Interturn short circuit (ITSC) fault diagnosis at its early stage is very essential to improve the security of permanent magnet synchronous motors. In this article, a Bayesian optimization (BO) based residual convolutional neural network (CNN) algorithm for ITSC fault diagnosis was proposed. There are mainly two aspects of the contributions. First, it is a challenge to apply a conventional CNN on time-series signal tasks as they can only look back on history with a liner size, which will limit the network depth. Moreover, to obtain enough characteristics from acquired signals, the proposed algorithm requires adequate network depth, which may result in degradation difficulties. To solve these problems, a residual connection is embedded into the dilated CNN. Second, with the increase of network depth, the possible combination of hyperparameters will grow geometrically, thus the hyperparameters tuning will cost considerable time. To overcome this, the BO algorithm is adopted to select the optimal hyperparameter combination automatically. To demonstrate the effectiveness of the proposed algorithm, a motor fault experiment with various operating conditions was conducted on the motor that can be set for 17 fault severity levels. The test results and comparisons with other five algorithms show the advantage of the proposed algorithm.

Recursive Least Square (RLS) algorithms are considered as a kind of accurate parameter identification method for Lithium-ion batteries. However, traditional RLS algorithms usually employ a fixed forgetting factor, which do not have adequate robustness when the algorithm is interfered. In order to solve this problem, a novel variable forgetting factor method is put forward in this paper. Comparing with traditional variable forgetting factor methods, it has higher stability and sensitivity by using some mathematic improvements. The improvements in robustness of Recursive Least Square with a Variable Forgetting Factor (VFF-RLS) algorithm is verified in this paper. A Thevenin model which is frequently-used in Battery Management System (BMS) is employed in the verification. A data loss battery working condition is designed to simulate the interference to the algorithm. A simulation platform is established in MATLAB/Simulink software, and the data used in the verification is obtained by battery experiments. The analysis indicated that the novel VFF-RLS algorithm has better robustness and convergence ability, and has an acceptable identification accuracy.

Although the demand of battery electric vehicle (BEV) growths fast as the requirement of reducing greenhouse gas emission and the usage of fossil fuels, the limited driving range and unfriendly retail price present barriers to BEV to provide a comparable performance as traditional vehicle. This study proposes a dual-motor two-speed direct drive BEV powertrain to boost average motor operational efficiency in daily driving without increasing any complexity of manufacturing or control, ultimately, saving limited battery energy and manufacturing cost. The specifications of proposed powertrain are firstly identified through mathematical and graphical calculations, which split traditional one propelling motor to two with separate permanent engaged gears to maximize the motor efficiency. Based on dynamic powertrain modelling in a Simulink/Simscape®, economic shifting strategy and dynamic torque transfer control are designed and tested. According to the simulation results, it is noticed that significant energy efficiency improvement can be achieved. Thanks to the optimized torque transfer control strategy, extremely low vehicle jerk is recorded during the shifting process. At last, conclusions can be made that the proposed dual-motor powertrain superior to the traditional single motor counterpart in terms of fuel economic, driving range and cost.

Due to its low loss and high switching frequency, the silicon carbide metal oxide field effect transistors (SiC MOSFETs) are more suitable as switching devices in power inverter for electric vehicles (EVs). However, the power losses of power inverter would be dramatically increased with the rise of switching frequency, which would result in the limitation of using the high switching frequency performance of SiC MOSFET. In view of this case, this paper would systematically analyze the power losses distribution of power inverter and further optimize its efficiency under the high switching frequency. It is demonstrated that the power losses of power inverter are linearly increased with the rise of switching frequency, which is mainly caused by the switching losses of MOSFET chips increment. Based on this situation, an adaptive discontinuous pulse width modulation (ADPWM) is proposed to optimize the power inverter efficiency under the high switching frequency, which can effectively reduce the total power losses of power inverter and improve its efficiency compared with that of the conventional PWM strategies. Finally, the power losses and efficiency of power inverter with the proposed ADPWM strategy is verified by the experimental results.

Rolling element bearings are widely employed in almost every rotating machine. The health status of bearings plays an important role in the reliability of rotating machines. This paper deals with the principle and application of an effective multi-sensor data fusion fault diagnosis approach for rolling element bearings. In particular, two single-axis accelerometers are employed to improve classification accuracy. By applying the improved detrended fluctuation analysis (IDFA), the corresponding fluctuations detrended by the local fit of vibration signals are evaluated. Then the polynomial fitting coefficients of the fluctuation function are selected as the fault features. A multi-sensor data fusion classification method based on linear discriminant analysis (LDA) is presented in the feature classification process. The faults that occurred in the inner race, cage, and outer race are considered in the paper. The experimental results show that the classification accuracy of the proposed diagnosis method can reach 100%.