threshold to detect fault.In order to further locate faults,especially for different fault modes under the same fault performance,this paper proposes a fault detection of electromechanical actuation system based on Principal Component Analysis(PCA)and Support Vector Machine(SVM).method.Firstly,the different failure modes under the same fault performance are analyzed and the fault data is extracted,then,the data is processed by the PCA to form the training database and test database of the SVM classifier for training and testing the model.Secondly,based on the fault detection rate,false alarm rate,robustness and other indicators,the fault detection system established above is evaluated.Finally,based on the laboratory EMA semi-physical experimental platform,the fault detection software is established,and the effectiveness of the fault detection method is further proved by experiments.Through the research of this paper,the fault detection method of electromechanical actuation system is given.At the same time,the feature extraction methods of different fault modes under the same fault performance are analyzed,and the differences between different fault mode data are sorted out to provide a judgment basis for fault separation,and further Provides theoretical support for system security management and remaining service life.