Longyuan Fan

Longyuan Fan

Auto-generated profile, based on publicly available information

Longyuan Fan

Institute of Artificial Intelligence, Huazhong University of Science and Technology, Wuhan, China

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
6 November 2024 in IEEE Transactions on Industrial Electronics
Verified
Journal Article
Closed access
LFLongyuan Fan
Longyuan Fan
Zicheng LiuZicheng Liu
Zicheng Liu
Dong JiangDong Jiang
Dong Jiang
RQRonghai Qu
Ronghai Qu
The reduction of vibration and noise in active magnetic bearings (AMBs) within electric motors has garnered increasing attention. However, there has been limited research on the high-frequency (HF) vibration introduced by AMB controllers in relation to the switching frequency. In this article, we pioneer an exploration into the mechanism and characteristics of HF vibration in magnetic bearings. We introduce a carrier phase-shifting (CPS) method to analyze its impact on the spatial distribution of electromagnetic force in the air gap and vibration response. Subsequently, we construct a deep neural network (DNN) to model and predict the actual vibration of magnetic bearings, and integrate the grey wolf optimization (GWO) algorithm to optimize the CPS mode, phase-shifting angles, and switching frequency. Finally, experimental validation confirms the effectiveness of our proposed CPS method and artificial intelligence (AI) optimization algorithm. Our results demonstrate a remarkable reduction in HF vibration of AMB, achieving a maximum reduction of 90% and an average of 84% through intelligent optimization.

5 August 2025 in IEEE Transactions on Industrial Electronics
Verified
Journal Article
Closed access
LFLongyuan Fan
Longyuan Fan
Zicheng LiuZicheng Liu
Zicheng Liu
HWHaijiao Wang
Haijiao Wang
Dong JiangDong Jiang
Dong Jiang
RQRonghai Qu
Ronghai Qu
This article presents a novel online vibration suppression strategy to mitigate the dominant vibration issues in active magnetic bearing (AMB)-based permanent magnet synchronous motors, particularly those related to the rotational frequency spectrum. A dynamic axial displacement control (DADC) method is introduced to address the degradation in AMB control accuracy caused by machining errors. The study also investigates the impact of machining-induced inaccuracies on axial displacement sensor readings, providing insights for the design of dynamic DADC signals. To further enhance the adaptability of the DADC algorithm, an online vibration suppression strategy is developed, which integrates real-time vibration spectral amplitude extraction using the adaptive gradient linear neuron (AdaGrad-LN) method and an improved gray wolf optimizer with evolutionary population dynamics (GWO-EPD) for self-tuning the DADC control parameters. Compared with conventional GWO, the proposed GWO-EPD enhances population diversity by eliminating poorly performing individuals and repositioning them, thereby improving convergence speed and tuning accuracy. Experimental results demonstrate the effectiveness of the proposed approach. The GWO-EPD-based online vibration control achieves a 93.1% reduction in vibration amplitude and a 23 dB decrease in vibration acceleration compared with traditional AMB vibration control methods.

30 December 2025 in IEEE Transactions on Industrial Electronics
Verified
Journal Article
Closed access
LFLongyuan Fan
Longyuan Fan
Zicheng LiuZicheng Liu
Zicheng Liu
RZRongpei Zhu
Rongpei Zhu
Dong JiangDong Jiang
Dong Jiang
RQRonghai Qu
Ronghai Qu
This article presents a novel mechanism-data fusion modeling approach for predicting high-frequency vibrations in active magnetic bearings induced by pulse-width modulation (PWM) power supply. First, an analytical method is developed to characterize the spatial distribution of circumferential electromagnetic force waves in the air gap, and a spatial-order-based transfer function is constructed to describe the mechanical vibration response. Building upon this, a mechanism-informed neural network architecture is proposed, incorporating prior knowledge into a self-supervised framework, with constrained hidden nodes and learnable edge activation functions. This hybrid design enhances both prediction accuracy and model interpretability. Experimental results demonstrate that the mechanism-informed architecture achieves over 98% prediction accuracy on experimental data sets. Moreover, the model provides interpretable insights into vibration transmission mechanisms, offering guidance for interaction analysis and collaborative optimization of variable PWM control parameters.

17 February 2025 in IEEE Transactions on Industry Applications
Verified
Journal Article
Closed access
Zicheng LiuZicheng Liu
Zicheng Liu
HWHaijiao Wang
Haijiao Wang
PWPengye Wang
Pengye Wang
LFLongyuan Fan
Longyuan Fan
Dong JiangDong Jiang
Dong Jiang
RLRui Li
Rui Li
Active magnetic bearing (AMB) can circumvent the mechanical vibration inherent in traditional bearings. However, it can exacerbate the electromagnetic vibration caused by unbalanced magnetic pull (UMP). This article investigates the effect of the position control of AMB on the electromagnetic vibration reduction of permanent magnet synchronous motor (PMSM). The serious dynamic eccentricity arising from rotor position fluctuation leads to an increase in UMP, inducing severe electromagnetic vibration at rotational frequency. The proposed zero displacement control with dynamic compensation not only repositions the rotor to decrease displacement fluctuation but also dynamically adjusts the air gap distribution of the PMSM, thereby mitigating UMP caused by dynamic eccentricity. Ultimately this method can achieve notable rotational vibration suppression. Experimental results confirm the theoretical analysis and verify that this method can effectively reduce the vibration at the rotational frequency by more than 20dB.

17 September 2025 in IEEE Transactions on Power Electronics
Verified
Journal Article
Closed access
RZRongpei Zhu
Rongpei Zhu
GYGuang Yang
Guang Yang
HLHongyang Liu
Hongyang Liu
LFLongyuan Fan
Longyuan Fan
Zicheng LiuZicheng Liu
Zicheng Liu
RQRonghai Qu
Ronghai Qu
In motor drive system fed by voltage source inverter, common-mode voltage (CMV) is commonly present because of pulsewidth modulation (PWM) techniques. It is proved to be feasible in research to collect energy from CMV and provide a dc power source for low-power sensors such as encoders. This approach can replace auxiliary power supply, thus, simplifying system wiring design. However, the practicality of this approach will be largely constrained by the maximum power available from CMV, as it directly affects the stable operation of sensors. This article aims to explore methods to improve the power capacity with low cost. A common-mode (CM) power transfer model is constructed to clarify the factors influencing CM power extraction and analyze possible methods for increasing power capacity. Then, from the perspective of power source, the characteristics of CMV are analyzed including its frequency composition and rms value. Further exploration is conducted to analyze the influence of zero-sequence injection during modulation on the characteristics of CMV. At last, a discontinuous PWM method DPWM3 is identified as the most effective PWM method to significantly enhance power capacity. Experimental results validate the accuracy of quantitative analysis and effectiveness of the proposed method.