34 research outputs found
Data-driven methodologies for Predictive Maintenance and TPM implementation: research approaches and applications
No full textNegli ultimi anni, la diffusione dell'Industria 4.0 (I4.0) ha reso disponibile una grande quantità di dati in real time riguardo le condizioni operative degli impianti industriali. Tali dati possono essere analizzati attraverso metodologie data-driven per il riconoscimento di comportamenti anomali e la definizione delle strategie di Manutenzione Predittiva (PdM). Allo stesso modo, l’applicazione delle tecnologie I4.0 in ottica Lean Production (LP), come ad esempio nell’ambito della Total Productive Maintenance (TPM), può risultare vincente in quanto la LP è ancora una delle soluzioni industriali più efficienti nel settore manifatturiero. Nell'era dell’I4.0, l'obiettivo della presente tesi è quindi quello di proporre nuove metodologie data-driven per l’applicazione di strategie di PdM e TPM. Per quanto riguarda le strategie PdM, sono state proposte tre diverse applicazioni. In primo luogo, è stato sviluppato un framework che integra diverse tecniche di Data Mining (DM) per la classificazione dei modi di guasto in base alla loro influenza sull'Overall Equipment Effectiveness (OEE), e la successiva identificazione delle relazioni considerate critiche, ovvero, quelle tra i modi di guasto appartenenti alla fascia OEE più bassa. Successivamente, l’analisi è stata complicata proponendo due nuovi frameworks basati sull’applicazione di algoritmi di Machine Learning (ML). In particolare, il Sequential Pattern Mining (SPM) è stato utilizzato per analizzare le sequenze di guasti e riconoscere i modelli ricorrenti tra i guasti che si verificano in una linea produttiva. Nell’ultima applicazione, viene utilizzata invece la Sequential Long Short-Term Memory (LSTM) per comprenderne le potenzialità da un punto di vista pratico. Infine, le tre metodologie data-driven sono state testate nel caso di una nota azienda italiana del settore automobilistico per identificare la metodologia con le migliori capacità predittive. L’obiettivo finale è stato quello di implementare una vera e propria piattaforma auto apprendente per la manutenzione predittiva nel caso studio in esame. Quattro ulteriori frameworks sono stati poi sviluppati nell’ambito del TPM. In primo luogo, è stato analizzato l’OEE, in qualità di indicatore principale del TPM, tramite l’applicazione delle Association Rules (AR) e la Network Analysis (NA). Successivamente, è stata proposta un'estensione del framework che mira alla progressiva implementazione del TPM tenendo conto sia dei fattori di perdita che di miglioramento dell'OEE, nonché della loro influenza sugli eventi del processo produttivo. Infine, è stato anche studiato il pilastro inerente all’Administrative TPM con l’obbiettivo di fornire supporto ai processi produttivi anche attraverso il miglioramento delle funzioni logistiche. In particolare, sono state proposte due nuove metodologie che mirano, da un lato, alla riduzione dei ritardi nell'evasione degli ordini in un magazzino tradizionale tramite una revisione dell’analisi ABC incrociata e, dall’altro lato, alla riduzione dei problemi di approvvigionamento in un Automated Storage and Retrieval System (ASRS) integrando tecniche di AR e di simulazione.In recent years, the diffusion of Industry 4.0 (I4.0) made available a vast amount of real-time data on the operating conditions of industrial plants. Such data can be analyzed through data-driven algorithms to recognise anomalous behaviours and define Predictive Maintenance (PdM) strategies. Likewise, using I4.0 technologies from a Lean Production (LP) perspective, such as in Total Productive Maintenance (TPM), can be successful since it is still one of the most efficient industrial solutions in manufacturing. The aim of this thesis is thus to propose new data-driven methodologies for implementing both PdM and TPM strategies in the era of I4.0. Concerning PdM strategies, three different applications are presented. First, a framework integrating multiple Data Mining (DM) techniques allows for classifying failure modes according to their influence on the Overall Equipment Effectiveness (OEE) and then identifying relationships among those in the lowest OEE range. Then, two additional frameworks are proposed based on Machine Learning (ML) algorithms to complicate the analysis. Sequential Pattern Mining (SPM) is first used to analyse failure-related sequences recognizing recurrent patterns among failure occurrences across the production line. In contrast, sequential Long Short-Term Memory (LSTM) is used in the third application to understand its practical potentiality. The three data-driven methodologies are, in fact, tested on a well-known Italian company in the automotive sector to identify the best predictive capabilities leading to developing and implementing a real self-learning platform for PdM. Four additional frameworks are then proposed from a TPM perspective. First, OEE is investigated as the leading TPM indicator applying Association Rules (ARs) and Network Analysis (NA) in the automotive company. Then, an extension of the framework aims to progressively implement TPM based on both the loss and better OEE factors and production events. Eventually, since TPM must embrace the entire company, the Administrative TPM pillar is achieved to support production processes by improving logistics operations. In particular, two new methodologies aim to reduce order processing delays in a traditional warehouse by revising the formal cross-ABC analysis and sourcing parts issues in an Automated Storage and Retrieval System (ASRS) through ARs and simulations techniques
A modified cross-abc analysis for direct inventory management
No full textData analytics represents an important step for manufacturing companies aiming at optimizing warehouse management. Moreover, in a Decision Support System perspective, companies require automated methods and processes for warehouse activities planning able to identify the most appropriate inventory management policy. Moving towards a data-driven environment, a deeper understanding of data provided by warehouse operations and supplier's orders is needed. In this context, a procedure aiming at reducing the problem of out of stock and overstock by developing a Decision Support System is deployed through a data-driven approach. The standard crossABC analysis is revised as a top-down methodology: in this way, cross-ABC analysis is applied iteratively by ranking items according to their physical features, e.g., color or size, in order to identify the weight of the items for each attribute classification, then combining the results for the final rating. The first step of the proposed procedure requires the identification of a weight for each feature, based on the influence of its importance. Then, a Decision Support System is developed in order to customize and automate the following daily activities: (a) application of the revised cross-ABC analysis, (b) warehouse management, and (c) orders planning. The main contribution of this work is the development of a versatile and dynamic procedure applying the well-known cross-ABC analysis in a different way, also developing and integrating a direct warehouse management system. A case study of a manufacturing company is also presented to explain the proposed procedure, as well as to analyze its performance and the different results compared to the standard analysis
Artificial Intelligence Within Food Processing Industry with the Scope of Enhancing Workplace Safety: A Literature Review
No full textThe pursuit of efficiency and productivity in the food processing industry often requires advanced machinery, complex procedures, and various potential hazards, highlighting the critical importance of workplace safety. Accidents and injuries not only can compromise the quality of products and the well-being of consumers, but it may also jeopardize the safety of workers. Within this framework, the utilization of artificial intelligence emerges as a promising solution to mitigate workplace hazards and create a safer environment for human operators. This research paper aims at providing a literature review in the present topic: how Artificial Intelligence can reduce workplace hazards in the food processing industry. Initially, the methodology employed for this literature review is outlined. Subsequently, an exhaustive examination is presented to generate bibliometric maps and tables that succinctly outline the principal scientific patterns within this subject matter, utilizing the assistance of the VosViewer software, which is also introduced
A rule-based machine learning methodology for the proactive improvement of OEE: a real case study
No full textPurpose: The Overall Equipment Effectiveness (OEE) is considered a standard for measuring equipment productivity in terms of efficiency. Still, Artificial Intelligence solutions are rarely used for analyzing OEE results and identifying corrective actions. Therefore, the approach proposed in this paper aims to provide a new rule-based Machine Learning (ML) framework for OEE enhancement and the selection of improvement actions. Design/methodology/approach: Association Rules (ARs) are used as a rule-based ML method for extracting knowledge from huge data. First, the dominant loss class is identified and traditional methodologies are used with ARs for anomaly classification and prioritization. Once selected priority anomalies, a detailed analysis is conducted to investigate their influence on the OEE loss factors using ARs and Network Analysis (NA). Then, a Deming Cycle is used as a roadmap for applying the proposed methodology, testing and implementing proactive actions by monitoring the OEE variation. Findings: The method proposed in this work has also been tested in an automotive company for framework validation and impact measuring. In particular, results highlighted that the rule-based ML methodology for OEE improvement addressed seven anomalies within a year through appropriate proactive actions: on average, each action has ensured an OEE gain of 5.4%. Originality/value: The originality is related to the dual application of association rules in two different ways for extracting knowledge from the overall OEE. In particular, the co-occurrences of priority anomalies and their impact on asset Availability, Performance and Quality are investigated
A data-driven framework for supporting the total productive maintenance strategy
No full textMaintaining machinery efficiency is a challenge in modern manufacturing, with companies adopting Total Productive Maintenance (TPM) strategies and using Overall Equipment Effectiveness (OEE) as a key metric for operational improvements. Industry 4.0 features have provided advanced data processing tools to assess machine performance, but effectively leveraging this data to enhance OEE remains challenging due to their multisource nature. To fully leverage the vast amount of data, new analytical methodologies are essential, not only to identify production events negatively impacting OEE but also to enhance positive events, thereby supporting strategic decision-making for TPM implementation. This study addresses this need by introducing a new data-driven methodology that integrates lean TPM principles with I4.0 analytical tools. Specifically, the proposed framework provides company managers with a structured approach, employing a dual application of Association Rule Mining (ARM) and Network Analysis (NA) to discover hidden relationships within OEE factors—availability, performance, and quality—and between key factors and operational variables. By assessing the impacts on equipment effectiveness, this methodology aims to prevent negative domino effects from adverse events while promoting positive events. Finally, a data-driven decision support system provides a long-term roadmap for determining which TPM pillars and associated actions should be prioritized for improvement. A case study demonstrates the framework's effectiveness, leading to key outcomes: implementation of 24 kaizen actions, achievement of 7 out of the 8 TPM pillars, and a 14% improvement in OEE
Implementation of Industry 4.0 techniques in Lean Production technology: A Literature Review
No full textLean thinking and Industry 4.0 have been broadly investigated in recent years in intelligent manufacturing. Lean Production is still one of the most efficient industrial solutions in business and research, despite being implemented for a long time. On the other hand, Industry 4.0 has been introduced referring to the fourth industrial revolution. This study aims to analyze the combination of both Industry 4.0 and Lean production practices through a systematic literature review from a Lean Automation perspective. In this field, 189 articles are examined using VOSviewer for cluster analysis. Then, a more detailed analysis is provided to explore how Industry 4.0 and Lean techniques are integrated from a practical perspective. Results highlighted Big Data Analysis and Value Stream Mapping as the most common techniques, also emphasizing a growing trend toward new publications. Nevertheless, few practical applications are identified in the literature highlighting six gaps in the correlation of LA practices
Effects of a new photoactivatable cationic porphyrin on ciliated protozoa and branchiopod crustaceans, potential components of freshwater ecosystems polluted by pathogenic agents and their vectors.
No full textThe increasing use of photosensitized processes for disinfection of microbiologically polluted waters requires a precise definition of the factors controlling the degree of photosensitivity in target and non-target organisms. In this regard, tests with protozoa and invertebrates which have a natural habitat in such waters may be used as first screening methods for the assessment of possible hazards for the ecosystem. A new cationic porphyrin, namely meso-tri(N-methyl-pyridyl)mono(N-dodecyl-pyridyl)porphine (C12), is tested in this work on the protozoan Ciliophora Colpoda inflata and Tetrahymena thermophila and the Crustacea Branchiopoda Artemia franciscana and Daphnia magna. The protocol involved 1 h incubation with porphyrin doses in the 0.1-10.0 μM range and subsequent irradiation with visible light at a fluence rate of 10 mW cm(-2). The results indicate that C12 porphyrin has a significant affinity for C. inflata and T. thermophila; this is also shown by fluorescence microscopic analyses. C. inflata cysts were resistant to the phototreatment up to a porphyrin dose of 0.6 μM. The effects of C12 on cysts have been evaluated at 3 and 24 h after the end of the phototreatment; a delay in the excystment process was observed. T. thermophila was fairly resistant to the phototreatment with C12 porphyrin. The data obtained with the two crustaceans indicated that the effects of dark- and photo-treatment with C12 need to be closely examined for every organism. A. franciscana is more resistant, probably owing to its ability to adapt to extreme conditions, while the high level of photosensitivity displayed by Daphnia magna represents a potential drawback, as this organism is often selected as a reference standard for assessing the environmental safety. Thus, while C12 photosensitisation can represent a useful tool for inducing a microbicidal or larvicidal action on polluted waters, the irradiation protocols must be carefully tailored to the nature of the specific water basin, and in particular to its biotic characteristics
The New Humanities Project—Reports from Interdisciplinarity
No full textNew Humanities is an international research and teaching project promoted by an interdisciplinary group of people from five different faculties and departments based at the University of Roma Tre. Initially set up as a forum for academic dialogue between the humanities and the sciences (including social sciences), the project became a transition space and platform for experiencing new research methodologies and teaching curricula that would question the present epistemological order of the European university system. In order to develop this approach, we have organized our work around a number of interdisciplinary clusters, each describing an epistemological node. In this paper we will discuss five interconnected case studies that emerged from an active collaboration between scientists and humanists. The first node, Protocols of Vision, investigates the cognitive nature of sensory perception and the different forms of knowledge it produces—empirical, artistic, and scientific. Memory: Mathematics, Computer Science, and Literature recapitulates many of the different threads in these discussions by exploring the interdependencies between the various kinds of memory: from external to subjective memory, from storage tools and techniques of self-construction to the invariance of mathematical structures. The third node, Signs and Bodies between Digital and Gendering, reflects on the problematic relationship between digital media and literary and linguistic gendering. Narrative Identity: Nature, Ontogeny and Psychopathology critically re-examines the main concepts and theories concerning the nature, ontogeny, and pathologies of the autobiographical self or narrative identity. Finally, the last node, Contribution of Quantum Physics to the Idea of Consciousness is a cross-cultural investigation into the phenomenon of consciousness tackled from the points of view of quantum field theory and ancient Indian philosophy
The Celiac Iceberg in Italy: Exploring Primary School Children by RIA Salivary Anti-Transglutaminase Antibodies
No full text
