Filter results

48188 results

PowerWeb Lecture: “Combining Deep Reinforcement Learning and Mathematical Optimization for Scalable Decision Making in Power Distribution Systems”

PowerWeb Lecture: “Combining Deep Reinforcement Learning and Mathematical Optimization for Scalable Decision Making in Power Distribution Systems” 30 May 2023 12:45 till 13:30 | Add to my calendar By: Masood Parvania, Associate Professor and Director of Utah Energy and Power Innovation Center (U-EPIC), University of Utah. Date: Tuesday 30 May 2023 Time: 12:45-13:30 (free lunch from 12:15) Location: Faculty of EWI, Mekelweg 4 (Chip Hall) Moderator: Dr Francesco Lombardi Register here : registration form . Abstract : The operation of power distribution systems has grown from passive and manual applications to include making complex decisions in real-time to facilitate the operation of large number of distributed energy resources for various applications. The limitations of mathematical optimization techniques in making scalable real-time decisions under uncertainty, has brought attention to techniques such as deep reinforcement learning (DRL) that has the potential to revolutionize the operation and enable further automation in power distribution systems. This talk will introduce a model that combines DRL with mathematical optimization for real-time operation applications in power distribution systems. This model brings the best of DRL and mathematical optimization, enabling the scalable and automated decision making in real-time while ensuring the physical feasibility of decisions in power distribution systems Short bio of the presenter : Dr. Masood Parvania is the Director of Utah Energy and Power Innovation Center, and Associate Professor of Electrical and Computer Engineering, at the University of Utah. His research interests include the operation, economics and resilience of power and energy systems, and modeling and operation of interdependent critical infrastructures. Dr. Parvania serves as an Associate Editor for the IEEE Transactions on Power Systems and the IEEE Power Engineering Letters.

ELLIS Delft Talk: Manuele Bicego and Michael Biehl

ELLIS Delft Talk: Manuele Bicego and Michael Biehl 04 May 2023 14:00 till 15:30 - Location: Hybride: Lecture Hall Boole (36.HB.T0.610) & Zoom - By: ELLIS Unit Delft | Add to my calendar On Thursday, 4 May the ELLIS Delft Unit is hosting an ELLIS Delft Talk in Lecture Hall Boole with two invited speakers. Manuele Bicego (Università degli Studi di Verona) will talk about ‘Random Forests beyond classification and regression’. Random Forests (RFs) represent a well-known data description tool for Pattern Recognition and Data Mining. RFs are mainly designed to face supervised tasks like classification and regression. However, there has been an increased interest in investigating their exploitation also in unsupervised scenarios like density estimation, clustering, anomaly detection, manifold learning and others. This talk is aimed at describing some of these unsupervised exploitations. Michael Biehl (University of Groningen) will talk about ‘The statistical physics of learning: Phase transitions in layered neural networks and the role of the activation function’. A particularly important aspect of designing neural networks is the choice of activation functions, which define the response of individual neurons to their actual input. In particular in deep learning, a variety of alternatives to the classical sigmoidal activations have been suggested, including the very popular ‘rectified linear unit’ (ReLU) and its modifications. If you're not able to join this session in person, follow link to Zoom here

Half Height Horizontal

TU Delft jointly wins XPRIZE Rainforest drone competition in Brazil

TU Delft jointly wins in the XPRIZE Rainforest competition in the Amazon, Brazil Imagine using rapid and autonomous robot technology for research into the green and humid lungs of our planet; our global rainforests. Drones that autonomously deploy eDNA samplers and canopy rafts uncover the rich biodiversity of these complex ecosystems while revealing the effects of human activity on nature and climate change. On November 15, 2024, after five years of intensive research and competition, the ETHBiodivX team, which included TU Delft Aerospace researchers Salua Hamaza and Georg Strunck, achieved an outstanding milestone: winning the XPRIZE Rainforest Bonus Prize for outstanding effort in co-developing inclusive technology for nature conservation. The goal: create automated technology and methods to gain near real-time insights about biodiversity – providing necessary data that can inform conservation action and policy, support sustainable bioeconomies, and empower Indigenous Peoples and local communities who are the primary protectors and knowledge holders of the planet’s tropical rainforests. The ETHBiodivX team, made of experts in Robotics, eDNA, and Data Insights, is tackling the massive challenge of automating and streamlining the way we monitor ecosystems. Leading the Robotics division, a collaboration between TU Delft’s Prof. Salua Hamaza, ETH Zurich’s Prof. Stefano Mintchev and Aarhus University’s Profs. Claus Melvad and Toke Thomas Høye, is developing cutting-edge robotic solutions to gather ecology and biology data autonomously. “We faced the immense challenge of deploying robots in the wild -- and not just any outdoor environment but one of the most demanding and uncharted: the wet rainforests. This required extraordinary efforts to ensure robustness and reliability, pushing the boundaries of what the hardware could achieve for autonomous data collection of images, sounds, and eDNA, in the Amazon” says prof. Hamaza. “Ultimately, this technology will be available to Indigenous communities as a tool to better understand the forest's ongoing changes in biodiversity, which provide essential resources as food and shelter to the locals.” . . . .

Students Amos Yusuf, Mick Dam & Bas Brouwer winners of Mekel Prize 2024

Master students Amos Yusuf, from the ME faculty (Mick Dam, from the EEMCS faculty and graduate Bas Brouwer have won the Mekel Prize 2024 for the best extra scientific activity at TU Delft: the development of an initiative that brings master students into the classroom teaching sciences to the younger generations. The prize was ceremonially awarded by prof Tim van den Hagen on 13 November after the Van Hasselt Lecture at the Prinsenhof, Delft. They received a statue of Professor Jan Mekel and 1.500,- to spend on their project. Insights into climate change are being openly doubted. Funding for important educational efforts and research are being withdrawn. Short clips – so called “reels” – on Youtube and TikTok threaten to simplify complex political and social problems. AI fakes befuddle what is true and what is not. The voices of science that contribute to those discussion with modesty, careful argument and scepticism, are drowned in noise. This poses a threat for universities like TU Delft, who strive to increase student numbers, who benefit from diverse student populations and aim to pass on their knowledge and scientific virtues to the next generation. It is, therefore, alarming that student enrolments to Bachelor and Master Programs at TU Delft have declined in the past year. Students in front of the class The project is aimed to make the sciences more appealing to the next generation. They have identified the problem that students tend miss out on the opportunity of entering a higher education trajectory in the Beta sciences – because they have a wrong picture of such education. In their mind, they depict it as boring and dry. In his pilot lecture at the Stanislas VMBO in Delft, Amos Yusuf has successfully challenged this image. He shared his enthusiasm for the field of robotics and presented himself as a positive role model to the pupils. And in return the excitement of the high school students is palpable in the videos and pictures from the day. The spark of science fills their eyes. Bas Brouwer Mick Dam are the founders of NUVO – the platform that facilitates the engagement of Master Students in high school education in Delft Their efforts offer TU Delft Master Students a valuable learning moment: By sharing insights from their fields with pupils at high school in an educational setting, our students can find identify their own misunderstandings of their subject, learn to speak in front of non-scientific audiences and peak into education as a work field they themselves might not have considered. An extraordinary commitment According to the Mekel jury, the project scored well on all the criteria (risk mitigation, inclusiveness, transparency and societal relevance). However, it was the extraordinary commitment of Amos who was fully immersed during his Master Project and the efforts of Brouwer and Dam that brought together teaching and research which is integral to academic culture that made the project stand out. About the Mekel Prize The Mekel Prize will be awarded to the most socially responsible research project or extra-scientific activity (e.g. founding of an NGO or organization, an initiative or realization of an event or other impactful project) by an employee or group of employees of TU Delft – projects that showcase in an outstanding fashion that they have been committed from the beginning to relevant moral and societal values and have been aware of and tried to mitigate as much as possible in innovative ways the risks involved in their research. The award recognizes such efforts and wants to encourage the responsible development of science and technology at TU Delft in the future. For furthermore information About the project: https://www.de-nuvo.nl/video-robotica-pilot/ About the Mekel Prize: https://www.tudelft.nl/en/tpm/our-faculty/departments/values-technology-and-innovation/sections/ethics-philosophy-of-technology/mekel-prize

New catheter technology promises safer and more efficient treatment of blood vessels

Each year, more than 200 million catheters are used worldwide to treat vascular diseases, including heart disease and artery stenosis. When navigating into blood vessels, friction between the catheter and the vessel wall can cause major complications. With a new innovative catheter technology, Mostafa Atalla and colleagues can change the friction from having grip to completely slippery with the flick of a switch. Their design improves the safety and efficiency of endovascular procedures. The findings have been published in IEEE. Catheter with variable friction The prototype of the new catheter features advanced friction control modules to precisely control the friction between the catheter and the vessel wall. The friction is modulated via ultrasonic vibrations, which overpressure the thin fluid layer. This innovative variable friction technology makes it possible to switch between low friction for smooth navigation through the vessel and high friction for optimal stability during the procedure. In a proof-of-concept, Atalla and his team show that the prototype significantly reduces friction, averaging 60% on rigid surfaces and 11% on soft surfaces. Experiments on animal aortic tissue confirm the promising results of this technology and its potential for medical applications. Fully assembled catheters The researchers tested the prototype during friction experiments on different tissue types. They are also investigating how the technology can be applied to other procedures, such as bowel interventions. More information Publicatie DOI : 10.1109/TMRB.2024.3464672 Toward Variable-Friction Catheters Using Ultrasonic Lubrication | IEEE Journals & Magazine | IEEE Xplore Mostafa Atalla: m.a.a.atalla@tudelft.nl Aimee Sakes: a.sakes@tudelft.nl Michaël Wiertlewski: m.wiertlewski@tudelft.nl Would you like to know more and/or attend a demonstration of the prototype please contact me: Fien Bosman, press officer Health TU Delft: f.j.bosman@tudelft.nl/ 0624953733