Xinshuai Zhang

Owing to the stringent reliability requirements of silicon carbide (SiC) MOSFETs, conventional gate-driving methods with fixed parameters are no longer sufficient to ensure reliable operation under varying operating conditions. To address this issue, this study proposes an adaptive gate-driving method based on model prediction dealing with operating conditions. First, the influence of operating conditions on crosstalk is comprehensively analyzed, starting with an explanation of the crosstalk mechanism. Then, a two-variable behavioral model is established to estimate the amplitude of the crosstalk voltage under the specified operating condition. Based on this prediction model, a lightweight optimization algorithm is developed to select the negative gate turn-off voltage according to the operating conditions. Finally, a double-pulse test platform is built to validate the proposed dynamic gate-voltage selection strategy. The experimental results show that, compared with the fixed −3 V turn-off voltage scheme, the proposed method selects the optimal turn-off voltage of −2 V under specific operating conditions, which can reduce unnecessary negative gate-voltage stress while maintaining the gate-source voltage within the allowable range. Additionally, the measured turn-off loss of the proposed method is 172.7 μJ, lower than the 222.3 μJ loss of the parallel gate-source capacitance method, indicating a better trade-off between crosstalk suppression and switching loss. The proposed behavioral model requires only 69.5 microseconds to predict positive and negative crosstalk peaks, whereas the analytical model requires 46.6 milliseconds. The low computational burden makes the proposed method suitable for potential real-time implementation on resource-constrained microcontrollers.

Remaining useful life (RUL) prediction of power semiconductor devices plays a crucial role in reliability design and predictive maintenance of power control system. This article introduces a data-driven methodology on predicting the RUL of the gate oxide layer in silicon carbide (SiC) MOSFETs. Firstly, a power cycling platform is established to collect the time-varying curves of threshold voltage and construct an aging dataset. Then, the successive variational mode decomposition (SVMD) algorithm is employed to adaptively decompose the signal of gate threshold voltage, helping suppress measurement noise and fluctuations caused by operating conditions while retaining degradation features. Subsequently, a Temporal Convolutional Network (TCN) is adopted to capture temporal dependencies in the degradation sequence, thereby improving the characterization of gate oxide health status assessment. Finally, the extended Kalman particle filter (EKPF) is employed to estimate the degradation state and quantify the associated uncertainty by recursively fusing model predictions with real-time measurements. The proposed method integrates the adaptive signal decomposition capability of SVMD, the temporal feature extraction capability of TCN, and the uncertainty quantification capability of EKPF. Their complementary integration improves prediction accuracy and robustness in gate oxide degradation evaluation for SiC MOSFET.

Stator winding reconfiguration technology has attracted increasing attention because it can modify the torque–speed output characteristics of motors online and extend the high-efficiency operating region. However, conventional winding switching methods suffer from several limitations, including long transient duration, typically at the millisecond level; difficulty in precise control; current interruption or asymmetric operation during the transient process; and the resulting torque loss and torque fluctuation. To address these issues, this paper first proposes a seamless winding switching method. The core idea is to complete the switching within the zero-vector interval of a single SVPWM period, thereby eliminating potential surge voltage without requiring a snubber circuit. The proposed method reduces the transient duration to only 5 μs, eliminates current interruption and current asymmetry, and suppresses the associated torque ripple. Furthermore, a deadbeat predictive control strategy is employed. By directly controlling the voltage vector, the proposed strategy enables fast and accurate current tracking within one sampling period after reconfiguration, thereby avoiding current overshoot, oscillation, and the corresponding torque and speed fluctuations. In addition, a voltage feedforward compensation method is proposed to pre-compensate the voltage applied during the PWM period in which the winding reconfiguration is performed, thereby preventing the application of voltage vectors that are inconsistent with the actual motor winding connection. With these three innovative measures, fast, smooth, and robust winding reconfiguration is achieved. The mechanisms and implementation procedures of the proposed strategies are described in detail, and their feasibility and effectiveness are verified through experiments.

Accurately assessing the remaining useful life (RUL) is a significant challenge to the reliability of Silicon Carbide (SiC) MOSFETs and is crucial for their safe operation. Consequently, this paper proposes a novel data-driven prediction method that integrates Successive Variational Mode Decomposition (SVMD), the Sparrow Search Algorithm (SSA), and the Transformer model. The threshold voltage V_th is selected as the degradation parameter for prediction. Firstly, SVMD is utilized to decompose the original V_th data into a degradation trend component and several fluctuation components with different central frequencies, thereby providing a more precise feature for prediction models. Subsequently, based on the Transformer model, trend predictions are conducted on each intrinsic mode function (IMF) derived from SVMD, and these results are aggregated as the final predicted value of V_th. The hyperparameters of the Transformer are optimized using SSA to enhance prediction accuracy. Ultimately, a power cycling platform is constructed to acquire the dataset of the device, where the device is subjected to rated current and 80 °C junction temperature fluctuation stress during testing. Building upon this, the difference between the number of cycles when V_th reaches its upper limit and the current number of cycles is determined as the predicted RUL value. Results demonstrate that compared to both a single Transformer model and the SVMD-Transformer model, the proposed method achieves a higher coefficient of determination (R²) and a lower root mean square error (RMSE), indicating superior prediction performance.