Distributed grid management requires real-time optimization for large-scale systems with renewable generators and controllable loads. AI techniques, such as consensus-based distributed computational intelligence, offer solutions to address the challenges of rapidly changing conditions, computation, and communication bottlenecks
[67]. AI has driven the development of decentralized and intelligent controllers, improving processing speed, reliability, and efficacy. These controllers distribute operations among distributed units, reducing the burden on centralized controllers and improving the resilience of the system
[68].
According to Zambrano and Giraldo
[70], predictive models based on AI for renewable energy hold the promise of revealing valuable glimpses into the expected energy enhancements in the near future. Ruhnau et al.
[71] believe that combining various approaches can refine these forecasts by making the most of the disparities in individual prediction models. These approaches encompass both standalone and integrated technologies that generate predictions based on distinct time series data derived from specific sources such as weather stations, wind turbines, or solar panels
[72]. To enhance forecast precision, the incorporation of information from nearby areas to the location of interest has become increasingly popular, particularly in recent years
[40].
AI has enabled the emergence of “prosumers”, allowing domestic energy users to both produce and consume electricity and share it with others. This shift from centralized, fossil-fueled generation to a decentralized, intelligent system enhances economic benefits for consumers, fostering energy sharing and trade
[73].
Sami
[74] described how to use AI in prosumers management. He pointed out that machine learning within the realm of artificial intelligence has the capability to assess and anticipate energy demand patterns and categorize irregular energy usage. By leveraging data collected through smart meters and subjecting it to AI analysis and data mining, it becomes feasible to discern various customer segments’ electricity consumption behaviors. Subsequently, this data can be employed to enhance statistical precision, facilitating the targeted delivery of advertisements and services. Fluctuations in the environment, such as variations in weather conditions, alterations in electrical appliance usage, and changes in consumer behavior, can impact the accuracy of anomaly detection results. Consequently, it is imperative to emphasize potential adverse aspects within the power grid that could influence the equitable distribution of power among consumers. The analysis of energy consumption is intrinsically linked to human characteristics, which can be addressed by extracting or taxonomy features. In this context, the development of deep learning models, particularly multilayered hidden neural networks, augments the predictive performance of energy demand and consumption.
Rodgers et al.
[75], in his study, underscores the significance of AI in smart grids, aligning goals with global sustainability objectives, emphasizing the role of ICT, and outlining practical requirements for smart grids. The study delves into the decision-making processes of experts and their knowledge transfer apparatus. It highlights the importance of information and communication technology (ICT) and AI usage in facilitating knowledge transfer for a greener environment. The researchers have identified three key goals for smart grids: universal access to electricity, environmental protection, and efficiency. These goals align with global sustainability objectives, such as those set by the United Nations Conference on Sustainable Development (Rio + 20). The AI-based solutions can be useful to realize them.
The AI can be also used to enable smart grid stability prediction. This possible usage was described by Ucar
[76]. He proposes an enhanced model using explainable AI and feature engineering for predicting the stability of the smart grid (SG). This model approaches the problem with both classification and regression, offering a holistic perspective on existing studies and proposing a novel structure to address their limitations. The GBM (gradient boosting machine) and deep learning models are introduced as effective tools for prediction, despite their drawbacks. The flexibility and practicality of GBMs make them valuable tools for model design and customization. The text concludes by emphasizing the importance of combining data analytics with smart grid research for future studies.
Summing up, artificial intelligence plays a pivotal role in optimizing the performance of smart grids by efficiently managing various grid operations and energy consumption. These AI-driven systems utilize load-forecasting techniques and predictive analysis to enhance stability, minimize downtime, and proactively maintain grid equipment, preventing costly outages. Moreover, AI facilitates real-time adjustments in energy consumption by communicating with smart appliances and electric vehicles, ensuring optimization during peak demand periods, and consequently reducing stress on the grid. AI’s integration within the energy market enables grid operators to coordinate distributed energy resources efficiently, enhancing grid reliability and managing distributed generation and storage capacity effectively. Furthermore, AI-driven decentralized controllers enhance system resilience and processing speed, optimizing operations among distributed units, and reducing the dependency on centralized controllers. Moreover, AI’s role extends to improving grid security by continuously monitoring and predicting potential threats, automating decision-making, and fostering collaboration among infrastructure components in smart cities.
5. Energy Storages
With the global shift toward renewable energy sources like solar and wind, the need for efficient and reliable energy storage solutions has become increasingly critical. AI plays an important role in addressing the challenges associated with energy storage, making it smarter, more cost-effective, and environmentally friendly
[77].
Energy storage technology has a role to play in enhancing the capabilities for utilizing new energy sources, ensuring the reliable and cost-effective power systems operation, and advancing the extensive adoption of renewable energy technologies
[78]. Various fresh innovations, concepts, methodologies, and technologies have been introduced in this domain, stemming from disciplines such as materials science, knowledge management, electrical engineering, control systems, and artificial intelligence
[79].
AI algorithms are being used to enhance the performance of energy storage systems, particularly lithium-ion batteries. By continuously monitoring and analyzing data from these batteries, AI can optimize their charging and discharging cycles, extending their lifespan and improving their efficiency
[54]. This not only reduces maintenance costs, but also reduces the environmental impact of battery disposal. This solution is employed to predict potential issues in energy storage systems before they lead to costly breakdowns
[80]. Through real-time data analysis and machine learning models, AI can detect anomalies in system behavior, enabling operators to perform timely maintenance and prevent unexpected downtime
[81].
Energy storage systems equipped with AI can respond rapidly to fluctuations in the grid. When the supply of renewable energy is inconsistent, AI can instantly adjust the flow of stored energy, stabilizing the grid and ensuring a consistent power supply
[82].
In the literature, many techniques of AI usage in energy storage can be found. Ahmed and Abdallia
[83] proposed hybrid differential evolution optimization of AI. The efficiency of the proposed controller is confirmed in an electrical grid that includes a synchronous generator, a photovoltaic power source, and a battery energy storage system. The controller’s parameters are adaptively tuned in real-time by training the artificial neural network (ANN) with datasets generated during the optimization phase of both controllers using the hybrid differential evolution optimization method under varying levels of disturbance, ranging from low to high. Athari and Ardehali
[84] used the fuzzy logic controller-based approach. The membership features of the fuzzy logic controller (FLC) are tailored to reduce operational costs in green energy hybrid systems. This reduction is achieved by utilizing weekly and periodic data predictions for factors such as water availability, electricity demand, and environmental conditions like wind speed, sunlight, and air temperature. This optimization process employs algorithms inspired by frog-spring shuffling. It is worth noting that accurate accounting of power grid costs plays a significant role in enhancing the efficiency of energy storage components for the hybrid renewable energy systems (HRESs) when connected to the grid. This efficiency improvement is achieved because the configured weekly and periodic FLCs help minimize the operating hours of fuel cells and gas-based generators while reducing state-of-charge (SOC) variability in the battery stack
[83].
Zahedi and Ardehali
[85] described the situation when a novel energy storage system (ESS) control system employing a multi-agent setup was implemented for a 100-megawatt system. The system’s control performance was verified through simulation analysis and practical testing. The AI-driven solution based on hierarchical control was described by Yunhao et al.
[86]. By employing balance regulation, the simulated impedance is dynamically adjusted to eliminate the impact of inaccurate line impedance on the precision of the current distribution. Subsequently, each power storage unit can fine-tune its current based on state-of-charge (SoC) balance control, taking into account its capacity and charging status. This helps reduce SoC discrepancies and facilitates a gradual state of charge (SoC) balance during both charging and discharging operations.
Summing up, artificial intelligence is revolutionizing energy storage solutions by optimizing the performance and longevity of storage systems. In energy storage, AI algorithms continuously analyze and fine-tune the charging and discharging cycles, notably enhancing the efficiency of lithium-ion batteries. By leveraging real-time data and predictive analytics, AI predicts potential system issues, enabling proactive maintenance, reducing downtime, and mitigating the environmental impact associated with battery disposal. These AI-driven solutions in energy storage effectively stabilize the grid by swiftly responding to fluctuations in renewable energy supply, ensuring consistent power flow and minimizing interruptions. Additionally, diverse AI-based approaches, such as hybrid optimization and fuzzy logic controllers, significantly improve system efficiency and reduce operational costs in hybrid renewable energy systems. These advancements underscore AI’s role in enhancing the reliability, efficiency, and sustainability of energy storage systems, offering promising avenues for smarter and more eco-friendly energy management.