ABSTRACT. Efficient and robust fault detection in PHM typically depends on high-quality run-to-failure data, yet such labeled historical data is often scarce. While unsupervised algorithms trained on data assumed to represent normal operation (such as early-stage operation data) can yield good performance, they may be limited by contamination or incomplete coverage of normal states, related to unseen operating scenarios and conditions. This study investigates the effects of incomplete coverage of normal scenarios on fault detection performance. To enhance robustness against such incompleteness without making assumptions on the dataset’s composition, employing weakly supervised anomaly detection by incorporating few labeled fault samples is proposed. Evaluations on degradation testing data show that providing an autoencoder with a few fault labels significantly improves detection performance compared to a fully unsupervised autoencoder. This improvement is shown to be robust under various conditions, including changes in the number and quality of labeled fault samples and variations in the training data's composition. The proposed approach realistically simulates incomplete coverage through selective dataset composition and builds on weak supervision to enhance fault detection, making it particularly suitable when only field data – with no information on its composition - and few labeled fault samples are available.

ABSTRACT. Proton Exchange Membrane Fuel Cells (PEMFCs) play a critical role in zero-emission electro-hydrogen generators (GEH2). Accurate remaining useful life (RUL) prediction for PEMFCs is essential to optimize operational efficiency and enable cost-effective predictive maintenance. A key challenge in RUL estimation lies in selecting an effective Health Indicator (HI) that reliably reflects the degradation state of the PEMFC stack. While voltage and power-based HIs are commonly used under static conditions due to their monotonic decline and ease of measurement, they become unreliable under dynamic operating conditions due to their sensitivity to mission profiles. To address this issue, we propose a machine learning model to extract a robust HI from voltage measurements, capturing the true degradation state of the PEMFC. Building on this HI, we propose a hybrid deep learning architecture combining Transformer networks and Gated Recurrent Units (GRUs) to model temporal dependencies in PEMFC degradation and predict the RUL of PEMFCs operating under dynamic conditions. The proposed approach is validated using a real industrial dataset comprising four PEMFCs. Comparative analyses against state-of-the-art machine learning methods demonstrate that our approach consistently outperforms existing models, exhibiting superior predictive accuracy and robustness. These results highlight the framework’s robustness in real-world scenarios, offering a scalable solution for predictive maintenance of PEMFC systems under dynamic mission profiles.

ABSTRACT. Maintenance strategies are pivotal in ensuring the reliability and performance of critical components within industrial machines and systems. However, accurately determining the optimal replacement time for such components under stress and deterioration remains a complex task due to inherent uncertainties and variability in operating conditions. In this paper, we propose a comprehensive approach based on Robust Markov Decision Processes (RMDP) to optimize component replacement decisions in machines with one critical component while addressing uncertainty in a structured manner. RMDP offers a robust framework for decision-making under uncertainty, allowing for the modeling of component degradation and variability in operating conditions. Our methodology uses data-driven ambiguity sets, including likelihood-based and Kullback-Leibler (KL)-based ambiguity sets, to capture and quantify uncertainty in the degradation process. We show the mathematical relationship between the KL-based and Likelihood-based ambiguity sets and provide statistical guarantees for the optimal cost. Through computational experiments, we demonstrate the effectiveness of our RMDP approach in identifying the optimal replacement time that minimizes the total maintenance cost while exhibiting greater stability compared to traditional methods.

ABSTRACT. Hybrid offshore renewable energy systems, which integrate technologies such as wind turbines and floating solar photovoltaics, present promising solutions to enhance the reliability of energy production while reducing investment and operational and maintenance (O&M) costs through shared infrastructure. However, design decisions, such as system sizing, bring critical O&M challenges, including operational accessibility, system reliability, availability, and maintenance strategies that influence the project's economic feasibility and long-term performance. Nevertheless, existing studies often overlook O&M challenges by relying on oversimplified assumptions such as fixed availability rates, maintenance costs as a percentage of overall capital expenditures, or a predetermined number of maintenance over the project's lifespan, neglecting their impact on system sizing decisions. To bridge this gap, this work proposes a novel framework to optimize the system configuration by incorporating a more realistic estimation of O&M costs. The model incorporates design decisions and relevant maintenance strategies, including corrective and preventive approaches, while accounting for component degradation, vessel fleet selection, accessibility, availability, and energy production. When applied to real case studies, this framework provides valuable insights for optimizing the design and operation of hybrid offshore systems. It highlights the interplay between operational and design decisions while maximizing investor returns.

ABSTRACT. Accelerated life tests are experimental methods used to estimate the lifespan or reliability of a product, component, or material by subjecting it to conditions more severe than normal operational environments. The goal is to induce failures more quickly, so that data can be collected in a shorter time, allowing engineers and designers to predict the product's performance under normal conditions over a longer period. The main objective of the present study is to define the lifetime of supercapacitors through accelerated life testing. An introduction is made regarding the mathematical characteristics of accelerated life tests applied to innovative energy storage technologies. To validate and implement the presented concepts, the referred tests were conducted on supercapacitor prototypes, subjected to various voltages, exceeding normal operating levels, to accelerate their degradation and induce failure, allowing to predict the lifespan of such devices under normal operating conditions. The experimental data collected in the study was obtained by carrying out floating tests, that were fitted to an appropriate degradation model and analyzed. Results showed that this model can be applied to supercapacitors' accelerated life tests, allowing the comparison of different materials and obtaining the best solutions.

ABSTRACT. Multi-stack Proton Exchange Membrane Fuel Cells (PEMFCs) have gained significant attention due to their high efficiency, scalability, and flexible operation, making them a promising solution for carbon-free, high-power applications. However, durability challenges remain a major barrier to commercialization, as fuel cell degradation directly impacts system reliability and operational costs. Prognostics and Health Management (PHM) approaches have shown great potential in extending fuel cell lifespan by enabling predictive maintenance and optimized load allocation strategies. This paper proposes a load-dependent degradation model based on a non-homogeneous stochastic gamma process to predict PEMFC degradation under varying load conditions. The model effectively captures uncertainties, variability, and non-linear degradation dynamics, providing a more accurate estimation of Remaining Useful Life (RUL). Leveraging this model, a health-aware energy management strategy is developed for multi-stack configurations, dynamically adjusting power distribution to mitigate degradation effects. Additionally, an optimal maintenance strategy is introduced, incorporating replacement decisions to minimize operational costs while maintaining system reliability. The proposed methodology is evaluated against traditional energy management approaches, demonstrating its potential to significantly enhance PEMFC durability and cost-effectiveness in real-world applications.

ABSTRACT. For the energy pipeline industry in particular, pipeline wall condition assessment is crucial to what they term integrity management. Failure to detect a defect (localised reduction in wall thickness) or to accurately predict defect growth can lead to catastrophic failure. Industry seeks practical modelling techniques compatible with available data. Relevant types of data informing defect prediction that is normally available includes environmental, operational, construction parameter, and various types of condition data. Increasingly, reliable condition data generated by ‘in-line-inspection’ (ILI) is available. The cost of ILI precludes frequent use resulting in scare data: perhaps 2 to 5 data points per site. This prevents the practical application of complex modelling techniques for the prediction of specific defect growth. We report a study exploring on the use of simple models of each with two or three parameters. Five simple models were tested using the ILI datasets collected from three energy pipelines. It was found that these models can present reasonable prediction accuracy, with the logarithmic model returning the best prediction accuracy of approximately 80%. Based on the test outcome, a framework is proposed that guides individual defect growth prediction in industrial practice. The framework can be integrated into existing industry-workflows, enabling practical implementation to enhance operational safety and decision-making in pipeline operation.

ABSTRACT. The study presented in this paper assess failure incidents and predict maintenance requirements for Electric Multiple Units (EMU) trains using time series forecasting techniques. A Seasonal Autoregressive Integrated Moving Average with Exogenous variables (SARIMAX) model approach was utilised, and an attempt is made to forecast the patterns of failure incidents using weekly and monthly CCTV failure data. The Augmented Dickey-Fuller (ADF) test is used for stationarity assessments of the failure data, while autocorrelation and partial autocorrelation function is used to determine the seasonality. Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) are applied to ensure model selection balances statistical accuracy and complexity, preventing overfitting SARIMAX risk-based model approach is used to forecast the weekly and monthly CCTV failure incidents and then categorise the risk of failure of CCTV systems into low, medium and highrisk levels, represented with a traffic light system. Graphical display of the model simulation results from three geographical locations in Scotland is presented in this paper. The outcome of this paper is to support informed decisions on maintenance scheduling, and the allocation of maintenance resources based on the risk of failure.

ABSTRACT. Motivated by the Right-to-Repair legislation, we propose a data-driven warranty pricing model that uses non-life insurance techniques. Our proposed model aims to address the complexity involved in determining warranty fees, which are influenced by two main factors: unpredictable return patterns and stochastic remanufacturing costs based on product quality. To address these challenges, we use historical return data and employ a survival regression model based on product characteristics to predict the time until a product is returned. Using the same product characteristics, a gradient-boosting machine predicts the remanufacturing costs of a return. The combination of these two techniques enables the customization of the warranty fee based on the product characteristics. We validate our approach using a dataset of about 35,000 returns over a five-year period from an automotive parts supplier. Our model provides a framework for industrial companies to manage return management strategies and warranty pricing.

ABSTRACT. Reconfigurable Manufacturing Systems (RMSs) are designed to adapt to fluctuating market demands and evolving product configurations. Unlike traditional manufacturing systems, RMSs are characterized by their ability to reconfigure in response to changes in production capacity and product design. However, maintaining these systems presents significant challenges due to their modular, scalable and dynamic nature. Conventional maintenance models do not fully exploit the unique modularity, scalability, customization and convertibility of RMSs, leading to suboptimal maintenance strategies and increased downtime. Although maintenance is critical for improving the performance and adaptability of RMSs, most existing research has focused primarily on their design and operation, leaving a gap in the literature. To address these limitations, this research proposes a novel maintenance framework considering key RMS characteristics, providing a structured methodology for maintenance modeling and decision-making. This framework allows the designing of maintenance strategies in RMSs to enhance system reliability and responsiveness. Various case studies were then conducted to investigate the feasibility and the robustness of the proposed framework, demonstrating its practical use in RMS maintenance planning and serving as an adequate tool for engineers and practitioners aiming to optimize maintenance in RMS environments.

ABSTRACT. Inverters are the key component in a photovoltaic operation, as their failures can lead to the disconnection of hundreds of photovoltaic modules. If the inverter fails, several hundred kW of power can be completely lost. Furthermore, this component is the most prone to failure in photovoltaic operations. Out of the all the possible failure modes for the inverter, the capacitor and power component failures are the most prominent. For these reasons, this paper presents a detection and prognostic method based on the creation of a feature using domain knowledge to understand the physics behind the failures.

ABSTRACT. Comparisons between Condition Based Maintenance (CBM) and Time Based Maintenance (TBM) are often made under a fixed cost assumption. However, wind speed is seasonal, hence opportunity cost of maintaining wind turbines follow a cyclic pattern too. In this talk the fixed cost assumption is therefore abandoned. To compare CBM and TBM, for both methods a discrete Markov Decision Process (MDP) is formulated, which is solved by means of linear programming. The one-step transition probabilities of the deterioration process are derived from a stationary discrete gamma process. When cost is time-varying, the optimal threshold policy for both TBM and CBM becomes dependent on the week of the year. The majority of preventive maintenance actions is conducted in campaigns at the cheapest moment of the year. It is found that CBM always outperforms TBM, also under time-varying cost. For a low Mean Time To Failure (MTTF), the relative performance of TBM improves when seasonality increases.

ABSTRACT. This study presented in this paper investigates failures contributing to increased downtime in an LPG plant gangway system. It aims to identify critical failure modes, assess the effectiveness of current maintenance strategies, and propose improvement initiatives to minimise downtime. The analysis utilises historical failure and maintenance data collected over a 4-year period. An exploratory analysis of the data is conducted using advanced statistical methods, and a failure mode and effect analysis approach is used to identify the critical failure modes and corresponding areas where the failure occur are classified as mechanical and instrumentation. The probability of failures and their impact on system reliability is calculated. Reliability analysis is conducted using the Fault Tree Analysis (FTA) approach. Inferential statistics are employed to evaluate the severity of downtime events systematically. The result provides an insight into more efficient operational and maintenance planning for the LNG plants critical systems.

ABSTRACT. Highly complicated machines contain many parts with few failures (up to 5 in the whole history). This makes predicting future failures challenging, hindering a cost-effective stocking and maintenance strategy. This article explores the benefits of switching from the traditional Poisson to the Zero Inflated Poisson (ZIP) distribution for the modeling of the occurrence of rare failures. Using real data, we compare the performance of the two modeling approaches and show that the proposed model significantly improves the forecasting accuracy of spare part storage management. ZIP also provides an improved prediction range that more accurately describes the magnitude of predicted failures.

ABSTRACT. Shannon introduced entropy in his seminal 1948 article entitled ``A mathematical theory of communications''. This notion has been used in a few papers at the turn of the century, for instance for the performance assessment of intelligent machines. By contrast, there has been in the last five years a fast growing literature on the application of Shannon entropy and its variants to system maintenance in various domains: radars, metro signaling, motor fault detection, etc. Yet the described systems are mostly either small or have a simple series or parallel architecture.

The purpose of this paper is to consider the different versions of entropy used in maintenance studies and compute their values for large and possibly meshed networks. Asymptotic expansions as a function of the system size are given and explained for various architectures of practical interest. The influence of the failure distribution density --- not necessarily exponential --- is also addressed, along with the changes brought by the introduction of common-cause failures.

ABSTRACT. We proposed a hybrid inspection and preventive replacement policy for technical systems subject to two types of opportunities: (1) opportunities to advance inspections at a lower cost, which occur more frequently since inspections demand less effort and time; and (2) opportunities to advance preventive replacements. The policy also allows for postponing planned preventive replacements in the case of operational constraints. To evaluate the proposed policy, we developed a mathematical model based on the delay-time concept, assuming perfect inspections. The expected cost-rate, availability, and mean time between failure events are evaluated, and the impact of uncertainties in parameter estimation is assessed through global sensitivity analysis. A numerical study is presented to demonstrate the policy's performance. Additionally, a web application is provided, allowing users to perform analyses. The policy was designed to integrate well with the working conditions of the system’s operational environment and proved to be a competitive alternative, especially in the context of manufacturing plants, where it enables better coordination between maintenance and production management.

ABSTRACT. Rotating equipment, such as pumps, gearboxes, and turbines, rely heavily on proper lubrication to ensure efficient operation and longevity. However, lubrication tanks can be susceptible to contamination from airborne particles, moisture, and other environmental factors, which can compromise the integrity of the lubricant and ultimately lead to equipment failure. Breathers play a vital role in preventing these contaminants from entering the tank, thereby safeguarding the quality of the lubricant and protecting the equipment. By allowing air to enter and exit the tank while filtering out impurities, breathers help maintain a stable atmosphere within the tank, reducing the risk of oxidation, corrosion, and microbial growth. This paper highlights the significance of breathers in rotating equipment lubrication tanks, discussing their design, functionality, and benefits, as well as best practices for selection, installation, and maintenance. By emphasizing the importance of breathers, this research aims to raise awareness among industry professionals about the critical need for effective contaminant control measures to optimize equipment performance, reduce downtime, and extend lifespan.

ABSTRACT. Effective maintenance strategies are essential for ensuring sustainability and cost efficiency in complex systems with heterogeneous components. A well-balanced maintenance policy must consider economic, social, and environmental factors by minimizing urgent interventions and reducing unnecessary replacements. By prioritizing planned maintenance over reactive tasks, organizations can lower maintenance costs, improve system reliability, and reduce after-hours workload, enhancing both operational efficiency and employee well-being. To achieve this, we propose a flexible maintenance policy that integrates condition-based maintenance (CBM) and failure-based maintenance (FBM), leveraging opportunistic maintenance to optimize intervention timing. Unlike traditional fixed-interval approaches, our method strategically incorporates semi-urgent and urgent visits as opportunities for preventive replacements, reducing reliance on costly urgent interventions.

To determine optimal maintenance actions, we formulate a Markov decision process (MDP) and address computational complexity through heuristic approaches. We propose two heuristics: Heuristic 0, which relies solely on corrective maintenance and incurs higher costs, and Heuristic 1, which incorporates semi-urgent and opportunistic maintenance to improve cost efficiency, sustainability, and computational feasibility. Numerical experiments demonstrate that the proposed policy significantly reduces maintenance costs, urgent interventions, and material waste. This study provides a scalable and cost-effective solution for industries aiming to enhance maintenance planning, system reliability, and sustainability by aligning economic, social, and environmental objectives.

ABSTRACT. As industries face increasing pressure from climate change and demographic shifts, traditional performance metrics like Overall Equipment Effectiveness (OEE) prove insufficient for addressing long-term sustainability. While OEE effectively measures productivity, it does not integrate environmental, social, and economic considerations that are crucial for sustainable operations. To bridge this gap, we propose the Overall Sustainable Equipment Effectiveness (OSEE) framework, which extends conventional performance evaluation by incorporating sustainability dimensions. The framework is developed through a comprehensive literature analysis and refined through expert validation to ensure applicability in real-world industrial settings. To navigate the complex interdependencies of sustainability factors, we employ causal AI methods, specifically Dynamic Bayesian Networks (DBN) and Knowledge Graphs (KG). These tools enable a structured representation of cause-effect relationships within sustainability contexts while facilitating quantitative optimization of their impacts on operational efficiency. By integrating AI-driven insights, the OSEE framework provides decision-makers with a systematic approach to balancing efficiency and sustainability. This research contributes to the broader twin transition, linking digital transformation with sustainability efforts to create resilient and future-oriented industrial systems.

ABSTRACT. Flexible manufacturing systems (FMS) face escalating challenges due to the vast number of components and potential failure causes, intensified by diverse product portfolios and fluctuating production volumes. This paper introduces an agent-based adaptive recommendation system that signifies a paradigm shift in data-driven maintenance management. By integrating the expertise of maintenance engineers and personnel, the system employs a Natural Language Processing (NLP)-based model to classify symptoms, causes, and solutions with high accuracy (achieving 100 % for symptoms, 90 % for causes, and 85 % for solutions). Leveraging Bayesian networks to model cause-and-effect relationships, the system enables probabilistic inferences and provides resource-efficient recommendations that minimize energy use, resource consumption, and waste. Continuous feedback mechanisms enhance the system's performance over time, improving diagnostic and repair processes while promoting sustainable maintenance practices.

ABSTRACT. Optimal cutting tool replacement policy is an essential subject in machining that balances the risk of producing out-of-tolerances pieces, and avoids excessive consumption of new tools and production interruption. The tool replacement policy is usually determined by striking an economic balance that does not consider the environmental impact of the policy and this situation contradicts the principles of the circular economy, which are now essential. In industrial practice, this often results in systematic preventive replacement of tools after a safely fixed number of produced pieces, wasting a portion of the tool's useful life (at an environmental cost) and increasing the machine's downtime. In this paper, based on advances in cutting tool monitoring, we assume that it is possible to monitor the tool wear. Under these hypotheses, we use a Monte-Carlo simulation method to show, through an example in Ti6Al4V milling, that a condition-based replacement allows improved economic and ecological performance, considering the single indicator of Global Warming Potential, expressed in kgeq CO2, of the tool replacement policy, even under uncertainty on the condition estimate.

ABSTRACT. Water distribution networks (WDNs) face increasing challenges due to urbanization, climate change, and aging infrastructure. This study presents a probabilistic framework to evaluate and enhance WDN resilience under multiple failure scenarios. Traditional pipe criticality assessments overlook dynamic hydraulic conditions; therefore, we integrate real-time pressure indicators to improve vulnerability analysis. Hydraulic simulations are conducted in EPANET, and pipe failure scenarios are implemented in a bespoke Python-based research code to assess network performance over time. A probabilistic failure and recovery model is developed to capture uncertainties in response times and operational constraints. Results of two case studies demonstrate that incorporating pressure-based rankings improves maintenance prioritization, shortens recovery times, and enhances resilience planning. The proposed framework provides water utilities with a robust decision-making tool to optimize resource allocation and maintain service continuity during disruptive events.

ABSTRACT. Disruptions on a rail network can halt operations at affected sections, leading to the rapid propagation of delays throughout the highly congested network, as observed in the UK rail network. Therefore, timely responses to disruptions are essential to minimising the delay propagation. This paper proposes a hybrid simulation-optimisation model that employs genetic algorithms to automate near real-time decision-making for disruption management. This research aims to determine optimal service alterations for targeted train services, facilitating faster recovery while minimising total delay minutes, additional costs incurred due to disruption management and the number of train cancellations across the network. The performance of the model is evaluated by comparing its outcomes with example empirical data from a historical disruption. Results demonstrate that the model effectively balances the three key objectives while adhering to operational constraints, therefore ensuring its practical applicability. The findings highlight the types of train services most beneficial to alter and quantify the improvement achieved, including reductions in service cancellations and compensation-eligible delays. In summary, this research contributes to modelling methods for strengthening the resilience of the rail network by enhancing its ability to recover from disruption efficiently.

ABSTRACT. This study focuses on the performance analysis of an M/M/1 retrial queueing system that incorporates the dynamics of an unreliable server and customers' impatience. When customers encounter the server in a busy state, they exhibit impatience or decide to join a retrial orbit, where they wait for an opportunity to access the server. After a random period, customers in the orbit may either attempt to retry for service or abandon the system altogether due to impatience. The server is subject to potential breakdowns, reflecting the real-world scenario of service interruptions. To analyze the system, we formulate the Chapman–Kolmogorov equations and solve them using probability-generating functions. These solutions yield the probability distributions, which are then used to derive various performance measures for the system. A numerical example is provided to demonstrate the impact of different system parameters on performance measures. Moreover, the study formulates a profit function based on the customers’ strategy of joining the system. This profit function is optimized using the quasi-Newton method (QNM) to maximize the system's profitability, thereby guiding decision-making to improve overall performance.

ABSTRACT. The increasing complexity of high-tech systems poses a significant challenge on service organizations tasked with the timely identification of the root causes of unexpected downtimes. While data-driven methods are effective for diagnosing frequently occurring issues, or those affecting a large number of systems, rare issues suffer from data scarcity, necessitating alternative approaches. This paper presents a two-step model-based methodology that leverages system architecture information to support diagnosing of systems for which little data is available. Firstly, system design and observability information, including diagnostic tests, is captured in a knowledge graph. Secondly, the knowledge graph is queried and transformed into a probabilistic graphical model. This is fully automated using transformation rules based on the ontology underlying the knowledge graph. The probabilistic graphical model then infers the most likely causes of failure using measurement data, and guides service engineers by suggesting cost-effective diagnostic actions. This paper outlines the proposed methodology, demonstrates its application on a small example system, and reports early-stage validation findings from high-tech system cases.

ABSTRACT. Fault Detection and Diagnosis (FDD) is crucial for both maintenance and control in chemical industries, where early fault detection can prevent costly failures and optimize operations. This is particularly critical in pilot units like those at IFP Energies Nouvelles (IFPEN), which operate under short-term experimental conditions with frequently varying operational parameters. This study conducts an extensive benchmarking analysis using the Tennessee Eastman Process (TEP), a widely used simulated chemical process dataset, to evaluate multiple approaches for fault detection and diagnosis, featuring numerous continuous operations with different sensors and faults. Several methods were implemented and compared including Multi Scale PCA (MS-PCA), AutoEncoder, Ensemble Learning, and LSTM models for fault detection, alongside Random Forest, XGBoost, and BLSTM (Bidirectional LSTM) for fault diagnosis. Using the TEP dataset, our results demonstrate that Ensemble Learning achieves detection rates ranging from 80% to 100% across various fault scenarios for the fault detection task. For the fault diagnosis task, BLSTM achieved a diagnosis accuracy of 98.76%. The study reveals that ensemble-based approaches consistently outperform individual models in handling the complex, multivariate nature of chemical process data, due to its robustness in combining multiple perspectives, comprehensive data capture, and localized detection capabilities. Furthermore, the superior performance of the BLSTM is due to its ability to capture both past and future temporal dependencies in the sequential data, particularly important in chemical processes where fault patterns may manifest with complex temporal relationships. This research contributes to the field of Fault Detection and Diagnosis by providing empirical evidence for the effectiveness of ensemble methods and Bidirectional LSTMs in order to address industrial FDD applications on chemical processes.

ABSTRACT. Machine groups play a pivotal role in modern industries such as wind farms and flexible production factories. These systems enable collaborative production through multiple interconnected machine nodes, promoting enhanced efficiency and quality in swarm production. Ensuring the reliable operation of these nodes is vital, making intelligent fault diagnosis a cornerstone for maintaining their health and functionality. The rapid development of deep transfer learning has further revolutionized fault diagnosis by enabling the transfer of knowledge from well-studied machine nodes (source domains) to less-explored nodes (target domains). This eliminates the need for extensive retraining on the target domain and underscores the importance of improving model generalization in fault diagnosis for machine group. However, machine group nodes often exhibit unique characteristics and varying degradation patterns, creating significant challenges. In many cases, multiple source domains are available, but none individually offer sufficient, highly transferable knowledge to the target domain. Multi-source transfer learning has emerged as a promising solution, focusing on extracting domain-invariant features from multiple sources and applying them to the target domain. Techniques such as maximum mean discrepancy (MMD) and generative adversarial networks (GANs) have been employed to address domain discrepancies, enhancing the generalization of diagnosis models across diverse domains. Despite their effectiveness, existing multi-source transfer methods often assume centralized domain data for distribution adaptation. This assumption is impractical in real-world engineering scenarios due to two primary challenges. First, data privacy concerns hinder data centralization, as monitoring data often contains sensitive information, including production quality metrics and manufacturing processes. This is particularly relevant for machine nodes belonging to different organizations, where sharing sensitive data is restricted by competitive and privacy considerations. Second, high costs associated with data centralization, including the need for high-bandwidth industrial internet and powerful data centers to process large volumes of data, conflict with the cost-efficiency goals of most enterprises. Federated learning offers a promising alternative, enabling global model training across multiple clients without requiring centralized data. Instead, lightweight, encrypted data—such as model parameters—is aggregated by a central server and shared with clients for further training. This architecture preserves data privacy while reducing transmission and computational costs. Federated learning approaches have recently been explored in multi-source transfer learning for fault diagnosis. For instance, adversarial networks have been employed to align domain distributions, while intermediate distributions or public datasets have been used to bridge domain gaps. However, existing federated methods face two significant limitations: First, most methods focus on marginal distribution alignment while neglecting conditional distributions in decentralized settings. This oversight is particularly problematic in scenarios involving distinct but related machines, where conditional distribution discrepancies remain unresolved. Second, the intermediate distributions used in prior works are often unrelated to the given domains or fixed, making them unsuitable for dynamic domain adaptation during federated training. To overcome these challenges, we propose a dynamic barycenter bridging network (DBBN) for federated multi-source transfer fault diagnosis in machine groups. DBBN introduces a federated learning framework that utilizes a central server to collect lightweight distribution parameters (e.g., means and covariances) from multiple source domains instead of raw data. Using these parameters, a dynamic distribution barycenter is calculated through fixed-point iteration and serves as an intermediate distribution to bridge diverse machine nodes. This barycenter is broadcast to all domain nodes, enabling targeted alignment of both marginal and conditional distributions during collaborative training. Unlike the static intermediates used in previous methods, DBBN dynamically adjusts the barycenter throughout the training process, ensuring effective adaptation to domain shifts. The contributions of this work are twofold. First, we propose a federated multi-source transfer learning framework, called DBBN, to facilitate lightweight and secure diagnosis knowledge transfer in machine groups. DBBN achieves directional alignment of multiple domain distributions by transmitting extremely lightweight data, such as distribution means and covariances, rather than high-dimensional features or raw data. Second, we introduce the Wasserstein barycenter as an intermediate distribution to bridge gaps between domains. By dynamically adjusting the barycenter during training, DBBN ensures its proximity to all domains, improving model generalization and transferability. Experimental results on a multi-source transfer fault diagnosis case using machine-used bearings demonstrate DBBN’s ability to improve adaptation accuracy and achieve superior fault diagnosis performance.

ABSTRACT. Embedded systems are vital to modern life and their reliable operation is crucial. While validation and verification during design are essential, they cannot fully address unexpected issues arising during operation, often due to environmental factors. To enhance system reliability, runtime monitoring is widely used to detect and address misbehavior in real-time and increase trustworthiness. Most runtime monitoring solutions, however, focus solely on functional or extra-functional aspects. This work introduces an assertion-based runtime monitoring approach using Time-Sensitive Behavioral Contracts (TSBCs), implemented as hardware components within a versatile VHDL library. The library provides low-level building blocks and readyto- use parameterizable monitor entities that simplify integration. To ensure minimal impact on the monitored system, the approach was evaluated for resource usage, performance, and power efficiency. In a motor inrush current monitoring scenario, the hardware solution reduced power consumption by 0.4% and achieved a frequency over 800 times higher than its software counterpart and comparable hardware solution.

ABSTRACT. Predicting the remaining useful life (RUL) is central for effective prognostics and health management. The current prognostics approaches for a network of assets are centralized, where aggregated data makes a richer dataset. When assets are located at different enterprises, data sharing is not preferred. One solution like federated learning, the local client data and training is on-site rather than being shared with a central server. For Phase I, we created benchmark results for centralized and decentralized learning approaches for RUL prediction. Benchmark studies are important since it gives researchers a benchmark and the ability to produce the same reported results. The results indicated that decentralized setups using Transformer as the final prediction model performed similar to centralized. Furthermore, previous prognostic prediction models have a single output RUL. For Phase II, we developed a single model cross-teaching multi-input-multi-output (MIMO) neural network framework, which accounts for more than one fault mode and produces simultaneously fault classifications and RUL predictions. The results indicated that this framework performed better against a non-cross-teaching MIMO, and single output approach. These findings help with proactive maintenance strategies, minimizing operational and maintenance disruptions while extending the shelf-life of assets.

ABSTRACT. This study develops a Mixed-Integer Linear Programming (MILP) model to optimize a manufacturing-remanufacturing-transport supply chain (MRTSC). The model integrates raw material procurement, transportation, inventory limits, and vehicle constraints which are factors often overlooked in previous research. It distinguishes between new and remanufactured products, leading to tailored manufacturing, transportation, and warehousing decisions. The model optimizes raw material procurement, production, remanufacturing, transportation, and returned product management under a carbon tax strategy, using carbon footprint as a sustainability metric. Implemented in CPLEX, the profit-maximizing model is validated through numerical results, demonstrating its effectiveness for MRSC (Manufacturing-Remanufacturing Supply Chain) optimization.

ABSTRACT. Despite the increasing number of studies dealing with mission abort policies (MAP), very few references considered the impact of maintenance on mission-abort decisions. This paper presents a novel model to jointly optimize selective maintenance scheduling and mission-abort decisions for mission-critical systems. The system is assumed to operate in an uncertain environment, which impacts its performance. Given that maintenance resources are limited, this paper develops a new integrated optimization approach where the classical SMP is extended to include mission-abort policies. In this approach, SM and mission-abort decisions are integrated into a single optimization model. A mixed-integer non-linear programming model is formulated to minimize expected total maintenance, mission failure, and penalty costs. A solution method is developed, and numerical experiments are then conducted to validate the proposed model and highlight its added value.

ABSTRACT. Mission-critical systems in sectors such as aerospace, defence, transportation, petrochemistry, and power generation require high reliability to prevent failures causing major economic losses, environmental damages, and safety risks. For such systems, solving the selective maintenance problem (SMP) yields optimal maintenance planning decisions during scheduled breaks. Its extension, the multi-mission SMP (MMSMP), focuses on optimizing component maintenance, maintenance levels, and repairperson assignments over multiple consecutive missions interspersed with maintenance breaks. While recent advances integrate predictive, resource-constrained, and fleet-wide strategies, they rely on the unrealistic assumption of fixed budgets, ignoring the reality of fluctuating and tight financial constraints faced by planners. This study investigates how different maintenance budget allocations across missions affect system performance. Using a two-phase decomposition model and binary integer programming, it explores various budget distribution strategies: uniform, linearly increasing, and inverted-V. The goal is to determine how allocating resources differently across missions can enhance asset reliability within fixed budget limits. The findings aim to guide maintenance planners in making budget decisions to improve overall system reliability while balancing resource constraints.

ABSTRACT. The rapid growth of e-commerce has significantly changed consumer purchasing behavior and preferences, pushing companies to adapt their supply model. Traditional single-channel supply chains have become inadequate, prompting businesses to integrate online platforms alongside their physical stores to survive in a competitive environment. This shift has led to the widespread adoption of dual-channel supply systems, allowing companies to expand their customer base and enhance market reach. Alongside direct selling, leasing has emerged as a viable alternative, offering customers flexible acquisition options. However, the management of leasing, remanufacturing and maintenance policies requires a well-defined strategy to ensure operational efficiency, reliability and sustainability of the different activities of the supply chain. This study focuses on optimizing operational strategies in dual-channel supply chains, particularly regarding pricing, remanufacturing, selling, leasing, maintenance and warranty. By simultaneously considering these activities and services, the studied system could describe numerous real cases of manufacturing and remanufacturing businesses in a dual-channel supply chain. This work aims to provide insights and support for managers and decision-makers by examining how integrating reverse logistics and different services, such as repair and warranty, into these systems can improve efficiency, sustainability, and reliability. The research examines a dual-channel supply chain comprising a Manufacturer, an Online platform and a Retailer. Both distribution channels allow the selling of new products and propose after-sales services such as repair and warranty. Conversely, the retailer proposes to lease products for a fixed duration and ensures its proper functioning by carrying out maintenance actions on the products during the leasing period. In this study, the Retailer performs periodic imperfect preventive maintenance directly on the leased products, and corrective maintenance action characterized by minimal repairs in case of failure. However, at the end of the leasing contract, the products are returned to the refurbishment process aiming to restore their state to as good as new, in order to get either resold or leased again. As a matter of fact, maintenance services are integrated to enhance customer satisfaction and extend product lifespans. The Manufacturer is responsible for the manufacturing activities of new products while also managing the refurbishment of leased products and the remanufacturing of products returned at the end of the life cycle. For each period, new products are obtained from directly transforming raw materials, and also integrating remanufactured subcomponents obtained from collected products in order to reduce the manufacturing costs. The main focus of this work is to develop pricing policies in a multi-periodic planning horizon for the selling and leasing processes taking into account customer preferences, market sensitivities and demands, and the costs of the proposed services. The optimization of the remanufacturing and maintenance strategies ensures the finest management of the different costs, encouraging competitive pricing in order to attract more customers while maintaining profitability, sustainability and reliability. The leasing pricing decisions are also optimized for each period based on market conditions and maintenance costs, ensuring that leasing remains an attractive option compared to direct purchase.

ABSTRACT. Modern football increasingly relies on data analysis and technological advancements, particularly artificial intelligence (AI) and machine learning (ML), to optimize performance, strategies, and player management. Recent research highlights various applications of these technologies, ranging from match result prediction to injury prevention, player performance evaluation, and collective strategy optimization. Match result prediction is a key area where researchers leverage historical data. Rodrigues and Pinto utilized models like random forests to analyze data from past matches, achieving an accuracy of 25.26% [1]. Anfilets, using deep multilayer neural networks, improved this accuracy to 61.14% [2]. Other works, such as those by Bomao, have applied several ML models to predict football match outcomes and scores. In-depth data analysis reveals that home teams have a significantly higher win rate compared to away teams [3]. These studies demonstrate progress but also highlight the limitations of these models in addressing football's unpredictability, which stems from factors such as player psychology, referee decisions, and environmental conditions. At the same time, injury prevention has become a major concern for clubs. Majumdar et al. developed models using physiological data to assess the risk of fatigue and injuries, enabling better training management [4]. FatigueNet, a model based on GPS data, integrates deep learning to predict player fatigue, thereby reducing injury risks while maintaining optimal performance [5]. These approaches not only help teams safeguard player health but also ensure that players perform at their peak during crucial matches, emphasizing the role of technology in modern sports science. Research also includes performance evaluation and estimating players' market values. Mustafa and Sakir proposed a method using machine learning to predict players' market value by incorporating objective data such as match performance [6]. Zanganeh et al. applied transfer learning techniques to improve these predictions, demonstrating increased reliability in evaluating player value [7]. These advancements allow clubs to make more informed decisions during player transfers, maximizing return on investments. Finally, other studies focus on optimizing collective strategies. Andriyanov compared player clustering models based on Gaussian mixtures and neural networks, revealing superior performance for Gaussian approaches [8]. Enhanced tracking systems now provide real-time data on player movements, which, when coupled with advanced models, offer coaches valuable insights to refine team strategies and adapt during matches. These studies highlight the potential of AI and ML to revolutionize football by providing powerful tools to analyze, predict, and enhance performance. In this regard, our work aims to address the challenges faced by coaches in player selection by developing a decision support tool for coaches, analysts, and recruiters. This tool will analyze both individual and collective performances, optimize tactical strategies, strengthen a team's weak areas, and facilitate the identification of ideal player profiles for recruitment. This approach enhances match efficiency, optimizes competitive performance, and significantly improves decision-making precision.

Références : [1] Rodrigues, F., & Pinto, Â. (2022). Prediction of football match results with Machine Learning. Procedia Computer Science, 204, 463-470. [2] Anfilets, S., Bezobrazov, S., Golovko, V., Sachenko, A., Komar, M., Dolny, R., ... & Oso- linskyi, O. (2020). Deep multilayer neural network for predicting the winner of football matches. International Journal of Computing, 19(1), 70-77. [3] Pan, B. (2025). Explore Machine Learning's Prediction of Football Games. In ITM Web of Conferences (Vol. 70, p. 04005). EDP Sciences. [4] Majumdar, A., Bakirov, R., Hodges, D., Scott, S., & Rees, T. (2022). Machine learning for understanding and predicting injuries in football. Sports Medicine-Open, 8(1), 73. for Understanding and Predicting Injuries in Football. Sports Medicine-Open, 8(1), 1-10. [5] Kim, J., Kim, H., Lee, J., Lee, J., Yoon, J., & Ko, S. K. (2022). A deep learning approach for fatigue prediction in sports using GPS data and rate of perceived exertion. IEEE Access, 10, 103056-103064. [6] Al-Asadi, M. A., & Tasdemır, S. (2022). Predict the value of football players using FIFA video game data and machine learning techniques. IEEE Access, 10, 22631-22645. [7] Zanganeh, A., Jampour, M., & Layeghi, K. (2022). IAUFD : A 100k images dataset for automatic football image/video analysis. IET Image Processing, 16(12), 3133-3142. [8] ndriyanov, N. (2020). Comparative analysis of football statistics data clustering algo- rithms based on deep learning and Gaussian mixture model. In CEUR Workshop Procee- dings (Vol. 2667, pp. 71-74).