Compensation capacitors are naturally susceptible to manufacturing defects and aging effects, leading to the degraded performance of a wireless power transfer (WPT) system. This article focuses on the compensation parameters optimization during the design stage and control strategy during the operation phase to improve the inherent capacitor error tolerance of the WPT system. The Sobol sensitivity method is applied to rank the importance of deviations of three capacitors on the transfer characteristics, and then the method of tracking the secondary resonance frequency is proposed. The numerical method is applied to find the optimal compensation parameters, with the constraint that the output voltage change caused by the shift of the designed compensation condition is limited to be less than ±5%. Experimental results show that with the proposed frequency tracking method and compensation parameter optimization, the deviation tolerance index is decreased from 0.485 to 0.363, showing an improvement of 25.2%, and the minimum power factor is increased from 0.78 to 0.89. Besides, the characteristics of constant primary coil current and voltage gain are almost not affected.

In this paper, a deep learning framework combined with isogeometric analysis (IGA for short) called IGA-Reuse-Net is proposed for efficient reuse of numerical simulation on a set of topology-consistent models. Compared with previous data-driven numerical simulation methods only for simple computational domains, our method can predict high-accuracy PDE solutions over topology-consistent geometries with complex boundaries. UNet3+ architecture with interlaced sparse self-attention (ISSA) module is used to enhance the performance of the network. In addition, we propose a new loss function that combines a coefficients loss and a numerical solution loss. Several training datasets with topology-consistent models are constructed for the proposed framework. To verify the effectiveness of our approach, two different types of Poisson equations with different source functions are solved on three datasets with different topologies. Our framework can achieve a good trade-off between accuracy and efficiency. It outperforms the physics-informed neural network (PINN for short) model and yields promising results of prediction.