Antonio Camacho

Modern power grids with high renewable energy penetration are vulnerable to fast voltage disturbances caused by grid faults. Among these, voltage sags are critical because they develop within milliseconds and require rapid reactive current support to maintain grid stability and power reliability. Reinforcement learning has previously shown potential for reactive current injection control during voltage sag events due to its fast response and adaptability to changing system conditions. However, existing approaches rely on separate policies for specific subsets of the operating space, which limits their ability to provide optimal actions when the system operates across broader or combined state regions. To address this limitation, this paper proposes a unified Soft Actor–Critic (SAC) target policy trained over the full state and action space by integrating multi-source transfer learning with potential-based reward shaping approach. Results show that the proposed multi-source transfer approach enables the target agent to converge faster and reach a higher reward solution than the baseline SAC and single-source transfer approach. The trained policy also improved prediction accuracy, achieving reactive-current errors below 0.2 A with respect to the ground-truth reference generated through extensive simulations over the full observation and action space. The reference follows the grid-code requirement for minimum reactive current injection during faults and provides a benchmark for evaluating prediction accuracy. This can help distributed generation sources respond more effectively during severe perturbations such as voltage sags, support voltage recovery, and reduce the risk of cascaded disconnections that could lead to unwanted blackouts. Additionally, the inference execution time is also sufficiently fast to satisfy the response-time requirement of voltage sag events, confirming the real-time feasibility of the proposed controller.

The presence of nonlinear loads in islanded microgrids produces harmful voltage harmonics that need to be addressed. Existing solutions established in the state-of-the-art aim to compensate the harmonic distortion at the node where the problematic loads are connected or at the inverters’ output. However, certain scenarios may require mitigation to be performed at a given node of the microgrid where sensitive equipment is installed. Given that the sensitive node could be located at any point of the network, it should be considered as a remote node from the inverters’ perspective. This article presents a novel control strategy to mitigate the voltage harmonic distortion from a remote sensitive node using the system’s already present grid-forming inverters. Its novelty lies in the use of complex-valued unilateral band-pass filters to reduce the voltage distortion of the different harmonic orders and their corresponding positive-negative sequences. Unlike conventional integrator-based approaches, the finite-gain behavior of the proposed controller allows inverters to collaboratively work toward the compensation, enabling seamless control integration. Furthermore, the use of optimally designed control parameters guarantees a fair distribution of the control effort between all inverters. The work includes the mathematical framework of the system followed by the design guidelines of the control to achieve its distinct characteristics. Finally, a selection of experimental results is presented to validate the behavior of the proposal.

In the event of a short-circuit fault in an islanded ac microgrid, grid-forming power inverters activate a resistive virtual impedance to protect themselves against over-current, whereas grid-feeding power inverters are tasked with providing voltage support. In this scenario, the sequence impedances that are responsible for voltage support exhibit significant variability depending on the values of both the virtual and short-circuit impedances. State-of-the-art control schemes do not take this variation into account, resulting in voltage support that could be enhanced. This article presents a control scheme designed for grid-feeding inverters that improve voltage support during unbalanced short-circuit faults. The scheme is based on an online measurement of the sequence impedances and the selection of the most appropriate current injection technique for voltage support using the measured impedance values. A theoretical analysis is also included, which was performed to verify the stability of the microgrid with the proposed control. Finally, experimental results from a laboratory microgrid are presented for validation purposes.

This article introduces a novel complex sliding-mode control strategy for grid-forming inverters in islanded microgrids. This approach offers an alternative to the traditional hierarchical control strategies that typically rely on droop control, communication-based secondary controls, and virtual impedance methods. To develop the complex sliding-mode control, we propose a complex phasor model for grid-forming inverters, presenting an alternative to the conventional power flow model used in ac microgrids. This model enables the derivation of a linear complex dynamic model for grid-forming inverters connected to a microgrid, based on the apparent power differential equation, thereby simplifying the control law design process. The proposed control system features three key advantages: first, it ensures precise active and reactive power sharing across both resistive and inductive power lines, eliminating the need for virtual impedance; second, it exhibits high robustness against line impedance mismatches and external disturbances, including load changes and the connection and disconnection of inverters; and third, it eliminates the requirement for secondary frequency control. Experimental results validate the effectiveness of the proposed control strategy.

Grid-connected ac microgrids (MG) are commonly exposed to voltage quality deterioration due to the presence of voltage imbalance. This article presents a control strategy that enables MG grid-feeding inverters (interfacing distributed generation units) to compensate for the voltage unbalance at a given remote node of the MG where for example a sensitive load is likely to be damaged. Each inverter is equipped with a new controller whose operation 1) allows to contribute in the control of the imbalance by only using a measure of the negative-sequence voltage at the remote node while 2) ensuring a fairly redistribution of the control effort among the participating inverters in terms of an inherent negative-sequence current sharing functionality. Since both control objectives are achieved without requiring the collaboration among the participating inverters, the presented solution offers an inherent robustness not found in previous state-of-the-art voltage unbalance control strategies. The new control strategy is accompanied by a controller parameter design receipt, and it is complemented by a formal analysis that permits evaluating how MG modeling errors and uncertainties affect stability and performance. Laboratory experimental results corroborate the properties of the presented control.

During voltage sags, three-phase three-wire power inverters can provide voltage support with several current injection strategies. In general, good results are obtained, except in overvoltage situations when one or more phase voltages exceed the allowed limit in grid codes. In three-wire inverters, this is a challenging problem because phase voltages are coupled. In this article, we present a strategy that performs independent reactive current injection for each phase, which solves the overvoltage problem. This approach also includes a current-limiting technique that guarantees the injection of the maximum active power under safe current conditions. A stability analysis is carried out based on the small-gain theorem. The properties of the proposed control are experimentally validated with selected tests in a laboratory setup. A performance comparison with state-of-the-art control strategies is also included in the experimental validation.

This paper introduces the application of a genetic programming (GP)-based method for the automated design and tuning of process controllers, representing a noteworthy advancement in artificial intelligence (AI) within the realm of control engineering. In contrast to already existing work, our GP-based approach operates exclusively in the time domain, incorporating differential operations such as derivatives and integrals without necessitating intermediate inverse Laplace transformations. This unique feature not only simplifies the design process but also ensures the practical implementability of the generated controllers within physical systems. Notably, the GP’s functional set extends beyond basic arithmetic operators to include a rich repertoire of mathematical operations, encompassing trigonometric, exponential, and logarithmic functions. This broad set of operations enhances the flexibility and adaptability of the GP-based approach in controller design. To rigorously assess the efficacy of our GP-based approach, we conducted an extensive series of tests to determine its limits and capabilities. In summary, our research establishes the GP-based approach as a promising solution for automating the controller design process, offering a transformative tool to address a spectrum of control problems across various engineering applications.

This paper introduces the implementation of a genetic programming (GP)-based procedure to the automatic design and tuning of process controllers. The proposed approach makes a significant contribution to the field of artificial intelligence (AI) in control engineering. Unlike other controller design methods, the GP-based program handles the entire design in the time domain, including differential operations like derivatives and integrals, without the need for intermediate inverse Laplace transformation. This approach not only simplifies the design process but also ensures that all generated controllers are implementable in physical systems. Furthermore, GP’s functions set includes various mathematical operations beyond basic arithmetic operators, such as trigonometric, exponential, and logarithmic operators. The performance and validity of the resulting controllers generated by the proposed GP-based approach are evaluated by verifying whether the generator can replicate the structure and performance of those produced by traditional controller design methods and, in some cases, achieve even better results. As a result, the GP-based approach presents a promising solution for automating the controller design process and addressing control problems in various engineering applications.