The maritime industry requires advanced condition monitoring and predictive maintenance strategies to enhance the reliability of marine diesel engine systems. This study presents a digital twin framework that integrates vibration analysis, physics-based modeling, and deep learning for real-time fault diagnosis and remaining useful life (RUL) estimation. High-frequency vibration signals are processed and analyzed using a hybrid convolutional neural network–long short-term memory (CNN-LSTM) architecture with an attention mechanism to automatically classify engine faults and predict degradation trends. The framework combines data-driven methods with thermodynamic and structural digital twin models to improve generalization across operating conditions. Experimental validation on a medium-speed marine diesel engine under controlled fault scenarios demonstrates fault classification accuracy above 95% and remaining useful life prediction error below 8%. Early degradation signatures were detected up to 150 operating hours prior to critical failure. The proposed approach supports intelligent condition monitoring and decision-making for maintenance scheduling, reducing downtime and operational risk. This research demonstrates the effectiveness of integrating digital twin technology, vibration analysis, and CNN-LSTM deep learning models for predictive maintenance of marine diesel engines.

Recent advances in deep learning have greatly promoted remaining useful life (RUL) prediction of rotating machinery, yet challenges remain in effectively capturing multi-scale temporal patterns from noisy vibration signals. Existing models generally struggle to represent both long-term degradation evolution and short-term dynamic fluctuations, and their feature extraction is further hindered by substantial noise and redundancy in raw signals. Moreover, many current frameworks lack coordinated optimization of spatiotemporal information, limiting their prediction reliability under complex operating conditions. To address these limitations, this study proposes a hybrid multi-scale TCN–BiLSTM–SE (MTBS) model that integrates temporal convolutional networks (TCN) and bidirectional long short-term memory (BiLSTM) in a parallel architecture. The model incorporates multi-scale convolutional branches for hierarchical temporal representation, a semi-soft threshold denoising module for adaptive noise suppression, and a channel attention mechanism to highlight informative features. Extensive experiments on the IEEE PHM2012 and XJTU-SY bearing datasets validate the effectiveness of the proposed model. MTBS achieves consistent improvements across RMSE, MAE, and Score on both datasets, demonstrating clear advantages over traditional and recent advanced models. These results demonstrate that MTBS more effectively captures discriminative degradation patterns and substantially enhances prediction accuracy and robustness in end-to-end RUL estimation.

ABSTRACT Remaining useful life (RUL) prediction of aero-engines plays a critical role in ensuring operational safety and guiding maintenance decision-making. Although data-driven approaches have achieved remarkable progress in recent years, existing models still face challenges such as insufficient cross-task generalisation, lack of physical consistency, and limited interpretability in the prediction process. This paper proposes a meta-learning framework for lifetime prediction (TFMPINN) that integrates time-domain latent state modelling with a frequency-domain physics-informed neural network. Specifically, a Temporal State Modelling Encoder (TSME) is constructed to build a continuous latent space via reparameterization, enabling capture of key dynamic features and uncertainty throughout the degradation process. Besides, the Fourier Physics-Constrained Network (FPCN) is designed to encode frequency-domain physical priors by introducing implicit PDE residual constraints, enhancing the model’s ability to capture consistent degradation dynamics, while an Adaptive Weighted Channels (AWC) module in the predictor adaptively fuses multi-sensor degradation features. By incorporating a meta-learning mechanism optimised over task distributions, the framework achieves rapid adaptation and improved generalisation under few-shot and varying operating conditions. Experiments on NASA C-MAPSS and N-CMAPSS turbofan engine datasets demonstrate that TFMPINN outperforms existing methods in prediction accuracy, cross-task generalisation, and interpretability of the degradation process.

Artificial Intelligence (AI) has demonstrated to be an effective method for predicting irregularities across various Industrial processes. In recent years, the development of predictive maintenance systems using AI techniques has attracted many researchers around the world. Maintenance planning has been effective with Remaining useful life prediction using AI in bearings. This review brings consideration to the role of AI in predicting the remaining useful life (RUL) of bearing components in various industrial processes. A systematic search was carried out in electronic databases such as Springer, IEEE, Elsevier, and the ACM Digital Library, with an emphasis on AI-based approaches for bearing RUL prediction. A brief summary of previous works is presented to show the development of technological advancement in this field of RUL prediction of roller bearing. Specifically, this review examines the types of bearing components studied, the sample sizes used for training AI models, the signal processing method and classification algorithm applied, and the outcomes achieved. The outcome of this review shows that hybrid approaches and deep learning models achieve better performance in predicting RUL in roller bearings. Finally, the review finds existing research gaps and provides recommendations for future improvements, aiming to guide future researchers toward more accurate and reliable RUL prediction models for bearings.

ABSTRACT Deep learning-based methods for Remaining Useful Life (RUL) prediction generally suffer from insufficient predictive dynamic stability and generalisation capability under real-world industrial conditions. To address this, this paper proposes the Degradation Manifold Dynamic Consistency Network (DMDCN). We initially define a differentiable embedded degradation manifold as the feature representation space. Addressing the limitation that existing embedding methods do not guarantee temporal evolution consistency, a dynamic consistency learning framework is devised to reframe RUL assessment as a joint inference problem of state estimation and dynamics estimation. Through the introduction of the manifold’s local geometric derivatives as the input domain for the dynamics estimator, dynamic consistency between the system’s state representation and its intrinsic evolutionary trend is achieved. Subsequently, this deterministic framework is extended to an approximate Bayesian paradigm, enabling uncertainty quantification via variational inference and Kernel Density Estimation. Empirical results on the public C-MAPSS dataset and a real-world industrial slurry pump dataset indicate that DMDCN improves prediction accuracy and dynamic stability compared to baseline models, validating the method’s potential for applications in high-reliability industrial scenarios.

Reliable prediction of the Remaining Useful Life (RUL) of lithium-ion batteries (LIBs) plays a pivotal role in maintaining safe operation, enhancing system dependability, and supporting economically sustainable lifecycle planning in electric mobility and stationary energy storage applications. However, battery aging is governed by highly nonlinear, interacting, and chemistry-dependent processes, which pose significant challenges for conventional data-driven prognostic models. In this study, a unified RUL prediction framework is proposed by integrating multi-domain feature engineering, a Multi-Criteria Adaptive Selection (MCAS) strategy, and a Bidirectional Long Short-Term Memory (Bi-LSTM) network enhanced with dual multi-head attention. Degradation-relevant descriptors extracted from time, frequency, and chaotic domains are employed to capture complementary aging dynamics across battery cycling. In addition, a novel degradation-consistency indicator, termed the M-score, is introduced to characterize the regularity and stability of degradation behavior using observable electrical, thermal, and statistical signals. The MCAS mechanism systematically identifies informative and temporally stable features while suppressing redundancy, thereby improving both predictive robustness and interpretability. The resulting architecture jointly exploits adaptive feature refinement and attention-based temporal modeling to enhance the RUL estimation accuracy. The proposed framework is validated using two widely adopted benchmark datasets: the Toyota Research Institute (TRI) dataset, representing fast-charging lithium iron phosphate (LFP) cells, and the Sandia National Laboratories (SNL) dataset, which includes multiple chemistries, such as LFP, NMC, and NCA. Experimental results demonstrate substantial improvements in the RUL prediction accuracy compared with baseline Bi-LSTM and single-attention models, while systematic ablation studies confirm the individual contributions of the M-score and MCAS components. Within the evaluated datasets and operating conditions, the results suggest that the proposed framework offers a robust and interpretable data-driven solution for battery RUL estimation. However, extending its generalizability and validating its performance on unseen datasets and in real-world scenarios remain important areas for future research.

Estimating the remaining useful life (RUL) of engineering systems is crucial for maintenance planning and the reliability of complex mechanical units. Accurate RUL predictions support timely interventions and help to prevent unexpected failures. This study proposes a statistically robust framework that models degradation signals up to the end of life using a hidden Markov model (HMM) with a simple-failure structure and an absorbing terminal state. The proposed method estimates state-dependent linear emission parameters and transition probabilities using a ridge-regularized expectation-maximization (EM) algorithm. The ridge penalty stabilizes slope estimates under limited data, while a robust Huber-based scale estimator reduces sensitivity to outliers in the sensor-derived health indicator. RUL is computed as a weighted expected time to absorption, combining transient-state survival characteristics with smoothed posterior-state probabilities obtained via the forward-backward algorithm. This yields a low-variance state-aware estimator that preserves the probabilistic structure of the HMM. Simulation studies show that the proposed ridge-regularized EM significantly reduces parameter variance and improves predictive accuracy compared with the baseline weighted least squares EM (WLS-EM). A real-data case analysis demonstrates further improvements in RUL estimation accuracy and smoother, more reliable prediction trajectories. Overall, the framework provides a robust and interpretable approach for practical prognostics applications.

Abstract Existing intelligent operation and maintenance (O&M) models require separately establishing regression and classification models for Remaining Useful Life (RUL) prediction and fault diagnosis near the end of adjacent equipment degradation, making it difficult to simultaneously address both tasks and establish an effective associative model with low information utilization efficiency. We propose a dual-task O&M model for the servo turret power head, which dynamically combines multi-scale shared features and private features. This method adopts a "shared-private" collaborative learning approach, processing dual tasks in parallel through a modular structure to ensure task information interaction while maintaining feature independence. The model comprises three core modules: a public multi-scale shared feature extraction module that utilizes a multi-scale convolutional kernel parallel architecture to mine multi-scale features, and outputs robust shared features after feature fusion and encoding; a dynamic private feature module that employs a dual-path parallel structure to construct independent branches for classification and regression tasks, outputting private features tailored to task requirements through "spatial-channel" dual-path feature extraction and adaptive weighted fusion; and a dual-task decision-making module that includes a classifier and a regressor to independently map private features.Experimental validation demonstrates that the proposed method achieves an average RMSE of 0.087 in RUL prediction and an average diagnostic accuracy of 98.52% in fault diagnosis for the servo turret power head. This proves that the proposed method can effectively accomplish both RUL prediction and diagnostic tasks simultaneously.

Abstract Rolling bearing Remaining Useful Life (RUL) prediction plays a critical role in equipment health management. However, noise in practical engineering applications can submerge early weak degradation features, making it difficult to capture early degradation patterns. This constrains the application of prediction methods across all degradation stages.To address these issues, this paper proposes a rolling bearing RUL prediction method based on a Vibration Signal Reconstruction-enhanced Method (VSRM) and a Time-Adaptive Fusion Network (TAFN). First, VSRM performs decoupled learning and non-linear reconstruction of the signal's amplitude and phase spectra in the frequency domain. This effectively suppresses noise and significantly enhances the robustness and representational power of degradation features. Subsequently, TAFN introduces timestamp encoding as an explicit global temporal prior and designs a temporal-position-driven adaptive weighting mechanism. This achieves a dynamic fusion of global trends and local features, thereby resolving the lag problem that existing models face when capturing non-linear changes in degradation rates. In this work, the VSRM-TAFN framework is deeply integrated with several mainstream time-series prediction networks and comprehensively validated on two full-life-cycle rolling bearing datasets. The results demonstrate that the proposed VSRM-TAFN framework significantly improves the prediction accuracy of all mainstream time-series networks, achieving a minimum RMSE of 0.042 and a minimum MAE of 0.0327. This general architecture effectively overcomes the challenges posed by noise and non-linear degradation rate variations, providing an effective and universal solution for achieving high-robustness RUL prediction.

System-level prognostics is crucial for ensuring reliability and enabling predictive maintenance in complex systems with interconnected components. This study presents a framework that integrates data-driven methods to predict the remaining useful life (RUL) of a subsystem under multiple and concurrent faults within a nuclear power plant system with explainable artificial intelligence (XAI). A nuclear power plant (NPP) operation was simulated to model the degradation behavior of NPP components, and four machine learning models—gradient boosting regressor (GBR), support vector regressor (SVR), fully connected neural network (FCNN), and long short-term memory (LSTM)—were evaluated for prognostics with a novel system RUL parameter. The LSTM model demonstrated potential superior repeatability, while SHAP (SHapley Additive exPlanations) for explainability provided consistent and trustworthy global explanations. In contrast, LIME (Local Interpretable Model-agnostic Explanations) offered localized interpretability but showed reduced stability for sequential data. Key findings include the interplay between component-level degradation and system-wide performance, with LSTM effectively capturing these dynamics through sequence-level predictions. The XAI techniques enhanced transparency by identifying critical features influencing model predictions and aligning with domain knowledge. This framework has significant implications for improving trust and understanding in predictive maintenance, particularly in safety-critical industries such as nuclear energy.

Early and accurate prediction of the remaining useful life (RUL), defined as the number of operational cycles a battery can continue to function before reaching its end-of-life threshold, is crucial for improving the reliability of new energy vehicles. To address noise contamination, capacity regeneration effects, and data scarcity in early-stage prognostics, this paper proposes a hybrid framework integrating signal decomposition, time series augmentation, and deep forecasting. The raw capacity sequence is decomposed using Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) to separate multi-scale components. A Transformer-enhanced time series generative adversarial network (HyT-GAN) is then employed to augment decomposed components, improving robustness under small-sample conditions. A CNN-BiGRU predictor is trained for capacity forecasting, and key hyperparameters are tuned via the Dung Beetle Optimizer (DBO). Experiments on NASA and CALCE benchmark datasets demonstrate that the proposed method achieves accurate early-stage prediction using only 20% historical data, with R2 ranging from 0.9643 to 0.9972 and RMSE/MAE below 0.0296/0.0198. These results indicate that the proposed framework can deliver reliable RUL estimates under data-limited and noisy measurement conditions.

Abstract Accurate prediction of the Remaining Useful Life (RUL) of rolling bearings is crucial for the safe and efficient operation of mechanical equipment. However, most existing studies only predict the RUL in ideal or in low-noise environments, and fail to take into account the complex noise interference in real environments. Therefore, a novel multiscale dilated attention convolutional neural network (MSDA-CNN) is proposed to predict RUL of rolling bearings. In our approach, the multiscale dilation convolution module (MDCM) leverages dilated causal convolutions to capture degradation characteristics across multiple scales. Furthermore, we incorporate a dynamic attention module (DAM) into the MDCM to refine feature selection and highlight key degradation patterns. To further enhance feature representation, a selective attention fusion module (SAFM) is designed to adaptively calibrate and fuse cross-scale features while suppressing redundant and noisy information. In addition, the dynamic feature fusion module (DFFM) integrates multi-path features through convex-weighted gating instead of high-dimensional concatenation, thereby improving efficiency. Extensive experiments on the PHM 2012 bearing dataset under varying noise scenarios demonstrate that the proposed MSDA-CNN substantially outperforms several advanced models in terms of prediction accuracy and noise robustness, thereby confirming its effectiveness and superiority for industrial applications.

Purpose The purpose of this study is to develop a scalable and privacy-preserving predictive maintenance framework for wind farms by addressing the challenges of heterogeneous multi-component degradation. By combining stochastic physics-based models with machine learning techniques under a federated learning (FL) paradigm, the framework aims to provide accurate estimation of remaining useful life (RUL) for critical components while preserving data confidentiality. Furthermore, a multi-objective optimization layer leverages prognostic outputs to minimize maintenance costs and maximize turbine availability, offering an intelligent solution for enhancing reliability and efficiency in modern wind energy infrastructures. Design/methodology/approach This study proposes an integrated framework for predictive maintenance in wind farms, addressing multi-component degradation through the combination of FL and hybrid degradation models. Under variable wind and load conditions, critical wind turbine components such as gearboxes, bearings and generators exhibit heterogeneous and nonlinear degradation behaviors that challenge conventional maintenance approaches. The proposed framework integrates stochastic physics-based degradation modeling with data-driven learning techniques to enable accurate and interpretable estimation of the RUL at the component level. FL is employed to facilitate decentralized model training while preserving data privacy, allowing turbine-level models to collaboratively improve prognostic performance without the need for centralized data aggregation. Simulation-based evaluation demonstrates that the proposed adaptive framework achieves a reduction in cumulative wear of approximately 35% compared with static maintenance strategies, while increasing the Weibull characteristic life by nearly 49%. In addition, the framework reduces total maintenance cost by about 22%, improves estimation accuracy by lowering the Kalman filter mean absolute error by approximately 44% and increases system availability by more than 140% relative to static policies. These results confirm that the proposed approach provides a scalable, privacy-preserving and cost-efficient solution for predictive maintenance and reliability enhancement in distributed wind farm environments. Findings The study demonstrates that integrating hybrid degradation models with FL significantly improves the accuracy of RUL estimation for wind turbine components under heterogeneous operating conditions. Simulation results reveal that the proposed framework reduces wear progression, lowers maintenance costs and enhances overall system reliability compared to traditional centralized or periodic strategies. The federated approach ensures data privacy and reduces communication overhead, while the optimization layer effectively balances cost efficiency and turbine availability. These findings highlight the framework’s potential as a scalable, intelligent and privacy-preserving solution for predictive maintenance in modern wind farms. Research limitations/implications This study is based on simulation experiments and may not fully capture the complexities of real-world wind farm operations, such as unmodeled environmental influences or rare failure modes. The FL framework requires consistent data quality and communication infrastructure across turbines, which may limit applicability in regions with poor connectivity. Future work should involve large-scale field validation, incorporation of more diverse degradation mechanisms and integration with real-time supervisory control systems. Despite these limitations, the framework offers a strong foundation for advancing intelligent, distributed and privacy-preserving predictive maintenance in modern wind energy infrastructures. Practical implications The proposed framework provides wind farm operators with a scalable tool to optimize maintenance scheduling while reducing downtime and operational costs. By accurately estimating the RUL of critical components such as gearboxes, bearings and generators, operators can shift from fixed-interval maintenance to condition-based strategies, extending component lifespan and improving asset reliability. The FL approach preserves data privacy, enabling collaboration across multiple sites without centralizing sensitive operational data. This enhances trust and data security while supporting widespread adoption, making the framework highly applicable for industrial deployment in modern wind energy infrastructures. Social implications The adoption of intelligent predictive maintenance in wind farms contributes to greater energy security and sustainability by ensuring higher system reliability and reduced downtime. Improved efficiency in wind energy production supports the global transition toward clean energy, lowering dependence on fossil fuels and reducing greenhouse gas emissions. By preserving data privacy through FL, the framework also fosters trust and collaboration between operators, regulators and stakeholders. Ultimately, this approach strengthens public confidence in renewable energy technologies, promotes green job creation and accelerates the achievement of climate and sustainability goals at both regional and global levels. Originality/value This study introduces a novel predictive maintenance framework that integrates hybrid degradation models with FL to address the challenges of heterogeneous, multi-component wear in wind farms. Unlike traditional centralized or periodic strategies, the framework enables accurate and interpretable RUL predictions while preserving data privacy and reducing communication costs. The addition of a multi-objective optimization layer ensures cost-efficient and availability-driven maintenance scheduling. This combination of decentralized learning, physics-informed modeling and optimization provides a scalable, intelligent and privacy-preserving solution, offering significant value for advancing maintenance practices in modern renewable energy infrastructures.

Abstract Accurate remaining useful life (RUL) prediction is essential for predictive maintenance. However, incomplete sensor data disrupts spatio-temporal dependencies and degrades prediction accuracy. The existing two-stage methods generate the amplification of errors by decoupling interpolation and prediction, and the end-to-end models struggle to capture both spatial coupling and temporal degradation continuity. Moreover, fixed task weights further limit adaptability under varying levels of missing data. To overcome these challenges, this paper proposes a multi-task spatio-temporal residual shrinkage network (ST-RSMTNet) combined a spatio-temporal residual shrinkage block (ST-RSBU) with a progressive multi-task learning strategy. The introduction of ST-RSBU adaptively filters degradation-related features across sensor channels and restores interrupted temporal trends via global sequence statistics. The proposed multi-task framework jointly optimizes imputation and prediction within an "imputation-predict" closed loop. Extensive experiments on the C-MAPSS and PHM2010 datasets show that the ST-RSMTNet maintains robustness under high missing rates. It achieves RMSE reductions of 7.29% on FD001 and 10.83% on PHM2010 C1 compared with the second-best approach. This network outperforms most state-of-the-art methods. The results demonstrate that the ST-RSMTNet is an effective solution for industrial prognostics under incomplete sensing condition. This work proposes a novel solution to the data deficiency challenges and provides crucial theoretical support for reliability of complex industrial systems in the future.

Purpose: The rechargeable batteries are their major element where the energy-storage systems are central to the modern power networks, electric transportation, and the portable electronic devices. The possibility to evaluate the battery condition and estimate the degradation with time is the key to the performance, reliability, and safety of these systems. Materials and Methods: Two significant measures of such a purpose are the state of health (SOH), which is the present capacity or power capability compared to original specifications, and the remaining useful life (RUL), which is an approximation of operation life until the battery fulfills end-of-life conditions. SOH and RUL because predictability is necessary in order to manage the battery, preventive maintenance, and cost-efficient system operation. The degradation of batteries is dictated by complex electrochemical and mechanical mechanisms with dependence on the conditions of operation like temperature, rate of charge, depth of discharge and patterns of usage. These time-varying nonlinearities are very difficult to deal with through conventional estimation methods. In order to overcome these issues, a broad selection of prognostic techniques has been designed, which can be narrowed down into model-based, data-driven, and hybrid. Model-based approaches are based on physical and electrochemical models of battery behavior, providing interpretability but in most cases, these models are sensitive to the identification of accurate parameters. Machine learning and deep learning models are data-driven approaches that allow the establishment of complex degradation trends at high levels of predictive accuracy using past operational data. Hybrid frameworks strive to build the merits of the two paradigms by blending physical wisdom and data-driven flexibility. Findings: When comparing previous research on the estimation of battery health and the remaining useful life, it becomes apparent that the performance trends are similar in various methodological types. Although, there is no universal method to be used in all of the operating conditions, the literature provides definite advantages and disadvantages related to model-based, data-driven, and hybrid prognostic methods. Unique Contribution to Theory, Practice, and Policy: The article is a thorough piece of work that provides an evaluation of battery health prediction and RUL estimation approaches both in terms of their principles of operation, implementation strategies, and performance attributes.

Purpose: Substation equipment and power transformers constitute vital parts of the modern power systems, and its proper functioning is the key to the system stability, efficiency, and safety. Sensors predictive maintenance uses predictive maintenance based on sensor data which is a time-series of data to propose proactive methods to monitor asset health, anomaly detection and predictive maintenance, thus minimizing unplanned outages and maximizing maintenance schedules. Materials and Methods: This review gives an overall overview of sensor technologies, data features and modeling techniques used in predictive maintenance of transformers and substation equipment. The classical models of statistics, machine learning methods, and deep learning systems are addressed in terms of condition monitoring, anomaly detection, and remaining useful life estimation. Problems such as data quality, model interpretability and deployment concerns are discussed and future research directions such as digital twins, physics-informed models, Edge-AI and secure cloud-edge are identified to inform the further development of the field. Findings: Additionally, the review highlights predictive models for estimating the remaining useful life (RUL) of assets to optimize maintenance planning. Unique Contribution to Theory, Practice, and Policy: This review provides a comprehensive understanding of predictive maintenance techniques for transformers and substation equipment. It contributes to theory by summarizing and evaluating various models and methods used in the field. In practice, it offers insight into current and future technologies for asset management and maintenance. The identification of future research areas like digital twins, Edge-AI, and secure cloud-edge will help to drive future developments and influence policy in the power systems sector.

The performance degradation of electronic power components during long-term operation can compromise system reliability and safety. Therefore, accurately predicting their remaining useful life (RUL) is critical for the reliability of safety-critical systems that utilize these components. This paper proposes a hybrid model integrating bidirectional long short-term memory networks (BiLSTM) and Gaussian process regression (GPR) for RUL prediction of electronic power components. The BiLSTM module provides high-precision point predictions, while the GPR module leverages the sequence features and trend information extracted by BiLSTM to deliver reliable interval predictions and high-confidence probabilistic outputs. The model’s predictive accuracy was validated using NASA’s publicly available lithium-ion battery dataset. Experimental results demonstrate that, compared to existing models, the proposed model achieves at least a 9.6% improvement in point prediction performance and a 63% improvement in interval prediction performance, fully validating the reliability and accuracy of the BiLSTM-GPR approach. The model was further applied to predict the RUL of DC-DC power modules. The predicted Continuous Ranked Probability Score (CRPS) reached a maximum of 0.050405, while the Probability Integral Transform (PIT) results exhibited a uniform distribution within the (0,1) range, further demonstrating the model’s high reliability and predictive confidence.

The rapid adoption of electric vehicles (EVs) has highlighted the need for intelligent systems that optimize performance, extend battery life, and ensure sustainable mobility. This study investigated AI-based energy management, battery health prediction, and adaptive control strategies for electrified mobility systems. A hybrid approach integrating machine learning, reinforcement learning, and predictive analytics was employed to monitor real-time driving conditions, forecast battery state-of-health (SoH) and remaining useful life (RUL), and dynamically adjust energy distribution. The methodology involved simulation-based evaluations across urban, highway, and mixed driving cycles to assess energy efficiency, system responsiveness, and predictive accuracy. Results demonstrated that AI-driven energy management significantly reduced energy losses during acceleration and deceleration, while predictive models accurately anticipated battery degradation, enabling proactive maintenance interventions. Adaptive control mechanisms improved vehicle stability, optimized load distribution, and minimized battery stress during dynamic driving scenarios. Comparative analysis indicated that AI-based systems outperformed conventional rule-based strategies in terms of efficiency, reliability, and scalability. These findings underscore the potential of intelligent electrified mobility systems to enhance operational performance, prolong battery lifespan, and support sustainable transportation solutions. Future implementations are recommended to integrate explainable AI techniques and real-world validation to further improve transparency, reliability, and adoption. Overall, the study establishes a framework for AI-enabled EV systems, highlighting their transformative role in achieving energy-efficient, adaptive, and resilient electrified mobility.

Abstract Rolling bearings, as a key transmission component of rotating machinery, require accurate life prediction and effective health management to enable intelligent operation and maintenance and ensure system reliability. A small-sample remaining useful life (RUL) prediction approach for rolling bearings based on meta-transfer learning is proposed in this paper. By fusing model-agnostic meta-learning (MAML) and domain adversarial neural networks (DANN), a MAML-DANN transfer learning (MDTL) framework is constructed to address the dual challenges of few-shot adaptation and cross-domain alignment. To enhance MAML’s small-sample adaptability, an outer-loop cosine annealing weight allocation strategy is designed to dynamically balance training priorities between task adaptation and domain alignment. For DANN, a feature spectrum penalty (FSP) regularization is introduced to constrain singular values of source/target domain features, preserving domain-invariant degradation information without compromising prediction performance. Combined with the maximum mean discrepancy (MMD) loss function, the model further reduces cross-domain distribution differences. Validated on IEEE PHM 2012 and ABLT-1A datasets, the proposed MDTL method reduces average RMSE by at least 31.45% compared to baselines (e.g., MAML, MAML-MMD). The results demonstrate its superiority in small-sample and varying-operating-condition bearing RUL prediction, providing a practical solution for industrial health management.

The accurate prediction of remaining useful life (RUL) of electric vehicle (EV) batteries is a critical aspect of intelligent battery management systems. Effective RUL prediction not only ensures vehicle safety and reliability but also plays a pivotal role in optimizing charging cycles, reducing maintenance costs, and extending the overall battery lifespan. This work presents a comprehensive Deep Learning (DL) framework for predicting RUL of EV batteries, using a novel Hyperscale-Cascaded Transformer Net architecture designed to capture long-term dependencies and degradation patterns in battery behavior. The proposed system initiates with data acquisition, wherein parameters such as cycle index, voltage, current, and time-based features are collected. Raw data undergoes preprocessing, which includes data cleaning to eliminate outliers and handle missing values, followed by Exploratory Data Analysis (EDA) to extract meaningful patterns through descriptive statistics, distribution analysis, and correlation heatmaps. Subsequently, the data is passed through a feature engineering pipeline, where feature scaling using Min-Max normalization is applied to enhance learning efficiency of model. Processed dataset is then split into training and testing sets, maintaining data integrity for unbiased evaluation. The core of the model lies in Hyperscale-Cascaded Transformer Net, a DL model that utilizes cascaded transformer layers to model complex temporal relationships and nonlinear degradation behaviors inherent in battery performance over time. Experimental esults demonstrate that proposed Transformer-based model outperforms traditional Machine Learning (ML) techniques in terms of accuracy and robustness in revolutionizing EV battery management systems.