Research Article
Analysis of 50 Hz Neutral-Point Voltage Fluctuation of
DC-Link Capacitors for Three-level Inverters
Cheng Luo
,
Jianbo Cao*,
Mi Qi,
Dapeng Lu,
Hongwei Zhou
Issue:
Volume 14, Issue 4, December 2026
Pages:
172-178
Received:
2 September 2026
Accepted:
15 October 2026
Published:
10 November 2026
Abstract: Grid-tied inverters serve as key equipment to guarantee the reliable operation of modern power systems dominated by renewable energy sources. Under asymmetric grid conditions, injecting negative-sequence current is a widely adopted control strategy for three-level grid-connected inverters to suppress grid voltage unbalance. However, this paper discovers an unreported phenomenon that obvious 50 Hz fluctuations arise in the half DC-bus voltage of three-level inverters during negative-sequence current injection. To clarify the generation mechanism of this voltage fluctuation, Fourier series decomposition is employed to analyze the coupling relationship between sequence components of voltage and current. The theoretical analysis demonstrates that the interaction between negative-sequence voltage and negative-sequence current constitutes the primary cause of the 50 Hz oscillation on the half DC-link voltage. Based on the revealed mechanism, the allowable safe operating range of the half DC-bus voltage is deduced quantitatively. Finally, simulation tests are carried out in the MATLAB/Simulink environment. The obtained results verify the correctness of the theoretical derivation and the proposed operating range constraint. This work provides a theoretical reference for parameter design and stable operation of three-level grid-tied inverters under unbalanced grid faults.
Abstract: Grid-tied inverters serve as key equipment to guarantee the reliable operation of modern power systems dominated by renewable energy sources. Under asymmetric grid conditions, injecting negative-sequence current is a widely adopted control strategy for three-level grid-connected inverters to suppress grid voltage unbalance. However, this paper disc...
Show More
Research Article
Rule-Based Operation of a Grid-Connected Hybrid Power System in Coastal Bangladesh
Fahim Faisal,
Cheng Peng*,
Md Shivly Mahmood
Issue:
Volume 14, Issue 4, December 2026
Pages:
179-191
Received:
6 October 2026
Accepted:
26 October 2026
Published:
10 November 2026
Abstract: Coastal communities in Bangladesh, such as Teknaf in Cox's Bazar, suffer from severe energy insecurity due to a fossil-fuel-dominated grid, long transmission lines, and chronic load shedding. This study aims to design and evaluate a grid-connected hybrid power system to improve local energy reliability, reduce grid dependence, and lower carbon emissions. A 24-hour hourly simulation was conducted using real commercial components: five 95 kW wind turbines (475 kW total), a 78.12 kWp bifacial solar PV array, a 128 kWh LiFePO₄ battery, a 100 kVA diesel generator, and bidirectional grid exchange. A transparent, rule-based energy management strategy with four time-of-day operating modules was developed to coordinate the sources and serve a consumer load with a 70-kW evening peak. The simulation results show that renewable generation (1,150 kWh/day) exceeded the consumer demand (929 kWh/day). This reduced gross grid imports by 41.3% and net grid purchase, along with associated CO2 emissions, by 84.5% compared to a CG-only system. However, the temporal mismatch between renewable generation and the evening peak meant that the 128-kWh battery only reduced the peak grid demand from 70 kW to 67.5 kW. Sensitivity analysis identified wind speed and battery capacity as the most dominant design parameters.
Abstract: Coastal communities in Bangladesh, such as Teknaf in Cox's Bazar, suffer from severe energy insecurity due to a fossil-fuel-dominated grid, long transmission lines, and chronic load shedding. This study aims to design and evaluate a grid-connected hybrid power system to improve local energy reliability, reduce grid dependence, and lower carbon emis...
Show More
Research Article
Diffusion–GAT Dual-Channel Framework for Extreme Few-Shot and Noise-Robust Pump Fault Diagnosis
Wu Heng,
Gao Lei,
Wang Wenchao*,
Zhang Yu,
He Biao
Issue:
Volume 14, Issue 4, December 2026
Pages:
192-202
Received:
2 September 2026
Accepted:
29 September 2026
Published:
13 November 2026
DOI:
10.11648/j.ajee.20261404.13
Downloads:
Views:
Abstract: Fault diagnosis of main circulation pumps in converter station valve cooling systems is severely challenged by strong electromagnetic interference and extremely limited fault samples. To address these issues, this paper proposes a physics-informed dual-channel collaborative framework that integrates multi-field coupling priors into both data generation and feature learning. First, feature sensitivity and signal correlation matrices are derived from fluid-mechanical-electrical coupling mechanisms to formulate computable physical priors. Second, a physics-informed conditional diffusion model is developed, embedding physical constraints into the reverse denoising process to generate augmented samples consistent with fault evolution laws. Third, a mechanism-initialized graph attention network is constructed with learnable balancing factors that dynamically fuse physical priors with data-driven features. The two channels collaborate implicitly through sensitivity matrices, forming a closed-loop optimization of generation and learning. Experimental results under extreme conditions—only five training samples per class and a signal-to-noise ratio of 5 dB—demonstrate that the proposed method achieves a diagnostic accuracy of 96.8%, outperforming single-channel physics-embedded methods by 6.3 percentage points and purely data-driven methods by 12.2 percentage points. These results validate the effectiveness, robustness, and physical interpretability of the dual-channel collaborative mechanism.
Abstract: Fault diagnosis of main circulation pumps in converter station valve cooling systems is severely challenged by strong electromagnetic interference and extremely limited fault samples. To address these issues, this paper proposes a physics-informed dual-channel collaborative framework that integrates multi-field coupling priors into both data genera...
Show More
