Higher Institute on Territorial Systems for Innovation
PORTO@iris (Publications Open Repository TOrino - Politecnico di Torino)Not a member yet
146173 research outputs found
Sort by
Block 19a, New Belgrade. A Tale of Three Agencies
No full textThis visual essay examines the architectural and urban evolution of Block 19a in New Belgrade, a Late Socialist residential complex distinguished by its environmental sensitivity and departure from the rigid orthogonal grid of the surrounding Modernist landscape. Through a diachronic and highly illustrated approach—combining diagrams, drawings, archival materials, and field documentation—the essay traces the block’s development from the 1949 master plan and the 1975 competition to contemporary profit-driven transformations. Visual mappings and close-up studies reveal how residents, planners, and developers negotiate space through tactical, profit-led, and subversive “agencies.” The essay interprets Block 19a as a dynamic micro-environment where planning legacies, ecological intentions, and post-Socialist urban pressures intersect
Gradient-informed neural networks: Embedding prior beliefs for learning in low-data scenarios
No full textWe propose Gradient-Informed Neural Networks (GradINN s), a methodology that can be used to efficiently approximate a wide range of functions in low-data regimes, when only general prior beliefs are available, a condition that is often encountered in complex engineering problems. GradINN s incorporate prior beliefs about the first-order derivatives of the target function to constrain the behavior of its gradient, thus implicitly shaping it, without requiring explicit access to the target function's derivatives. This is achieved by using two Neural Networks: one modeling the target function and a second, auxiliary network expressing the prior beliefs about the first-order derivatives (e.g., smoothness, oscillations, etc.). A customized loss function enables the training of the first network while enforcing gradient constraints derived from the auxiliary network; at the same time, it allows these constraints to be relaxed in accordance with the training data. Numerical experiments demonstrate the advantages of GradINN s, particularly in low-data regimes, with results showing strong performance compared to standard Neural Networks across the tested scenarios, including synthetic benchmark functions and real-world engineering tasks
Fundamental Investigation of Electrochemical Urea Synthesis
No full textL'abstract è presente nell'allegato / the abstract is in the attachmen
Physics-informed identification of nonlinear dynamical systems - A unified framework combining domain knowledge and data-driven sparse modeling
No full textL'abstract è presente nell'allegato / the abstract is in the attachmen
Parallel Computation of the Nonlinear Forced Response of a Bladed Disk With Friction Contacts Using the FETI Method
No full textThe size of the finite element (FE) models used for turbomachinery applications is rapidly growing to consider the whole engine dynamics and the complex dynamic interactions between subassemblies, which are connected by means of removable friction contacts. In this context, the aim of this paper is the introduction of a parallel harmonic balance method (HBM) for the nonlinear forced response analysis of bladed disks with localized contact nonlinearities. The proposed technique is based on the finite element tearing and interconnecting (FETI) method, which requires the decomposition of the FE model into several domains coupled by interface forces that are mathematically modeled using the Lagrange multipliers method. This paper offers a comprehensive overview of the FETI method and its implementation for predicting the nonlinear forced response of a bladed disk assembly with frictional interfaces. It will be shown that the FETI method effectively addresses the limitations and assumptions of state-of-the-art approaches that solve the nonlinear equation of motion using a reduced-order fashion
Online Joint Identification of Structural Dynamic Response Anomalies and Structural Damage Using Limited Data
No full textStructural Health Monitoring (SHM) systems aim to enable real-time assessment of structural conditions by continuously collecting structural response data through sensor networks, while leveraging advanced techniques for data processing and analysis. However, their practical application remains hindered by several challenges: sensor faults frequently degrade data reliability; the lack of adaptive model updates restricts performance under changing environmental and operational conditions; and accurate damage diagnosis is often impeded by the scarcity of labeled damage data, which limits model generalizability and diagnostic accuracy. To address these challenges, this thesis proposes a limited-data-driven framework for the online joint identification of structural response anomalies and structural damage. The research includes three main components: (1) To address the limitations of current sensor fault diagnosis methods, particularly their poor performance under limited labeled data and lack of online adaptability, an online meta-learning approach is proposed for sensor fault diagnosis using limited data. A 1D convolutional neural network (1D-CNN) is employed to detect and locate faulty sensors, with initial weights optimized via model-agnostic meta-learning to enhance adaptability across sensor fault classification tasks. After detecting and locating the faulty sensors, an online updating algorithm based on a dual Kalman filter is used to estimate the severity of sensor faults and structural states simultaneously. The effectiveness of the method is validated through both numerical simulations and the Canton Tower benchmark, showing superior performance over conventional deep learning approaches. (2) To balance identification accuracy and computational efficiency, this study proposes a hierarchical damage identification approach that integrates Bayesian model selection with meta-learning. At the coarse identification stage, potential damaged substructures are detected by generating candidate models and computing their posterior probabilities based on the Bayesian information criterion. At the fine-grained identification stage, a meta-learning-based artificial neural network is employed to locate specific damaged elements and estimate their severity. The effectiveness of the proposed method is demonstrated through numerical simulations and a benchmark application on the Canton Tower, showing superior performance compared to conventional deep learning techniques. (3) To overcome the inherent inflexibility of threshold-based anomaly detection, an adaptive Bayesian inference framework is proposed for real-time simultaneous anomaly detection and system identification. Statistical models for random and gross errors are introduced to represent typical measurement anomalies, and Bernoulli random vectors are used for anomaly detection. An adaptive Bayesian scheme updates both the Bernoulli and model parameters, allowing for real-time simultaneous anomaly detection and system identification. The effectiveness of the method is demonstrated through numerical simulation, laboratory experiment, and practical implementation in the Canton Tower monitoring system. Overall, this thesis presents a framework based on meta-learning and Bayesian inference, capable of achieving accurate, adaptive, and efficient online joint identification of structural dynamic response anomalies and structural damage using limited training dat
Ammonia-hydrogen blends combustion in turbulent high temperature co-flow
No full textIn response to the stringent vehicle emission regulations, ammonia, with its potential as a carbon-free alternative fuel for reducing carbon emissions, faces application challenges due to higher ignition energy requirements and lower flame stability. Adding hydrogen is one of the effective ways to improve combustion performance of pure ammonia. This study focuses on the auto-ignition characteristics and jet flame stability of ammonia-hydrogen fuel blends under various conditions, such as different injection pressures, co-flow velocities, co-flow temperatures, and hydrogen blending ratios, employing a controllable active thermal atmosphere burner. Hydrogen addition increases flame brightness, area, and crinkly morphology due to enhanced NH2 production and higher combustion temperatures. The flame length increases together to the hydrogen ratio and the co-flow temperature, and it has been verified that it is primarily governed by jet momentum. Above 1073 K of co-flow temperature, the heat transfer becomes dominant for auto-ignition, reducing the effect of hydrogen presence. Combustion efficiency improves for higher co-flow temperatures, while hydrogen enhances propagation until a threshold is reached for XH2 = 20 %, beyond which a sensible increment in propagation cannot be detected. When the injection pressure augments, the flame is enlarged but auto-ignition can be hindered. Hydrogen addition reduces fluctuations, ensuring optimal stability for XH2 = 20 %
Diagnosis of Resting Tremor in Parkinson’s Disease Using Accelerometer and Gyroscope Sensors Built into a Smartwatch
No full textMonitoring resting tremor in Parkinson’s disease (PD) can be performed using wearable technology and machine learning. Smartwatches offer a cost-effective and non-intrusive way to track tremors remotely. However, to ensure precise monitoring in free-living environments, optimized systems are needed. This chapter discuss about the performance of inertial sensors to identify resting tremors and its classification according to MDS-UPDRS III. Six PD patients wore a smartwatch on their wrists while performing different exercise based on MDS-UPDRS. During eight weeks, data from triaxial accelerometers and gyroscopes were collected simultaneously and analyzed using machine learning techniques. In tremor presence detection, using binary classification, the use of only accelerometer gives the best results in terms of accuracy (97%) and training time (47 s) compared accelerometer and gyroscope combined (96.4% and 67 s) and only gyroscope alone (93% and 59 s). In the MDS-UPDRS scale detection, using multi-class models, the best accuracy is offered by the combination of accelerometer and gyroscope (96.5%) but offers the worst training times (77 s), while accelerometer is slightly worse (96.1%) but require the less training time (57 s). These results show the performance and training times of Machine Learning models for the detection of resting tremor and prediction of the MDS-UPDRS assessment for the correct decision making of sensors and models to be used in future application developments. The results could be used to contribute to the development of reliable tremor monitoring systems using devices equipped with inertial sensors and Machine Learning algorithms
Database-Driven Analysis of Energy Geostructures using a Global Dataset: Diffusion, Efficiency, and Environmental Performance
Get PDFEnergy Geostructures (EGs) are multifunctional systems that combine structural support with thermal energy exchange using low-enthalpy geothermal energy. This study presents a comprehensive analysis based on a global database of 972 case studies from 27 countries, primarily in Europe, including real world installations, test sites, and simulations. It focuses on the development and performance of various EG types – particularly energy piles (789 cases), energy walls (79), and energy tunnels (27) – making it the most extensive EG database to date. Geographically, Austria, Switzerland, Germany, and the UK lead in EG adoption, with Italy and France also contributing significantly. The analysis highlights both established technologies and emerging types, such as energy quay walls and barrettes, which show promising potential despite limited representation. The study reveals consistent geometric and design features: energy piles are used in small to medium-scale projects, energy walls offer large, activated surfaces, and tunnels are installed at intermediate depths. Thermal performance is linked to pipe configuration, diameter, spacing, materials, and environmental conditions – most systems are in stratified, moist soils in cool-temperate climates. EGs also offer environmental benefits, notably CO2 emissions reduction, reinforcing their value in sustainable infrastructure and heating and cooling network development
Explainable Prediction of Recurrence After Prostate Cancer Radiotherapy Using in Silico digital twin model and machine learning
No full textBiochemical recurrence (BCR) for prostate cancer (PCa) patients treated with External Beam Radiation Therapy (RT) has an incidence rate of up to 20 %. Thus, predicting BCR after PCa RT appears crucial for personalising treatments. Current approaches, such as radiomics and deep learning, applied to clinical and in vivo imaging data, suffer from limited explainability. This paper introduces a pipeline for predicting BCR by integrating clinical data with biologically grounded features derived from in silico digital twin simulations, supported by two explainability analyses. Specifically, we leverage a previously developed in silico digital twin model to simulate tumour growth and response to radiation for 315 PCa patients retrospectively treated with RT. A logistic regression model was identified as the best predictor, integrating clinical characteristics and biologically interpretable features extracted from simulations (AUC = 0.73). To enhance explainability, a local perturbation analysis is performed to quantify the influence of individual radiobiological parameters within the in silico model. Additionally, SHapley Additive exPlanations (SHAP) were applied to evaluate the contribution of each feature to the BCR prediction. By linking simulation-driven parameter importance with feature-level explanations, the pipeline provides coherent insights at the mechanistic and statistical levels