Keynote Speakers

Keynote Speaker: Yo-Ping Huang 

 Title: "Applications of AIoT to Green Energy Industry"

Abstract:

The growth of renewable energy has become a major focus worldwide due to various factors such as environmental concerns about global warming, abnormal weather conditions, and the depletion of fossil fuels. Among the different types of renewable energy sources, solar energy has gained significant attention due to its abundance, sustainability, and lack of pollution. Most countries have set targets of increasing the share of renewable energy in total electricity generation to a certain percentage by 2030. As part of this plan, the production of solar energy needs to be raised from the earlier goals to new levels. Therefore, obtaining accurate information regarding the planning, monitoring, and technical aspects of photovoltaic (PV) power plants is essential to improve their performance.

Aquaculture is the fastest-growing food production sector around the world contributing approximately 50% of animal protein for half of the world’s population. According to the Food and Agriculture Organization (FAO), it predicts that aquaculture production will rise to 53% by 2030.

Applications of AI models, drone technologies, and systems play important roles and can be found everywhere, including widespread usage in industry. Furthermore, AI can be integrated with other techniques, such as Internet of Things (IoT), control methods, and edge computing to become powerful tools for industry and medical domains. This talk will focus on addressing the applications of AIoT to green energy industry.

Keynote Speaker: Emanuele Giovanni Carlo Ogliari 

 Title: "Machine Learning for Renewable Energy: Improving Solar Radiation Nowcasting"

Renewable energy sources, especially photovoltaic (PV), are essential to meet growing energy demands, but their variability poses significant challenges to electrical grid operations. Advanced forecasting techniques are necessary to mitigate these challenges and ensure grid stability. This keynote presents a novel approach to solar radiation nowcasting using an Enhanced Convolutional Neural Network (ECNN). By integrating infrared all-sky images with exogenous parameters, the ECNN model achieves higher accuracy in predicting short-term solar radiation, especially during periods of high variability. This advancement enables better management of PV systems, facilitates increased adoption of solar energy, and supports the transition to a more sustainable energy future. 

Development of predictive multiphysical models of Architected Materials and metamaterials 

Jean-François Ganghoffer. LEM3 – CNRS, Université de Lorraine, France

Predicting the effective multiphysical properties of architected materials (AM in short) is crucial because these materials derive their unique performance not just from their composition but from their engineered micro- and macro-structures. AM exhibit tunable properties (mechanical, thermal, electrical, acoustic, magnetic) based on their internal architecture. Accurate prediction helps optimize their design for specific applications, such as maximizing strength-to-weight ratios, enhancing thermal insulation, or improving energy absorption. Many modern applications require materials to perform across multiple domains, such as mechanical strength, thermal insulation, or electric conductivity. Predictive modeling ensures these materials can be tailored to meet multi-functional requirements, such as load-bearing structures also providing thermal and acoustic insulation. Moreover, predictive models enable rapid virtual prototyping, guiding material selection and design before fabrication, thereby reducing trial-and-error approaches. By accurately predicting the effective properties, engineers can minimize material usage while maintaining performance, leading to lightweight, high-efficiency structures with lower environmental impact. This is particularly valuable for green buildings, aerospace, and biomedical applications. With the rise of 3D printing and topology optimization, predicting effective properties ensures that printed materials meet design specifications before fabrication, preventing material failures and optimizing production. AM are often used in high added value applications like aerospace/aeronautics (3D interlocks), biomedical implants (soft or hard AM can be designed), civil engineering (mitigation of vibrations/ noise) or energy harvesting (use of auxetics). Accurate predictions of their effective properties help ensure reliability under real-world loading, thermal, and environmental conditions, preventing failures and improving safety. Predictive models, combined with Bayesian inference and machine learning, enable inverse design, where optimal material architectures can be discovered based on desired properties. This accelerates the development of next-generation AM. In this talk, we will focus on the elaboration of effective models of architected materials and metamaterials recoursing to homogenization and topology optimization methods in the static or dynamic range. The talk highlights specific features of inner architectures like chirality effects or the lack of centrosymmetry, requiring the development and identification of enriched continuum models of Cosserat or micromorphic types. A portfolio of applications in civil engineering, aeronautics (3D interlocks), bone biomechanics (biobased AM, like chitosan, mimick bone microstructure and properties, making them ideal biosubstitutes) will be exposed to illustrate some of the abovementioned key aspects of AM, as well as incursion into multiphysical couplings like piezoelectricity, flexoelectricity and magnetoelasticity. Ongoing research challenges will be mentioned to conclude the presentation.