In order to solve the problem of weak characteristic signals and difficult detection caused by incipient faults of IGBTs in inverters, a fault diagnosis method is proposed, which extracts fault characteristic parameters from the inverter output and utilizes PCA (principal component analysis) for fault detection. By analyzing the relationship between incipient fault in Insulate-Gate Bipolar Transistor (IGBT) and inverter output voltage, overshoot, peak time and peak voltage slope are selected as the transient fault parameters, and output steady-state voltage is selected as the steady-state fault characteristic parameters. These fault characteristic parameters are extracted by the method of time-slice segmentation. On this basis, PCA is used to decompose the fault characteristic parameter samples into the principal subspace and residual subspace, where statistical indices T² and SPE are calculated to determine whether there is a fault. Simulation and experimental tests results demonstrate that, compared with the existing methods, the proposed method achieves accurate diagnosis of incipient faults in IGBT, offering significant value for applications.