An Intelligent Enterprise Asset Management Framework for Railway Maintenance Prioritization: A Machine-learning Proof of Concept Using Benchmark Analogue Data

Integrating Machine Learning and AHP for Data-Driven Asset Renewal Prioritization in Power Distribution Systems

This paper proposes an integrated framework for predictive asset management in electric power distribution systems by combining machine learning with the Analytic Hierarchy Process (AHP). Distinct utility datasets were used according to the characteristics of each prediction task. Annual preventive- and corrective-maintenance records, including maintenance cost and frequency, covered the period from 2016 to 2023, whereas the SAIDI and SAIFI annual time series covered the period from 2014 to 2023. Remaining Useful Life (RUL) estimation was based on static asset-level records from the entire available equipment stock and therefore was not associated with a single time-series interval. Multiple machine learning algorithms were evaluated independently for each Type of Utility Component (TUC), and the best-performing models were selected according to their validation errors. The resulting predictions were incorporated into a two-level AHP hierarchy that integrates technical, economic, operational, and regulatory criteria for portfolio-level asset-renewal prioritization. For RUL estimation, the selected TUC-specific models reduced the mean MAE from 11.33 years for the historical-mean baseline to 2.70 years, corresponding to an absolute improvement of 8.63 years and a relative reduction of 76.15%. The resulting priority ranking was evaluated retrospectively using moving-average, cumulative-gain, and lift analyses. The top 20% of ranked assets captured 29.61% of the observed corrective-maintenance occurrences, corresponding to a lift of 1.4806 relative to random selection. The cumulative-gain curve yielded an area under the curve of 0.6337 and a ranking Gini coefficient of 0.2674, indicating moderate positive discrimination across the asset portfolio. These results show that the proposed ML–AHP framework can support the allocation of inspection, engineering-assessment, and renewal resources toward assets with comparatively greater maintenance demand and operational relevance. The framework provides a structured, data-driven decision-support approach for utility asset-renewal planning, while prospective field implementation remains necessary to quantify reductions in failures, corrective maintenance, continuity penalties, and total expenditure.

A Machine Learning-Based Predictive Model for Maintenance Management of Combustion Engines in the Agricultural Sector

Maintenance management of stationary combustion engines in the agricultural sector remains largely manual, increasing the risk of unplanned downtime. This study developed a machine learning-based predictive model to anticipate failures within a 60-day horizon, enabling the transition from reactive to proactive maintenance. Following the CRISP-DM (Cross-Industry Standard Process for Data Mining) framework, a sliding-window feature engineering pipeline was built from 2250 historical records spanning 59 engines. Four ensemble learners (Random Forest, LightGBM, XGBoost, and CatBoost) were then compared under two complementary protocols: a strict 60/40 chronological split simulating deployment, and a stratified leave-engine-group-out cross-validation withholding entire engines from training. Nonparametric testing (DeLong test and engine-level cluster bootstrap) showed that the four learners are statistically equivalent, whereas the feature engineering layer contributes a large, significant discrimination gain (ΔAUC ≈+0.06, p≈10−27) on engines unseen during training. Random Forest, selected as the final model, achieved an AUC of 0.90 with 84.2% recall under the deployment protocol and 0.96 with 90.7% recall under engine-grouped validation. Temporal extrapolation, rather than cross-engine generalization, appeared to be the primary challenge, indicating that rigorously engineered degradation features, more than the choice of ensemble algorithm, drive predictive performance in agricultural maintenance planning.

A Lightweight Machine Learning Framework Predictive Maintenance and Operational Safety in Offshore Energy Systems

Mechanical failures in critical offshore components, such as pumps and compressors, constitute a major challenge and can contribute non-productive causes to the field in deepwater operations, estimated to account for up to 20% of total downtime, with substantial production and environmental impacts. Existing data-driven predictive maintenance solutions are often computationally intensive and ill-suited for operators with limited digital infrastructure. To address these limitations, we developed a novel, lightweight predictive maintenance framework that combines gradient-boosted decision trees (LightGBM) with real-time anomaly detection to forecast equipment failure risks in resource-constrained offshore environments. This solution, called OffshoreGuard, is validated on the NASA C-MAPSS FD001 benchmark dataset and integrates a prototype dashboard with a Large Language Model (LLM) interface to enable natural-language querying and automated maintenance report generation. The framework achieved a risk classification F1 score of 86.58% at 92.29% precision, a RUL estimation RMSE of 18.42 cycles with a relative error of 18.55%, and an anomaly detection F1 score of 73.39%. With a total serialised model footprint of 1.6 MB, OffshoreGuard is deployable on standard web infrastructure at zero licensing cost, offering an affordable and scalable solution for asset managers in the West African energy sector to enhance operational safety, reduce non-productive time, and extend equipment life cycles.

Interchangeability and cannibalization in industrial maintenance: perspectives for asset management with machine learning, big data and artificial intelligence

Automating Infrastructure Maintenance Contracting: A Hybrid Markov–Machine Learning Framework for Predictive Asset Management and Risk Financing

Uncertainty in infrastructure deterioration and maintenance cost forecasting continues to weaken the effectiveness of conventional maintenance contracts, contributing to budget overruns, delayed interventions, and inefficient risk allocation. This study develops a machine learning (ML)–enhanced adaptive performance-based maintenance contract (APBMC) framework that integrates hybrid Markov–ML deterioration forecasting, dynamic payment adjustment logic, and parametric risk-sharing provisions within a unified decision support structure for bridge maintenance. The framework is calibrated using a 10-year dataset from 68 US bridges (2014–2023), incorporating inspection records, structural health monitoring (SHM) summaries, traffic loading, maintenance histories, and climate-exposure variables. A hybrid Markov–ML architecture is used to predict condition-state transitions and quantify uncertainty, and the resulting forecasts are translated into contract trigger points, adaptive maintenance responses, payment adjustments, and risk financing activation rules. The framework is evaluated through historical back-testing, scenario-based simulation, and sensitivity analysis of trigger thresholds, calibration settings, and risk-sharing configurations. Results indicate 91% overall predictive accuracy, with calibration error below 3% in the contract-relevant trigger region. Relative to the lump-sum baseline, the proposed APBMC framework reduces mean life-cycle maintenance cost by 22%, improves response time to critical repairs by 31%, and lowers major rehabilitation frequency by up to 18%. The integrated risk financing layer performs effectively under both condition-based triggers, with annual activation probabilities of 0.14–0.36, and flood-based triggers, with probabilities of 0.05–0.09, reducing the likelihood of unfunded major rehabilitation from 18% to below 5%. Sensitivity analysis further identifies stable operating regions for condition thresholds and model-mixing settings that improve cost certainty while maintaining resilience. The proposed APBMC framework demonstrates the value of coupling predictive analytics, adaptive contracting, and risk financing to support more resilient, transparent, and performance-oriented infrastructure maintenance governance.

Improving the Productivity of a Production Line through Explainable Artificial Intelligence for Data Analysis

Improving the performance of industrial production systems through data analysis is a recurrent topic in both research and industry. However, most existing data-driven approaches in manufacturing focus on engineer level decision support, such as predictive maintenance or strategic resource planning and pay little attention to operator-centred decision support. At the same time, production data exhibit specific characteristics that make their analysis challenging: high-dimensionality, often heterogeneous, and naturally structured as time series and point processes. These gathered properties require dedicated modelling strategies insufficiently addressed in the literature.An additional and increasingly recognised challenge is the adoption of digital decision-support tools on the shop floor. Operators and line adjusters need to trust the models’ outputs in order to integrate them into their daily practices. This calls for the design of explainable artificial intelligence approaches made for production contexts and for the needs of human operators.The proposed approach is developed in a real factory, to support operators decisions based on the state of the production system.. We detail the decision model building process from dimensional reduction to combination with field knowledge.Our goal is to build models that learn to recognise production situations similar to past ones that have already led to stops, so as to suggest relevant corrective actions to implement. A key aspect of our approach is the continuous confrontation of the modelling choices with field experience, in order to provide interpretable reasons for the recommendations.