In recent decades, with the rapid development of the economy, electrical power demand has significantly increased, and the per capita electricity consumption has been much greater than in any previous cultures throughout history. To this end, massive fossil fuels (crude oil, natural gas, and coal) are utilized in gas- and coal-fired power plants for electrical power generation, which leads to uncontrolled carbon emissions and the greenhouse effect. Recent studies show that if the global temperature continues to rise with the following current trends and no other strategies are adopted, it will cause losses of up to 10% of the global economy by the middle of this century [1]. To address it, more and more governments are paying attention to sustainable development and combating climate change. In this context, due to the clean, low carbon emission, and recycled nature characteristic of renewable energy sources, it is considered a foremost promising alternative to traditional fossil fuels in the transition toward a low-carbon economy. Specifically, renewable energy sources like wind, solar, and hydropower have been widely utilized for electrical power generation, as shown in Figures 1.1 and 1.2.

The integration of artificial intelligence (AI) into power electronics represents a transformative leap forward in the field of electrical engineering. Power electronics, a critical component of modern electrical systems, is responsible for the efficient conversion and control of electrical energy, serving as the backbone of various applications, from renewable energy systems to electric vehicles and smart grids. The traditional paradigms in power electronics have primarily relied on deterministic control techniques and rule-based algorithms to optimize energy conversion and distribution. However, the ever-increasing complexity and dynamic nature of today's electrical grids and energy systems demand a more adaptive, intelligent, and responsive approach. The introduction of AI into power electronics signifies a pivotal shift towards enhancing the efficiency, reliability, and sustainability of electrical systems. AI, powered by machine learning algorithms, neural networks, and advanced data analytics, empowers power electronics devices to operate with unprecedented levels of sophistication. This integration allows power converters, inverters, and other key components to learn from real-time data, anticipate fluctuations in power demand, and make dynamic adjustments, optimizing energy conversion and distribution in ways that were once inconceivable.

In this chapter, two network partition-based decentralized voltage control methods are proposed. The conclusions are summarized as follows:

A distributed coordination control method is first proposed for distribution system Volt-VAR control, considering both PV inverters and SVCs. The spectral clustering algorithm allows for a partition of the large distribution system into several sub-networks from a voltage control perspective. Then, the control of each sub-network is formulated as the MGs and solved by an attention-based MATD3 algorithm. The proposed method is centralized-trained and distributed-implemented and can easily be used for real-time voltage regulation. Compared with centralized control, the proposed approach mitigates issues, such as communication bottlenecks and privacy concerns. Compared with other distributed control methods, only local information is needed without communications between agents. The proposed method can be adapted to the flexible network partition requirements of the operator rather than the typical MADRL algorithm. Comparative results with several other existing model-based and data-driven methods demonstrate that the proposed method can achieve 96.4% optimality based on local information while also considering uncertainties. However, the model-based approach could not achieve satisfactory outcomes in the case of rapid variations in PV outputs. A novel MADRL framework is introduced for voltage regulation in distribution systems with a high PV penetration, featuring centralized training and decentralized execution without physical models. The SPGP is first leveraged to build a surrogate model that learns the mapping relationship between the active and reactive power injections and the voltage magnitude of each node using few-shot recorded data. This surrogate model is further integrated with the MADRL to assist in the formulation of a coordinated control strategy. In particular, the voltage regulation problem is cast into the MADRL framework by partitioning the whole network into several sub-regions, considering both the regional voltage regulation ability and the electrical distance. Each sub-region is treated as an agent and solved by the DRL algorithm. All agents are trained in a centralized framework to learn the coordinated control strategy guided by the reward given by the surrogate model. The proposed method can achieve real-time scheduling using only local information. Comparative results demonstrate that: (1) the proposed decentralized control strategy can achieve similar performance as the centralized one; (2) the performance of the suggested model-free approach closely matches that of one relying on a flawless physical model; (3) the control strategy can be taken in real-time with the usage of the battery storage system to mitigate the influence of violent PV fluctuations.

The future works include the development of a new control method that can coordinate the inverters and utility-owned equipment, which is a two-timescale control problem. We will also propose a meta-learning-based MADRL algorithm to deal with topology changes in the distribution networks.

Fuel scarcity and escalating carbon emissions pose a dual challenge to conventional energy sources. Photovoltaic (PV) solar and wind energy systems serve as renewable substitutes for traditional energy systems due to their eco-friendliness, versatility, and green nature. Furthermore, regulations adopted by numerous governments encouraging the installation of PV and wind power systems have reduced the cost of electricity production. The development of renewable energy sources, such as solar, wind, fuel cells, and biomass, has enabled the current electric power industry to advance technologically. A smart grid is a digital technology that integrates various microgrids with the grid and monitors them with appropriate management and control to reduce or eliminate power quality concerns. The likelihood of individual microgrid stability criteria being met is increased by interconnected microgrids. Therefore, self-sustaining smart grid technology is necessary to decrease carbon emissions and provide an energy management solution. The use of passive or active filters is one method to enhance the grid's power quality. Despite their higher cost, active filters are preferred due to their size and performance advantages. Utility grid-interfaced PV and wind energy system installations as active filters have gained popularity as a significant research topic worldwide since the turn of the millennium. The active filtering capabilities of smart grid systems connected with microgrids have received significant attention in the literature. Instantaneous reactive power theory (IRPT), synchronous reference frame theory (SRFT), and the improved linear sinusoidal tracer (ILST) are some of the methods used for managing shunt active filters (SAFs). These algorithms can reduce current harmonics on the source side and offer reactive power compensation when utilized to manage the voltage source inverter (VSI) in grid-connected systems.

A "smart microgrid" refers to an intelligent electricity distribution system that connects loads, distributed energy resources, and storage within clearly defined electrical boundaries, functioning as a single, controllable entity with respect to the main grid, according to India's Model Smart Grid Regulations. Microgrids (MGs) are small-scale power plants that can operate in both grid-connected and island modes, boasting high levels of energy security, dependability, storage capacity, and economic efficiency for demand-side and load-side control. The electricity sector can self-recover by assessing and addressing issues, as they are capable of local control via automated methods. To reduce costs and return excess energy to local microgrids, the electric industry has been revolutionized through the use of solar and wind renewable energy sources (RESs) and smart grid technologies. Smart microgrids have the potential to integrate with the grid, but they are also self-sufficient and can service a local community without relying on centralized power systems. Based on comparable electrical system research, future generations will require intelligent controller-based smart grid systems, with many microgrids integrating with the grid without affecting the quality of electricity.

A smart microgrid system comprises one or more interconnected smart microgrids, facilitated by a robust controller that can be integrated into or operated independently from the main grid. This comprehensive system encompasses diverse renewable and non-renewable energy sources, along with load centers catering to residential, commercial, and industrial sectors. Real-time management of electric power systems necessitates data analytics and artificial intelligence (AI) techniques for statistical analysis of consumer energy usage data and weather forecasts pertaining to various RESs. Furthermore, the AI-based controller incorporates modules such as tariff control and power flow management while processing customer power data.

In summary, this book has focused on applying AI technology to deal with a series of challenges caused by the integration of renewable energy. Among them, we first overview the development trend of renewable energy in the power system and introduce several important AI technologies in Chapter 1. After that, Chapters 2-8 provide some examples of utilizing AI methods to address some issues like fault diagnosis, power electronic reliability, dual active bridge modulation, voltage control of an active distributed network, energy management, and stability-oriented control. Specifically, Chapter 2 analyzes the characteristics of electric machine faults, and applies the few-shot learning method and Gaussian-process-enabled method to do fault diagnosis. In Chapter 3, deep learning methods are utilized for power electronic monitoring, optimization, and control. Moreover, to improve energy conversion efficiency, Chapter 4 explores reinforcement learning-enabled modulation strategies for dual active bridges. In Chapter 5, the voltage fluctuation in an active distributed network caused by the integration of photovoltaic (PV) and wind turbine (WT) is considered, and a reinforcement learning multi-agent deep reinforcement learning (DRL) algorithm is used to solve it. Chapters 6 and 7 apply the DRL method for the energy management of hybrid energy systems and microgrids, respectively. Chapter 8 pays attention to the small signal stability of renewable energy systems, and the DRL-enabled controllers are designed to enhance the small signal stability of the systems.

Overall, the integration of renewable energy will contribute to reducing carbon emissions to alleviate the environmental energy crisis, but it increases the nonlinearity and uncertainty in the power system, which makes the operation, management, and control of such a system more difficult. Especially, most conventional approaches use physical model-based methods, which highly depend on accurate system parameters and topology. However, due to the large-scale, high-order, and nonlinear time-varying characteristics of the actual power grid, it is difficult to establish a full-scale, detailed model. With the use of phasor measurement units (PMU), a large amount of electricity data with high volumes, mutual correlations, and complex structures can be observed and saved, which can be applied for the extraction of experience and further used for the operation and planning of the power system with a renewable energy system. In this way, the disadvantages of the physical model-based method can be overcome. To this end, data-driven methods are proposed. Among them, the AI method can learn from historical data to construct an adaptive agent to deal with the newly encountered conditions. In recent years, AI has achieved rapid development in optimization and control problems, around power electronics, active distribution networks, and the main power network. In this context, this book has demonstrated some state-of-the-art applications of AI technology for renewable energy systems, and power electronics. The results of these studies indicate that AI can better deal with the uncertainties and intermittency of complex systems, which makes it demonstrate better control and optimization performance using renewable energy systems.