Jingxuan Wu

The promotion of renewable energy technologies introduces multisource uncertainty into microgrid systems, posing increasing requirements to the intelligence of energy management systems (EMSs). This article proposes an intelligent energy management decision agent (EMDA) based on ResNet for a multienergy islanded microgrid (iMG) supported by hydrogen energy and PV system. Different from conventional EMSs that handle system stability constraints and economic optimization separately, the proposed EMDA forms a novel energy management framework that employs modern computer vision models to mine the potential mapping between optimal decisions and multidimensional system states. A ResNet-based fuzzy decision agent is trained on historical datasets to ensure the intelligence of the EMDA, and a cascaded defuzzification algorithm is designed to extract crisp decisions while maintaining the energy balance of the microgrid. The EMDA verifies the feasibility of the proposed innovative framework that bridges modern artificial intelligence (AI) models and energy systems, providing pretrained AI models with a novel interface method to energy systems. The comprehensive experimental validation demonstrates that the proposed EMDA can stably interface with the distributed energy resource control system and operate the energy flow in the iMG for more than 216 h with voltage oscillations within 2.5%. The results indicate that the EMDA outperforms existing rule-based and fuzzy logic online EMS approaches on system stability. Compared with theoretical optimum decisions, the proposed EMDA can achieve 89.9% of the economic performance regarding hydrogen consumption.

For a reliable integration of power-electronic-based generation in the future energy system, the grid-forming (GFM) converter has been extensively considered as a potential solution. However, while operating as a voltage source that is synchronized via active power regulation, the GFM converter exhibits insufficient damping, particularly when synthetic inertia is incorporated for frequency support. To address this concern, this article has developed active-damping control schemes using fractional-order regulators. Differing from the existing control approaches, which may significantly alter the inertial response quantified using an inertia constant, the proposed controls are developed to introduce adequate damping for both power set-point tracking and external disturbances, while maintaining the original explicit relationship between the GFM converter’s inertial response and the inertia constant. In this context, a preferable inertial response, precisely defined by the inertia constant as specified by certain grid codes, along with adequate damping, is achieved simultaneously. The effectiveness of the proposed controls has been validated though both theoretical analyses and experimental results.

For virtual synchronous generator (VSG) based microgrids, it is challenging to address the adverse effect of nonlinear loads and grid background harmonics with existing harmonic suppression methods. To address this issue, a decomposed grid harmonic current suppression (DHS) method with proper feedforward point of common coupling voltage and feedback VSG output current as well as nonlinear load current is proposed for VSG-based microgrids. The proposed DHS method can separately mitigate the grid harmonic current caused by nonlinear loads and grid background harmonics, obtaining a promising harmonic suppression effect without potential system instability problem. Compared with conventional harmonic control methods, the DHS method reduces the control sensitivity to system parameters including line impedance, grid impedance, and system shunt capacitance, improving the adaptivity of harmonic control in distorted grids. The proposed method's effectiveness and reliability are verified by theoretical analysis and comparative experiments.