Artificial Intelligence-Aided Model Predictive Control for a Grid-Tied Wind-Hydrogen-Fuel Cell System

This paper considers the optimal economic dispatch of power generators in a smart electric grid for allocating power between generators to meet load requirements at a minimum total cost. We present a decentralized algorithm where, each generator independently adjusts its power output using only a measurement of the frequency deviation of the grid and minimal information exchange with its neighbors. Existing algorithms assume that frequency deviation is proportional to the load imbalance. In practice this is seldom exactly correct. We assume here that the only thing known about this relationship is that it is an unknown, odd, strictly increasing function. We provide a proof of convergence and simulations verifying the efficacy of the algorithm.

The increased complexity of modern energy network raises the necessity of flexible and reliable methods for smart grid operation. To this end, this article is centered on the economic dispatch problem (EDP) in smart grids, which aims at scheduling generators to meet the total demand at the minimized cost. This article proposes a fully distributed algorithm to address the EDP over directed networks and takes into account communication delays and noisy gradient observations. In particular, the rescaling gradient technique is introduced in the algorithm design and the implementation of the distributed algorithm only resorts to row-stochastic weight matrices, which allows each generator to locally allocate the weights on the messages received from its in-neighbors. It is proved that the optimal dispatch can be achieved under the assumptions that the nonidentical constant communication delays inflicting on each link are uniformly bounded and the noises embroiled in gradient observation of every generator are bounded variance zero mean. Simulations are provided to validate and testify the effectiveness of the presented algorithm.

This paper studies the dc power supply system to combine with the hybrid renewable energy sources. A multi-input power converter (MIPC) which operates in four types: an operation type wherein power is delivered to dc Bus from hybrid renewable energy sources; a single type wherein only one renewable energy source supplies power to the dc Bus; an inverter type wherein power is delivered to dc Bus from ac grid via inverter module, and a battery type wherein power is delivered to dc Bus from batteries without renewable energy source. In this paper, the integration of the dc Bus and a hybrid renewable power supply system is proposed. A hybrid renewable power energy which integrates the solar energy, wind-power, and city power, etc will be exploited to supply the household appliance and loads.

In this paper, a distributed optimization algorithm based on state-based potential game (SPG) is proposed for the real-time economic dispatch (RTED) problem in smart grids. Under the DC power flow approximation, we formulate the RTED with coupled operational constraints as a centralized optimization problem (C-RTED). By treating each node in the grid as an agent, we convert the C-RTED into an SPG by augmenting its objective function in the designed game with a local augmented Lagrange-like function, leading to a distributed algorithm for solving the C-RTED problem. We reveal that the stationary-state Nash equilibrium of the SPG exactly identifies the global optimum of the constrained C-RTED problem. The proposed algorithm is capable of handling both equality and inequality constraints in complicated forms, and can be further simplified to entail lower communication burden. Simulations on the IEEE 9-, 39-, and 118-bus systems confirm that the algorithm can quickly converge to the global optimum even under unreliable communication and plug-and-play operations.

Event-triggered scheme for finite-time distributed economic dispatch in smart grids

This paper is dedicated to solving the economic dispatch problem in smart grids, which aims at minimizing the total generation cost while satisfying the power supply-demand balance and generation capacity constraints. A distributed algorithm is proposed to solve the economic dispatch problem over directed communication topologies that are time-varying unbalanced. Its features are that the proposed algorithm is push-based, and the model of line losses is integrated into the problem of our interest to reduce the power losses. Resorting to geometric graph theory and convex analysis, it is proved that the incremental costs achieve consensus at a convergence rate O(1 / √t ) with t being the number of iterations. Furthermore, robustness of the proposed algorithm is investigated solving the EDP when the communication information undergoes arbitrary large but bounded constant delays and the local gradients involve observation noises with zero-mean and bounded variance. Finally, three case studies implemented on an islanded microgrid, the IEEE 30-bus system and the IEEE 118-bus system are tested to demonstrate the effectiveness of the proposed algorithm.

Sustainable Energy Systems: Planning, Integration and Management provides integrated perspectives on engineering, management, and technological perspectives to examine how renewable energy sources, smart grids, artificial intelligence, and advanced forecasting methods can be effectively planned, integrated, and managed within contemporary energy infrastructures. The book begins by introducing AI-enabled smart grid systems and their role in improving energy efficiency, reliability, and grid optimisation. The book highlights the technology, storage and working principles behind major renewable energy sources, including solar, wind, hydro, geothermal, and biomass. Chapters also focus on the use of computational intelligence in forecasting, management and optimisation of major energy sources. The book examines frameworks, principles, supply chains, and policy-driven approaches to sustainable energy governance. Real-world case studies on energy management provide practical insights into scalable and implementable solutions. By combining theoretical foundations with applied case studies and modern digital technologies, this book serves as an essential resource for understanding the planning, integration, and management of sustainable energy systems in the era of intelligent and renewable-powered infrastructure. Key Features: -Comprehensive coverage of renewable energy technologies, including solar, wind, hydro, geothermal, biomass, and hydrogen energy -Exploration of AI-enabled smart grids, energy forecasting, and intelligent energy management systems -Integration of Industry 4.0, machine learning, and digital technologies in energy optimisation and planning -Detailed discussion on hybrid renewable systems, energy storage, and grid integration strategies

Shanghai is one of China's highest levels of industrialization and most economically developed regions. In the backdrop of the development of smart grid, smart electric power industry will be to play a very important position in the Shanghai urban area's electric power industry. This paper will focus on the analysis of the present situation of Shanghai urban area's power grid, especially the basic skills and business models of the power demand side. It is researched that the blueprint of the Shanghai urban area's future power grid and achieving methods based on the related smart electricity work carried on in the current Shanghai urban area and its effectiveness, especially, the pilot project. The main work is as follows: the description of the current Shanghai urban area power grid, including the history, development and evolution, load characteristics level, reliability, power distribution automation level and so on. The presented situation of smart power, which mainly includes the information collection system pilot project, electric vehicle charging facilities pilot project, intelligent building pilot project, flexible DC power transmission pilot project, the power FTTH (Fiber to The Home) to households pilot project, especially combine with the smart grid key pilot projects, such as the smart power design and practice of Shanghai World Expo smart grid pilot projects, and point out the current shortage and inadequate of the smart grid. The blueprint and achieving methods of the smart electricity describes the design blueprint of the future smart electricity in Shanghai urban area, and the means and direction to realize, such as to adapt and promote clean energy development, realizing highly smart grid dispatching to meet the electric vehicle and other new electricity consumers of electricity service requirements, efficient use and life cycle management of grid assets, convenient interaction between power users and grid, grid management informationalization and delicacy and playing the role of value-added services potential of the grid infrastructure.

The Economic Dispatch Problem (EDP) plays a critical role in power system operations by trying to allocate power generation across multiple units at minimal cost while satisfying complex operational constraints. Traditional optimization techniques struggle with the non-convexities introduced by factors such as valve-point effects, prohibited operating zones, and spinning reserve requirements. While metaheuristics methods have shown promise, they often suffer from convergence issues and constraint-handling limitations. In this study, we introduce a novel application of Group Relative Policy Optimization (GRPO), a reinforcement learning framework that extends Proximal Policy Optimization by integrating group-based learning and relative performance assessments. The proposed GRPO approach incorporates smart initialization, adaptive exploration, and elite-guided updates tailored to the EDP’s structure. Our method consistently produces high-quality, feasible solutions with faster convergence compared to state-of-the-art metaheuristics and learning-based methods. For instance, in the case of the 15-unit system, GRPO achieved the best cost of USD 32,421.67/h with full constraint satisfaction in just 4.24 s, surpassing many previous solutions. The algorithm also demonstrates excellent scalability, generalizability, and stability across larger-scale systems without requiring parameter retuning. These results highlight GRPO’s potential as a robust and efficient tool for real-time energy scheduling in smart grid environments.

This paper gives an overview of the Particle Swarm Optimization technique with the aim of investigating its applicability to minimizing losses in a power distribution system. The suitability of this method is investigated by emphasizing the impact of its intrinsic parameters on the results. The investigation employs particle swarm optimization on a test system by controlling the generator voltages, transformer taps and the reactive power in the capacitor banks for each load bus. In total it contains 33 control variables: 5 generators (not including the terminal bus), 4 transformers and 24 capacitor banks. A theoretical approach of the PSO is given in the beginning, followed by the simulation results and analyses.

SummaryThis paper studies distributed resource allocation problem for agents with input dead‐zone, which is not considered in the existing work. At first, the primal problem is transformed to an auxiliary problem by using the exact penalty method to deal with local inequality constraints. It is assumed that the explicit expressions of cost functions and local inequality constraints are unknown to agents but the values of the cost and constraint functions can be obtained. Under such a setup, the extremum seeking control is used to estimate the gradient information. Thus, to obtain the optimal allocation, a novel distributed algorithm is designed by the virtue of the extremum seeking control and a dynamic compensating mechanism which is used to handle the effects of the input dead‐zone. Due to a two time‐scale structure of the designed distributed algorithm, the semi‐globally practically asymptotical convergence of all agents' decisions to the optimal allocation is obtained by the singular perturbation technique. Finally, numerical examples of economic dispatch in smart grids are given to verify the effectiveness of our proposed method.

As the proportion of flexible load resources in smart grids continues to rise, resulting in increasingly complex grid structures, significant changes in grid characteristics, and increased risks to grid operation and control, it will be difficult to intelligently regulate the grid solely by relying on traditional resource regulation methods, and the dispatchable space for traditional resources will become smaller and smaller. To this end, this paper conducts an exploratory investigation on the modeling techniques for flexible loads participation in smart grid dispatching. First, a classification of the flexible loads involved in grid regulation is made. Secondly, according to the flexible load classification, the modeling techniques of different classes of flexible loads are reviewed and studied; then, the flexible load dispatching modes for different operating states and different control tasks, and under different control methods are discussed deeply. Moreover, the technological economics and feasibility of these different flexible load dispatching modes are compared. Finally, an outlook and conclusion are made.

Short-Term Residential Load Forecasting (STRLF) is a core task in smart grid dispatching and energy management, and its accuracy directly affects the economy and stability of power systems. Current mainstream methods still have limitations in addressing issues such as complex temporal patterns, strong stochasticity of load data, and insufficient model interpretability. To this end, this paper proposes an explainable and efficient forecasting framework named KAN+Transformer, which integrates Kolmogorov–Arnold Networks (KAN) with Transformers. The framework achieves performance breakthroughs through three innovative designs: constructing a Reversible Mixture of KAN Experts (RMoK) layer, which optimizes expert weight allocation using a load-balancing loss to enhance feature extraction capability while preserving model interpretability; designing an attention-guided cascading mechanism to dynamically fuse the local temporal patterns extracted by KAN with the global dependencies captured by the Transformer; and introducing a multi-objective loss function to explicitly model the periodicity and trend characteristics of load data. Experiments on four power benchmark datasets show that KAN+Transformer significantly outperforms advanced models such as Autoformer and Informer; ablation studies confirm that the KAN module and the specialized loss function bring accuracy improvements of 7.2% and 4.8%, respectively; visualization analysis further verifies the model’s decision-making interpretability through weight-feature correlation, providing a new paradigm for high-precision and explainable load forecasting in smart grids. Collectively, the results demonstrate our model’s superior capability in representing complex residential load dynamics and capturing both transient and stable consumption behaviors. By enabling more accurate, interpretable, and computationally efficient short-term load forecasting, the proposed KAN+Transformer framework provides effective support for demand-side management, renewable energy integration, and intelligent grid operation. As such, it contributes to improving energy utilization efficiency and enhancing the sustainability and resilience of modern power systems.

Climate change is one of the most pressing issues of our time and mitigating it requires a reduction of CO2 emissions. A big step towards achieving this goal is increasing the share of renewable energy sources, as the energy sector currently contributes 35% to all greenhouse gas emissions. However, integrating these renewable energy sources challenges the current power system in two major ways. Firstly, renewable generation consists of more spatially distributed and smaller power plants than conventional generation by nuclear or coal plants, questioning the established hierarchical structures and demanding a new grid design. Restructuring becomes necessary because wind and solar plants have to be placed at favorable sites, e.g., close to coasts in the case of wind. Secondly, renewables do not provide a deterministic and controllable power output but introduce power fluctuations that have to be controlled adequately. Many solutions to these challenges are build on the concept of smart grids, which require an extensive information technology (IT) infrastructure communicating between consumers and generators to coordinate efficient actions. However, an intertwined power and IT system raises great privacy and security concerns. Is it possible to forgo a large IT infrastructure in future power grids and instead operate them purely based on local information? How would such a decentrally organized system work? What is the impact of fluctuation on short time scales on the dynamical stability? Which grid topologies are robust against random failures or targeted attacks? This thesis aims to establish a framework of such a self-organized dynamics of a power grid, analyzing its benefits and limitations with respect to fluctuations and discrete events. Instead of a centrally monitored and controlled smart grid, we propose the concept of Decentral Smart Grid Control, translating local power grid frequency information into actions to stabilize the grid. This is not limited to power generators but applies equally to consumers, naturally introducing a demand response. We analyze the dynamical stability properties of this framework using linear stability methods as well as applying numerical simulations to determine the size of the basin of attraction. To do so, we investigate general stability effects and sample network motifs to find that this self-organized grid dynamics is stable for large parameter regimes. However, when the actors of the power grid react to a frequency signal, this reaction has to be sufficiently fast since reaction delays are shown to destabilize the grid. We derive expressions for a maximum delay, which always desynchronizes the system based on a rebound effect, and for destabilizing delays based on resonance effects. These resonance instabilities are cured when the frequency signal is averaged over a few seconds (low-pass filter). Overall, we propose an alternative smart grid model without any IT infrastructure and analyze its stable operating space. Furthermore, we analyze the impact of fluctuations on the power grid. First, we determine the escape time of the grid, i.e., the time until the grid desynchronizes when subject to stochastic perturbations. We simulate these events and derive an analytical expression using Kramer's method, obtaining the scaling of the escape time as a function of the grid inertia, transmitted power, damping etc. Thereby, we identify weak links in networks, which have to be enhanced to guarantee a stable operation. Second, we collect power grid frequency measurements from different regions across the world and evaluate their statistical properties. Distributions are found to be heavy-tailed so that large disturbances are more common than predicted by Gaussian statistics. We model the grid dynamics using a stochastic differential equation to derive the scaling of the fluctuations based on power grid parameters, identifying effective damping as essential in reducing fluctuation risks. This damping may be provided by increased demand control as proposed by Decentral Smart Grid Control. Finally, we investigate discrete events, in particular the failure of a single transmission line, as a complementary form of disturbances. An initial failure of a transmission line leads to additional load on other lines, potentially overloading them and thereby causing secondary outages. Hence, a cascade of failures is induced that propagated through the network, resulting in a large-scale blackout. We investigate these cascades in a combined dynamical and event-driven framework, which includes transient dynamics, in contrast to the often used steady state analysis that only solves static flows in the grid while neglecting any dynamics. Concluding, we identify critical lines, prone to cause cascades when failing, and observe a nearly constant speed of the propagation of the cascade in an appropriate metric. Overall, we investigate the self-organized dynamics of power grids, demonstrating its benefits and limitations. We provide tools to improve current grid operation and outline a smart grid solution that is not reliant on IT. Thereby, we support establishing a 100% renewable energy system.

The electricity production from Renewable Energy (RE) in isolated locations requires long-term energy storage systems. To that end, Hybrid Energy Storage Systems (HESS), through a combination of hydrogen and batteries, can benefit from the different advantages of both technologies. This paper presents a hybrid Power-to-Power (PtP) Optimal Power Dispatch (OPD) model for isolated systems with no electric grid access. Currently, the electricity supply in such cases is usually based on a mix of RE as the primary energy source sustained by a diesel genset acting as a backup generator. In this context, the model delivers the hourly energy flows between renewable production sources, energy storage devices and the electrical load, which minimises costs and Green House Gases (GHG) emissions. For validation purposes, the model was tested through its application to a case study in an isolated area in the Canary Islands, Spain. The results show that the algorithm calculates the hourly OPD successfully for a given plant sizing, considering the defined operational states of the different assets. These operational constraints showed a decrease in the PtP round-trip efficiency of 5.4% and a reduction of the hydrogen production of 9.7%. Finally, the techno-economic analysis of the results proves that the combination of hydrogen and batteries with RE production is a feasible alternative to phasing out fossil fuels for the selected case study – reducing the diesel generator usage down to 1.2% of the yearly energy supply.