Although the effectiveness of machine learning (ML) for machine diagnosis has been widely established, the interpretation of the diagnosis outcomes is still an open issue. Machine learning models behave as black boxes; therefore, the contribution given by each of the selected features to the diagnosis is not transparent to the user. This work is aimed at investigating the capabilities of the SHapley Additive exPlanation (SHAP) to identify the most important features for fault detection and classification in condition monitoring programs for rotating machinery. The authors analyse the case of medium-sized bearings of industrial interest. Namely, vibration data were collected for different health states from the test rig for industrial bearings available at the Mechanical Engineering Laboratory of Politecnico di Torino. The Support Vector Machine (SVM) and k-Nearest Neighbour (kNN) diagnosis models are explained by means of the SHAP. Accuracies higher than 98.5% are achieved for both the models using the SHAP as a criterion for feature selection. It is found that the skewness and the shape factor of the vibration signal have the greatest impact on the models’ outcomes.

Additive Manufacturing (AM) processes enable their deployment in broad applications from aerospace to art, design, and architecture. Part quality and performance are the main concerns during AM processes execution that the achievement of adequate characteristics can be guaranteed, considering a wide range of influencing factors, such as process parameters, material, environment, measurement, and operators training. Investigating the effects of not only the influential AM processes variables but also their interactions and coupled impacts are essential to process optimization which requires huge efforts to be made. Therefore, numerical simulation can be an effective tool that facilities the evaluation of the AM processes principles. Selective Laser Melting (SLM) is a widespread Powder Bed Fusion (PBF) AM process that due to its superior advantages, such as capability to print complex and highly customized components, which leads to an increasing attention paid by industries and academia. Temperature distribution and melt pool dynamics have paramount importance to be well simulated and correlated by part quality in terms of surface finish, induced residual stress and microstructure evolution during SLM. Summarizing numerical simulations of SLM in this survey is pointed out as one important research perspective as well as exploring the contribution of adopted approaches and practices. This review survey has been organized to give an overview of AM processes such as extrusion, photopolymerization, material jetting, laminated object manufacturing, and powder bed fusion. And in particular is targeted to discuss the conducted numerical simulation of SLM to illustrate a uniform picture of existing nonproprietary approaches to predict the heat transfer, melt pool behavior, microstructure and residual stresses analysis.

Mechanical vibrational energy, which is provided by continuous or discontinuous motion, is an infinite source of energy that may be found anywhere. This source may be utilized to generate electricity to replenish batteries or directly power electrical equipment thanks to energy harvesters. The new gadgets are based on the utilization of piezoelectric materials, which can transform vibrating mechanical energy into useable electrical energy owing to their intrinsic qualities. The purpose of this article is to highlight developments in three independent but closely connected multidisciplinary domains, starting with the piezoelectric materials and related manufacturing technologies related to the structure and specific application; the paper presents the state of the art of materials that possess the piezoelectric property, from classic inorganics such as PZT to lead-free materials, including biodegradable and biocompatible materials. The second domain is the choice of harvester structure, which allows the piezoelectric material to flex or deform while retaining mechanical dependability. Finally, developments in the design of electrical interface circuits for readout and storage of electrical energy given by piezoelectric to improve charge management efficiency are discussed.

Managing the lifecycle of the elements of a complex and safety-critical system, from conceptual design to support in operation, is still a relevant challenge in the industrial domain. Starting from research and engineering activities, up to production and delivery, the need of maintaining the trace of dependence, changes, and possible updates regarding the final product is a crucial aspect for success. Particularly, conceptual and preliminary design activities are affected by a high level of uncertainty and a certain degree of project instability, usually caused by the need to explore alternative solutions and variants. Different tasks are performed concerning the identification of system requirements, functional and physical behavior using model-based design (MBD), and model-based systems engineering, which, generally, are subjected to extensive trade studies aimed at identifying the best solution. In this context, this paper presents how model-based digital twin and threads concepts will likely change the way in which the MBD process is managed, overcoming the issues associated with federated IT infrastructures and with tools integration. Notably, lifecycle and nonlifecycle related interoperability aspects are described, with particular focus on the exploitation of standards for lifecycle collaboration and heterogeneous simulation. The design of an ice protection system for a regional aircraft is selected as case study, starting from the work performed by the authors within the EC-funded research project CRYSTAL.

Today’s deep learning strategies require ever-increasing computational efforts and demand for very large amounts of labelled data. Providing such expensive resources for machine diagnosis is highly challenging. Transfer learning recently emerged as a valuable approach to address these issues. Thus, the knowledge learned by deep architectures in different scenarios can be reused for the purpose of machine diagnosis, minimizing data collecting efforts. Existing research provides evidence that networks pre-trained for image recognition can classify machine vibrations in the time-frequency domain by means of transfer learning. So far, however, there has been little discussion about the potentials included in networks pre-trained for sound recognition, which are inherently suited for time-frequency tasks. This work argues that deep architectures trained for music recognition and sound detection can perform machine diagnosis. The YAMNet convolutional network was designed to serve extremely efficient mobile applications for sound detection, and it was originally trained on millions of data extracted from YouTube clips. That framework is employed to detect bearing faults for the CWRU dataset. It is shown that transferring knowledge from sound and music recognition to bearing fault detection is successful. The maximum accuracy is achieved using a few hundred data for fine-tuning the fault diagnosis model.