Fault detection methods for voltage source converters based on state characterisation and gap-metric

To ensure the operational reliability and fault-tolerance accuracy of voltage source converter (VSC), this paper proposes a diagnostic method for fault clustering and quantitative description. This method aims to achieve both fault classification and quantitative diagnosis of VSC systems. First, the state space model of the VSC system is established, and a Kalman filter is designed for state estimation to achieve the current. Second, fault clustering is accomplished by highlighting the fault characteristics of sensor faults, internal system component faults, and transistor faults through the deviation between actual values and predicted values getting from the Kalman filter. Thirdly, sensor faults and internal system component faults are quantified using data-driven gap metric to continuously characterise the degree of faults. Finally, the proposed method is applied on a three-phase inverter experimental platform. The experimental results demonstrate that the method proposed in this paper can effectively highlight VSC fault information and provide quantitative fault detection results.