The state-of-charge (SOC) is key to the safe operation and energy management of electric vehicles. The traditional SOC estimation algorithm based on Kalman filter and recurrent neural network requires a period of data to ensure the convergence of estimation results and cannot accurately estimate the SOC value near the starting point. A convolutional neural network SOC estimation method using U-Net structure is proposed, which can deal with variable-length input data and output the equal-length SOC estimation results. Meanwhile, it can also accurately estimate SOC at the starting point. In addition, a full-variance loss function is put forward, which can improve the estimation stability only by optimizing the loss function without increasing the model complexity and significantly reduce the maximum error. This model is trained with dynamic driving cycle data under five constant-temperature conditions, and the SOC estimation results are highly accurate under both the constant-temperature and variable-temperature conditions. Under the constant-temperature condition, the mean absolute error (MAE) and root mean square error (RMSE) of SOC estimation were within 1.1% and 1.4%, respectively. Under the variable-temperature condition, the MAE and RMSE were within 1.5% and 1.8%, respectively.

Owing to the advantages of lithium-ion batteries such as high energy density, long service life and low internal resistance, they have been widely applied to the fields of electric vehicles and energy storage. During their utilization process, their temperature will increase due to the internal chemical reactions. However, the traditional equivalent circuit model of batteries based on the voltage and current characteristics does not consider the thermal process, which cannot satisfy the precision requirement for battery models in practical applications. To solve this problem, a construction method for high-precision thermoelectric coupling model with measurable parameters is proposed. First, a thermal model of batteries based on back propagation neural networks is established using the Bayesian regularization method, and the battery surface temperature under different charge and discharge conditions is obtained. Then, the obtained characteristics of battery temperature are further input into a dual-polarization equivalent circuit model, so as to construct a thermoelectric coupling model of lithium-ion batteries. Compared with the traditional equivalent circuit model method which does not take into account the thermal effect, the established model can more accurately predict the changes in the temperature, capacitance and internal resistance of lithium-ion batteries with time, with an average reduction of 44.395% in output voltage error.

Breakover diode (BOD) has become a commonly used solution for thyristor overvoltage protection owing to its excellent volt-ampere and temperature characteristics. At present, high-voltage BOD is mainly monopolized by some foreign companies. To solve the bottleneck problem, a high-speed overvoltage triggering circuit without BOD is proposed, which uses a high-speed signal processing and amplification circuit consisting of components such as operational amplifiers and output transformer less (OTL) complementary symmetric amplification circuits to achieve fast overvoltage detection and protection. Finally, an experimental platform of 1 200 V thyristor overvoltage triggering circuit was built in the laboratory. Moreover, a comparison with the BOD triggering scheme fully verified the accuracy, speed and reliability of the proposed scheme.

Extracting the junction temperature of an insulate-gate bipolar transistor (IGBT) module is an essential part for its health monitoring and lifetime evaluation. The commonly used open-loop observer suffers from a lack of disturbance rejection capability, while the closed-loop scheme is infeasible since the states of a heat transfer system are not completely observable. Therefore, a composite observer suitable for online IGBT junction temperature extraction is proposed, where a closed-loop part is built therein via the observable states and the residual unobservable states construct an open-loop part. Aimed at an H-bridge inverter constructed with FF225R12ME4 IGBT chip, the design procedure of the composite observer is given in detail. Finally, real-time simulations of the composite observer were conducted on a hardware-in-loop (HiL) system, i.e., dSPACE DS1202. The HiL simulation results verified the superiority of the proposed composite observer in suppressing the effect of loss calculation errors on the estimation accuracy over the conventional open-loop observer.

The research on the driving technology for SiC devices has always been a hot issue for scholars both at home and abroad. The existing active drive technology realizes the regulation of switching transient by changing the drive resistance, drive voltage or current. It requires many components, and the drive board is bulky and costly. Therefore, a full-bridge active drive technology based on resonant drive is proposed. By switching on and off the drive switch tube only, different drive circuits are constructed to form different drive currents, thus realizing the switch transient adjustment. First, the 1/4-period full-bridge resonant drive is taken as a reference drive mode for the power tube. Then, a specific design method for active drive switch timing is given by analyzing various stages during the switching on process of the power tube. Finally, through simulations and experimental verification, it is found that the drive mode can reduce 87.0% of current overshoot during the switching on process and 4.2% of voltage overshoot during the switching off process, and it can also achieve flexible speed regulation in a wide range of specific stages.

Aimed at the problem that it is difficult for an LLC resonant converter to strike a balance between its dynamic performance and disturbance rejection capability in applications with high dynamic demand and frequent load changes, a charge active disturbance rejection control strategy is proposed. In this method, charge control is carried out in the inner loop of the resonant converter by collecting the resonant capacitor voltage, and active disturbance rejection control is introduced in the outer loop to form a compound control strategy of dual-closed-loop control. By deducing the relationship between the resonant capacitor voltage and resonant current of the LLC resonant converter, the controller parameters in the inner voltage loop of the converter are designed by using the charge control method, and an outer voltage loop controller is designed based on linearized active disturbance rejection control, thus further improving the anti-interference capability of the whole system. An experimental prototype with rated power of 300 W was designed and built, and the feasibility and effectiveness of the improved control strategy was verified by comparing with the traditional PID-PI control.

The DC-DC converter in a switch mode power supply (SMPS) is the core part that affects the device's volume, weight and working efficiency. As a classic DC-DC topology, the LLC resonant converter uses the soft switching technology and magnetic integration technology, which has characteristics such as a high efficiency, a high power density and harmonics suppression. The research status of optimization methods for LLC resonant converters applied in SMPS is reviewed, starting from the transformer winding structure schemes and control topology optimization schemes. In addition, suggestions on the effects of synchronous rectification and the planar transformers under an all-primary-referred (APR) magnetic integration model on the circuit are given. Finally, the optimization methods based on the third-generation wide bandgap material and electromagnetic compatibility are prospected.

The air gap in an integrated high-frequency transformer of an LLC resonant converter is usually placed at the center of magnetic core, and a single air gap structure is adopted. The nonlinear three-dimensional transformer models with different air gap positions are established using a finite element analysis method based on the software Ansys, and the influence of air gap position on loss is studied. Through simulations, the curves of loss versus air gap positions are obtained, and it is proved that the loss of transformer can be reduced and the converter efficiency can be improved if the air gap is located in the region of secondary windings, providing a reference for the optimal design of the integrated high-frequency transformer. On this basis, an integrated high-frequency transformer structure with three air gap magnetic circuits on the secondary side is proposed, so as to further reduce the transformer loss. Finally, an experimental prototype of a 160 W LLC dimmable LED driver was built, and experimental results verified the correctness and feasibility of the proposed method.

Aimed at the problem of backflow power generated by a half-bridge CLL resonant converter, a parameter design method based on the equivalent resonant cavity is proposed to optimize the backflow power and improve the conversion efficiency. First, according to the operating characteristics of an LLC converter, it is concluded that the value of backflow power can be characterized by the zero-crossing time of resonant current. Under the conditions of satisfying the voltage gain and zero-voltage switching of a switch on the primary side, the optimal resonant cavity parameters are selected according to the function curve representing the backflow power. Second, through the quantitative relationship between the CLL and the LLC resonant cavity, the resonant cavity is adjusted to be equivalent, so that the CLL has the same backflow power characteristics as the LLC. Finally, two sets of resonant cavity parameters meeting the design conditions were used for comparison and verification through simulations and experiments. Results show that under the condition of equivalent resonant cavity, both converters had the same zero-crossing time of resonance current, and the conversion efficiency was improved by about 1.3% under the conditions of quasi-resonance and full load, thus verifying the correctness and feasibility of theoretical analysis.

The typical isolated dual-active bridge (DAB) converter with a simple circuit structure and easy control is widely applied under scenarios where the bidirectional energy flow is required. Therefore, it is particularly important to study how to improve the working efficiency of the converter. First, the efficiency optimization strategies proposed by domestic and foreign scholars are compared and analyzed, and it is found that the intra- and inter-bridge phase shift angles on two sides have a strong correlation, which causes the difficulty in analyzing the working characteristics of the converter. Second, the rectified average current of the converter based on extended-phase-shift (EPS) control is optimized, a new phase shift angle is defined, and the transmission power and average current of the converter under the new phase shift angle are analyzed. Finally, a simulation platform based on Simulink and an experimental platform were set up for verification.