Yuchao Liu

In the motor-driven system, the utilization of a supercapacitor energy storage unit contributes to energy saving. However, the multi-link time-delay in the system makes the supercapacitor unable to immediately track and respond to the sharp change of motor power. This will cause the large DC bus voltage fluctuation, the peak current at the power grid side, and unnecessary energy loss. In this paper, a power advance prediction control strategy considering multi-link time-delay is proposed. The multi-link time-delay mechanism in the system is analyzed, in particular for the different conditions of motor power change. Afterward, with the motor speed, torque, and the inverter acceleration/deceleration rate obtained in advance, the motor steady power and power change rate are predicted. In the dynamic process, by adjusting the tracking rate in real time, the motor reference power function is derived. The reference current of the supercapacitor is given online, so as to match the power between the motor and the energy storage unit. Both simulation and experimental results verify the rationality and validity of the proposed control strategy. In the dynamic process, the system stability and energy utilization are successfully improved in comparison with multi-parameter collaborative power prediction control and double closed-loop control.

Modern Electric Vehicles (EVs) and grid-level energy storage systems rely on accurate lithium-ion battery capacity forecasting to ensure operational safety. However, conventional recurrent neural network (RNN) approaches often struggle to capture long-range dependencies, resulting in suboptimal predictions. Accurately predicting battery health over long-sequence remains a key challenge for developing advanced Battery Management Systems (BMS) and improving Remaining Useful Life (RUL) estimation. In this paper, a lightweight Transformer-LSTM (TF-LSTM) framework is developed. Compared to the traditional TF-LSTM, it integrates multi-head self-attention and positional encoding with LSTM's gating mechanism, while simultaneously leveraging Automatic Mixed Precision (AMP) to further reduce computational overhead. The performance of the proposed model has been validated through numerical simulations and leave-one-out cross-validation on four cells from the CALCE Li-ion Battery Aging Dataset. The experimental results indicate that the proposed approach achieves a 39.2% lower relative error (RE) compared to conventional RNNs for extended sequence lengths. Therefore, it demonstrates the enhanced generalization of the model, making it well-suited for large-scale EVs battery monitoring and industrial energy storage prognostics requiring precise long-horizon forecasting.