Foundations of active noise control
The course wants to introduce the bases and fundamentals of active noise control.
The following topics will be discussed: adaptive filters; broadband and narrowband, single-channel and multi-channel, feedforward active noise control systems; feedback and hybrid active noise control systems; secondary channel estimation.
Alberto Carini
28/05/2024 (10-13)
29/05/2024 (10-13)
30/05/2024 (10-13)
Non destructive testing with X-ray Computed Tomography
The principles of Non-destructive Testing (NDT) with X-ray Computed Tomography (CT) will be presented. X-ray CT is well known in the medical field but it is also widely exploited in industrial applications, such as e.g. airport security or automotive industry for actual/nominal comparison.
The course is suggested to PhD students in engineering (both information and industrial engineering) as well as to students of the PhD program in nanotechnology and physics.
Francesco Brun
Evolutionary Robotics
Foundations of Evolutionary Computation (EC). Topics.
High-level working scheme of an Evolutionary Algorithm (EA); terminology.
Generational model; selection criteria; exploration/exploitation trade-off; genetic operators with examples; fitness function; multi-objective optimization and Pareto dominance; debugging of an evolutionary search; EA issues (diversity, variational inheritance, expressiveness); fitness landscape.
Examples of common EAs: GA, GP, GE. Brief foundations of Artificial Neural Networks and EC
EA for neuroevolution Significant examples ○ Evolution of Soft Robots morphologies (body)
Evolution of robotic agents controllers (brain)
Simultaneous evolution of body and brain Simulation: tools and benchmark tasks
Eric Medvet
12/02/2024 9-12 in aula A, secondo piano, edificio C2
13/02/2024 9-12 in aula A, secondo piano, edificio C2
15/02/2024 9-12 in aula A, secondo piano, edificio C2
16/02/2024 9-12 in aula A, secondo piano, edificio C2
Formal methods for identifying malicious behavior
The objective of this course is to provide learners with an insight into the main cyber threats and provide a range of methodologies, from the use of deep learning to formal methods, applied to malware detection.
Furthermore, simulations of network attacks will be shown in order to identify and resolve any vulnerabilities and improve the security of computer systems. Real malware examples will be provided, through a case study related to the Android operating system, in order to study its behavior.
Francesco Mercaldo
13/01/2024 (10-13)
17/01/2024 (10-13)
20/01/2024 (11-13)
DC ship power systems: evolution & research challenges
Ships power system evolution.
Shipboard electrical applications (Integrated Power Systems).
MVDC power systems on ships.
Integrated Electrical/Electronics ships Power Systems design (methods and tools).
Integrated Power & Energy Systems Dependability on ships.
Analysis, evaluation and re-design of different types of shipboard power systems
Giorgio Sulligoi
Introduction to Uncertainty Quantification in Computational Science and Engineering
Computer simulations have become a useful tool for the mathematical modeling of many complex physical systems in different fields, such as engineering, physics, chemistry and a lot of others.
Most of computer simulations perform numerical solution of deterministic differential equations (either partial or ordinary). Such deterministic models however differ from reality due to the inevitable presence of uncertainty in the systems. In order to take uncertainty into account in the model, when studying complex systems, Uncertainty Quantification (UQ) techniques are introduced.
The UQ is a process that aims at quantitatively describing the origin, characterization, and propagation of different sources of uncertainty in complex systems. This course will focus on the solution of so-called forward problems, where the uncertainty in the input parameters is propagated through the model to give information about uncertain outputs, providing an overview of the most popular UQ techniques.
Lucia Parussini
14/03/2024 (9-12)
15/03/2024 (9-12)
18/03/2024 (10-12)
Artificial Intelligence
Course topics Artificial Intelligence and Agents Searching for Solutions Reasoning with Constraints Local Search and Optimization
Enrico Franconi
27/02/2024 (15-17)
28/02/2024 (15-17)
29/02/2024 (15-17)
01/03/2024 (15-17)
Experimental methods for fuel cells characterization
The lecture will start from the fundamental electrochemistry and thermodynamics, with emphasis on the performance of fuel cells systems.
The objective of the course is to give the students a solid foundation upon which they will be able to experimentally asses the performance of single cell, stack and fuel cells power plant. Contents Basic Electrochemical Principles Basic Thermodynamics of Fuel Cell Systems Polarization Curve Experimental characterization of single cell, stack and fuel cell systems
Rodolfo Taccani
An overview on linear optimization
Mathematical optimisation Definitions, history and context Applications to the real world Simple examples
Formulation of very popular combinatorial optimisation problems - Knapsack problem - Assignment problem - (Perfect) matching problem - Set covering, set packing, and set partitioning problem - Travelling salesman problem
An exact algorithm for integer linear programming - Brief introduction to continuous linear programming - Bounds and linear relaxation - The Branch & Bound algorithm
Heuristic algorithms for integer linear programming - Greedy algorithms - Local search algorithms - Approximate algorithms and their applications to the Knapsack problem and to the Travelling salesman problem
Lorenzo Castelli
Soft electro-active materials
Fundamentals of electrostatics
- Dielectric elastomers
- Linear and nonlinear electroelasticity
- Instabilities
- Applications: soft capacitors and other electroelastic devices
Massimiliano Gei
Systems Engineering: design processes and methods
Systems Engineering: definitions and context.
Main references for the application of Systems Engineering theories: INCOSE Handbook, NASA Handbook.
The product development cycle: from concept generation to engineering design.
Design methods: the axiomatic design.
The TRIZ methodology for systematic innovation.
Design for assembly, reliability, maintainability, availability and safety (RAMS).
Virtual Reality simulation in the systems engineering context: prototyping from CAD/CAE to physical mock-up.
Domenico Marzullo
Computer-aided design in product lifecycle management
The course focuses on modern product lifecycle management methodologies and advanced computer aided design (CAD) techniques within a PLM system.
PLM is an integrated management methodology, based on collaborative tools, aimed at managing and sharing the product data guaranteeing their reliability and consistency at every stage of the product lifecycle.
Advanced aspects related to the associativity of product data and the management of product skeletons with the CATIA V5 software by Dassault systems are explored.
Andrea Tarallo
Concept design based on MBSE approach supported by MATLAB
The course is designed to offer knowledge, tools, and methods for concept design of complex systems.
In the first part, the course introduces the principles of Model-Based Systems Engineering (MBSE) and highlights its advantages compared to traditional Systems Engineering.
Subsequently, the course introduces the main stages of the RFLP method and focuses on the structured Natural Language for writing well-formed requirements. By the end of the first module, students will have the ability to: (i) write well-formed individual requirements, (ii) specify a consistent and coherent set of requirements, (iii) use Simulink Requirements for organising the whole requirement list.
Andrea Rega
Graphical Interfaces of Matlab
The course introduces students to the development and implementation of software applications equipped with Graphical User Interfaces (GUI) by providing the skills and procedures for MATLAB application programming.
Ferdinando Vitolo
Some features of industrial fluid dynamics
Students will be introduced to some cases of applied research in the framework of industrial fluid dynamics.
The course will focus on how to approach a fluid dynamic problem critically, identifying the main physical features and defining a proper modeling approach, this procedure will take advantage of real applicative cases. First it will be analyzed a manyfolds, within a drying hood, with the aim of an equal flow distribution. The system will be analyzed analytically, by the use of the energy equation and of the momentum conservation equation to then test the findings numerically.
A second test case will be that of an air purification system for indoor air quality, here a mixed Eulerian-Lagrangian approach will be used. In both the problems a particular attention will be used on how to treat turbulence.
Federico Roman
Bayesian Optimization and Reinforcement Learning
Four modules. Every module includes a Guided lab session with jupyter notebook. First module (2h)
Background: normal distribution (univariate and multivariate), expectation, variance, covariance, conditional distributions.
Definition of Gaussian processes.
Gaussian process regression.
Hyperparameter optimization.
Motivation and applications to Gaussian process regression.
Limitations. Second module (2h)
Introduction to black box optimization, global optimization, and surrogate modeling.
Brief recap of Gaussian process regression.
Concept of utility functions.
Concept of exploration vs exploitation in Bayesian optimization.
Acquisition functions and its properties, e.g, Upper Confidence Bound (UCB), Expected Improvement (EI), etc.
Motivation and applications of Bayesian optimization.
Limitations.
Advanced topics: constrained Bayesian optimization, multi-objective Bayesian optimization, preferential Bayesian optimization. Third module
What is Reinforcement Learning?
Tabular approach: Multi-armed Bandits
Markov Decision Process, environment, state, agent and reward
Policy and Value Functions: optimality and approximation Fourth module (2h)
Dynamic Programming: Policy and Value Iteration
Temporal Difference Learning
Juan Ungredda - Simone Silvetti
Learning-based Controllers and the Reality Gap
Foundations of Learning-Based (LB) Control.
High-level formulation of Learning-Based control schemes.
Terminology. – Reinforcement Learning (RL) and Optimal Control.
Examples of common RL tasks. • Introduction to the Reality gap (RG) problem.
High-level formulation of RG in general LB control schemes.
RG in RL.
Domain Randomization.
Adversarial RL.
Transfer Learning.
Practical examples of RG.
Erica Salvato
07/05/2024 (14-16)
09/05/2024 (14-16)
14/05/2024 (14-16)
16/05/2024 (14-16)
21/05/2024 (14-16)
23/05/2024 (14-16)
State estimation and fault detection: a set-based approach
Reachability analysis - an overview
Constrained Zonotopes: description and main features
Set-based state estimation
Set-based fault isolation
Applications
Davide Martino Raimondo
