1,721,230 research outputs found
Alterazioni delle materie coloranti nelle pitture murali prodotte dalle alte temperature: fonti storiche ed indagini scientifiche
No full text
Physics-Informed deep Autoencoder for fault detection in New-Design systems
No full textThe industrial application of data-driven methods for fault detection of new-design systems is limited by the inevitable scarcity of real data. Physics-Informed Neural Networks (PINNs) can mitigate this problem by integrating data and physical knowledge. In this work, we develop a novel fault detection method that combines physics-based simulations for data generation with a Physics-Informed Deep Autoencoder (PIDAE) for reproducing the system behaviour in normal conditions; the Sequential Probability Ratio Test (SPRT) is, then, used for detecting abnormal conditions. The proposed method is applied to new-design electro-hydraulic servo actuators used in turbofan engine fuel systems. The results show that it can provide more satisfactory fault detection performance, in terms of false and missed alarms, than state-of-the-art methods based on traditional autoencoders only and pure physics-based models only. Furthermore, the PIDAE outcomes are physically consistent and, therefore, more acceptable and trustworthy
Long-term survival in myocardial infarct. Analysis of the post-infarct course in relation to the clinical aspects of the acute phase
No full textOptimization Method for an Improved Training of Physics Informed Neural Networks
No full textPhysics Informed Neural Networks (PINNs) are promising methodologies to improve accuracy and extend applicability of Deep Learning in engineering applications, providing a hybrid modeling that combine physics domain knowledge with data-driven methods. Nevertheless, the adoption of PINNs in practice is still limited by the complexity of the training process, where the unbalanced loss gradients in the back propagation step can lead to inaccurate training. This paper discusses a method to improve PINNs training by optimizing the hyperparameters of the model in a standard NN scenario, before applying physics-based constraints, and then balancing the backpropagation by weighting the PINN loss gradients to target the simplified NN loss decay. The result is a faster and improved PINN training that can enable hybrid modeling in industrial applications where both data and domain knowledge are available. A case study is a fluid-dynamics problem taken from literature, described by the Navier Stokes equations that drive thermosenergy processes
I dipinti duecenteschi dell’edicola funeraria di Malgerio Sorello nella Abbazia di Santa Maria di Ferraria (Caserta): vicende storiche, tecniche esecutive,conservazione
No full textAutomatic Extraction of a Health Indicator from Vibrational Data by Sparse Autoencoders
Full text linkWe present a method for automatically extracting a health indicator of an industrial component from a set of signals measured during operation. Differently from traditional feature extraction and selection methods, which are labor-intensive and based on expert knowledge, the method proposed is automatic and completely unsupervised. Run-to-failure data collected during the component life are fed to a Sparse AutoEncoder (SAE), and the various features extracted from the hidden layer are evaluated to identify those providing the most accurate quantification of the component degradation. The method is applied to a synthetic and a bearing vibration dataset. The results show that the developed SAE-based method is able to automatically extract an efficient health indicator
Fault diagnostics by conceptors-aided clustering
Full text linkFault diagnostics in practice faces the challenge of dealing with unlabelled time series that have long-term temporal dependencies. Inspired by the idea of representing temporal patterns by a mechanism of neurodynamical pattern learning, called Conceptors, we propose an unsupervised clustering method for identifying the degradation state of industrial equipment. Conceptors are used to represent the dynamic behaviour of the degradation trajectories and spectral clustering is used to group the Conceptors in homogenous classes of similar degradation states. The proposed method is applied to a case study of literature. The results show that the accuracy of the fault diagnosis is satisfactory
A data-driven framework for identifying important components in complex systems
No full textComplex technical infrastructures are systems of systems characterized by hierarchical structures, made by thousands of mutually interconnected components performing different functions. Given their complexity, it is difficult to derive their functional logic using traditional risk and reliability analysis methods based on engineering knowledge. In this work, we propose to address the problem in an innovative way that makes use of the large amount of data available from monitoring those systems. Specifically, we develop a data-driven framework to identify the critical components of a complex technical infrastructure. The criticality of a component with respect to the safe/failed state of the infrastructure is assessed considering a feature selection technique which employs Random Forest (RF) classification and a feature importance score. The proposed data-driven framework is applied to a nuclear power plant system and a synthetic case study, which mimics the complexity of a technical infrastructure
- …
