Open-circuit fault diagnosis method for t-type inverter based on space trajectory classification network model

Compared with traditional inverters, T-type three-level inverter has the advantages of high power and withstand voltage level, low electromagnetic interference and low switching loss. It is widely used in the fields of medium and high voltage AC drive, active power filter and reactive power compensation of power system. Subsequently, the increase of switch brings the increase of device failure rate. The main fault of switch is open-circuit fault. This fault is not easy to find, which can seriously affect the safe operation of the system, and even can cause the paralysis of the whole converter system. This paper proposed an open-circuit fault diagnosis method for T-type inverter based on space trajectory classification network model. Firstly, phase currents were used as the fault diagnosis variables. The space trajectory under the conditions of normal, single-, and double-switch open-circuit faults were analyzed. For different open-circuit faults, the distortion of space trajectory and fault characterization were different. According to the fault characteristics of space trajectory, a fault diagnosis method based on space trajectory classification network model was proposed. The space trajectory was normalized and transformed into data point set. Then, symmetry function and permutation network were used to form a fault diagnosis model. The model was verified on a dSPACE platform. The results show that it can distinguish both single-and double-switch open-circuit faults effectively. Moreover, the proposed method is robust to different switching frequencies, load condition, and modulation depth. With minor modifications, it can be used in the railway application. © 2022, Central South University Press. All rights reserved.