The rapid growth of cloud computing, artificial intelligence, and data-intensive applications has significantly increased the energy consumption of computing infrastructures. Modern servers and data centers require substantial electrical power not only for computation but also for cooling, networking, storage, and supporting infrastructure. Therefore, improving the energy efficiency of computing systems has become an important challenge for both sustainable computing and renewable energy integration.

This tutorial provides a comprehensive introduction to techniques for reducing the energy consumption and environmental impact of computers, servers, and data centers. It covers energy-efficient hardware and software design, dynamic power management, virtualization and containerization, workload consolidation, energy-aware scheduling, and resource optimization.

Special attention will be given to the increasing energy requirements of artificial intelligence workloads, including GPU-based computing, AI model training, and inference. The tutorial will also demonstrate how artificial intelligence itself can be employed to predict workloads and optimize server resource allocation and energy consumption.

Finally, the tutorial introduces renewable-aware and carbon-aware computing, where computational workloads are dynamically scheduled according to renewable energy availability, electricity demand, and carbon intensity. Practical examples and case studies will demonstrate how computing workloads can be shifted in time or location to improve energy efficiency and increase the utilization of renewable energy.

The tutorial aims to provide participants with both conceptual foundations and practical approaches for designing more energy-efficient, intelligent, and sustainable computing infrastructures.

DC–DC converters based on differential connections have attracted increasing interest for applications requiring flexible voltage conversion, improved modularity, and enhanced power-processing capability. In these configurations, two or more converter modules are interconnected through differential input or output connections, enabling voltage and current sharing while allowing the overall system to achieve operating characteristics that are difficult to obtain with a single converter.

This presentation provides an overview of the operating principles, main topological arrangements, and control requirements of differential-connected DC–DC converters. Particular attention is given to voltage conversion ratio, power distribution between modules, semiconductor voltage stress, efficiency, dynamic response, and the influence of parameter mismatches.

A particularly relevant feature of some differential-connected configurations is the possibility of partial power processing. In this operating mode, the converter processes only a fraction of the total load power, typically corresponding to the voltage or power difference between the source and load, while the remaining power is transferred directly through an appropriate connection path. Consequently, the processed power, semiconductor ratings, conduction losses, and converter volume can be reduced compared with conventional full-power-processing architectures.

The challenges associated with balancing currents, maintaining stable operation, and coordinating the switching actions of the interconnected converters are also discussed. Finally, the advantages and limitations of these converters are discussed in terms of efficiency, voltage gain, current sharing, isolation, regulation range, and fault tolerance.

Power systems have evolved over more than a century from relatively small and isolated networks into large, interconnected systems designed around centralized generation and synchronous machines. Their reliable operation has traditionally relied on well-established principles of generation–demand balance, frequency regulation, voltage control, reactive power management, and system stability. However, the growing integration of renewable energy resources and power-electronic-interfaced generation is gradually changing the fundamental characteristics and dynamic behaviour of modern power systems.

This tutorial begins with the historical development, basic structure, operating principles, and key dynamics of conventional power grids, with particular emphasis on frequency and voltage stability and their associated control mechanisms.

It then explores the transition toward renewable-rich power systems and discusses the technical challenges introduced by variable generation, reduced system inertia, changing power flows, and increasing reliance on power electronic converters.

Finally, the tutorial introduces the role of smart grid technologies, advanced monitoring and control, energy storage, and flexible resources in addressing these emerging challenges and supporting the stable, reliable, and sustainable operation of future power systems.

The rapid expansion of renewable energy is transforming modern power systems while introducing significant technical challenges associated with intermittency, uncertainty, nonlinear characteristics, changing environmental conditions, and real-time operation. Artificial Intelligence (AI) has emerged as a powerful enabling technology for addressing these challenges through improved forecasting, monitoring, fault diagnosis, optimization, control, and decision-making.

This tutorial provides a technically oriented overview of AI applications in renewable energy systems, with particular emphasis on photovoltaic (PV) and wind energy conversion systems. The evolution from conventional machine learning and deep learning toward emerging approaches such as explainable AI, edge intelligence, generative AI, and agentic AI will be introduced in the context of practical renewable energy applications.

Key AI applications in PV systems, including solar irradiance and power forecasting, PV performance prediction, maximum power point tracking (MPPT), fault and anomaly detection, condition monitoring, predictive maintenance, and AI-assisted converter control, will be presented and discussed. Similarly, major AI applications in wind energy systems, such as wind speed and power forecasting, wind turbine performance assessment, maximum power extraction, fault diagnosis, condition monitoring, predictive maintenance, and intelligent control of wind energy conversion systems, will be addressed. The role of AI in energy storage, power electronic interfaces, and coordinated renewable energy management will also be discussed.

Particular emphasis will be placed on the progression from data-driven forecasting and condition monitoring toward intelligent optimization, adaptive control, and autonomous operation. The tutorial will demonstrate how advanced AI techniques can move beyond prediction to support real-time operational decisions, optimize energy conversion, adapt control strategies to changing environmental and operating conditions, and coordinate renewable generation and storage.

Finally, key challenges including data quality and availability, model generalization, explainability, computational requirements, real-time implementation, reliability, cybersecurity, and trustworthy AI will be discussed. Future research directions toward self-monitoring, self-optimizing, adaptive, and increasingly autonomous PV and wind energy systems will be highlighted.

The rapid growth of renewable energy technologies is transforming electrical power systems toward a cleaner and more sustainable future. However, photovoltaic installations, battery energy storage systems, power electronic converters, and high-voltage DC networks introduce new electrical stresses and failure mechanisms that may increase the risk of fire.

This tutorial explores the “other side” of renewable energy generation, focusing on the mechanisms that can initiate and sustain electrical fires. Particular attention is devoted to DC arc faults, insulation degradation, connector failures, hot spots and ground faults.

Through real failure scenarios and practical examples, the tutorial discusses how electrical fires develop, why conventional protection devices may sometimes be insufficient, and how monitoring, protection, proper installation, and fault diagnosis can reduce these risks.

The objective is not to question the safety or value of renewable energy technologies, but to understand the emerging risks associated with their large-scale deployment and how electrical engineering can address them.

The increasing demand for energy-efficient electrical drive systems has stimulated extensive research on advanced control strategies aimed at reducing power losses while maintaining satisfactory dynamic performance. This aspect is particularly relevant in variable-speed electrical drives, which frequently operate over wide speed and load ranges where conventional control strategies do not necessarily ensure maximum efficiency.

This tutorial provides a comprehensive overview of loss minimization techniques for electrical drives, starting from the fundamental relationship between machine operating conditions, electromagnetic torque, and the main loss components. The basic principles of loss minimization are initially introduced with particular reference to induction motor drives, where the trade-off between copper and iron losses determines an optimal magnetization level for each operating condition. The discussion is then extended to permanent-magnet synchronous motor drives, highlighting how the same general loss-minimization principle can be applied to different machine topologies through the appropriate selection of the optimization variables.

The main families of efficiency optimization strategies are then introduced and compared, including Loss Model Control, Search Control, and Hybrid Control approaches. Their operating principles, advantages, limitations, computational requirements, convergence properties, and dependence on machine parameters are discussed, together with the evolution of loss minimization techniques and their integration into modern drive control systems.

Special attention is devoted to Loss Model Algorithms and to the main challenges associated with their practical implementation, including magnetic saturation, parameter uncertainties, frequency-dependent iron losses, thermal variations, and real-time computational constraints.

Finally, the tutorial presents the evolution from conventional model-based approaches toward adaptive and system-level loss minimization. Experimental studies are discussed to illustrate how the optimal operating point may change with load, speed, and machine thermal state. A self-adaptive reticular loss model approach is then introduced to dynamically refine the minimum-loss operating point while remaining suitable for implementation on low-cost embedded controllers. The analysis is finally extended beyond the electrical machine by considering power-converter losses, addressing the more general question of whether efficiency optimization should target the electrical machine alone or the complete drive system.

Inverter-Based Resources represent a key technology to achieve carbon-free electrical energy production. Since they must comply with power quality standards, they require power filters to mitigate the harmonics which propagate into the grid. However, the optimal power filter design when Multilevel Inverters are considered, i.e. choosing the lowest filter parameters, resulting in the power quality standards compliance represents an open challenge in literature.

Moreover, since the control strategy has a great impact on the output voltage harmonic spectrum, novel inverter control techniques are currently investigated in literature. In addition, in literature, new inverter control strategies are emerging, named Grid-Forming, aim to manage the active and reactive power injection to fulfill the power grid requirements and are often based on the emulation of Synchronous Generators. However, in case of a grid short circuit, since the power electronics devices can withstand an overcurrent which is way less than the one of Synchronous Generators, current limitation methods of Grid-Forming inverters during these abnormal conditions are currently under study.

Given the above, the topics investigated in this thesis aim to fill these literature gaps and are the following: Optimal LCL filter design for grid-connected Cascaded H-Bridge Multilevel Inverters; Enhancing low switching frequency performance using Finite-Control-Set Model Predictive Control for grid-connected Cascaded H-Bridge Multilevel Inverters; Enhanced ride-through method for Grid-Forming Cascaded H-Bridge Multilevel Inverters under symmetrical and asymmetrical grid faults and phase jumps.

Time series play a central role in monitoring energy systems, where the early detection of anomalies and drifts is essential to operational reliability. This tutorial offers a structured overview of methods for anomaly detection, concept drift detection, and diagnostics, covering both unsupervised and supervised approaches.

The unsupervised part covers statistical methods (rolling z-score, control charts, EWMA, CUSUM), distance- and density-based approaches (k-nearest neighbors, Local Outlier Factor, Dynamic Time Warping), reconstruction (PCA, Autoencoder), prediction (ARIMA/SARIMA residuals), as well as Isolation Forest and One-Class SVM. Drift detection is addressed through ADWIN, Page-Hinkley, DDM/EDDM, and KSWIN. The supervised part covers anomaly classification and the metrics adapted to strong class imbalance (precision, recall, F1, AUC-PR).

Particular attention is paid to diagnostics: temporal localization of change points (PELT) and explainability (SHAP), turning a statistical alert into actionable information for the domain expert. All concepts are illustrated and put into practice through a reproducible, hands-on notebook built around a running case study, intended for PhD students and faculty.