Jul 2026· Transactions of the Institute of Measurement and Control· 0 citations· 28 references
TL;DR
An experimental/synthetic hybrid, data-driven FDI framework that leverages supervised machine learning (ML) integrated with an experimentally validated second-order electro-thermal battery model to generate a mixed experimental–synthetic dataset enables fast, real-time diagnosis of complex multi-fault scenarios at the cell or module level in series–parallel LIB pack architectures.
Abstract
Accurate and reliable fault detection and isolation (FDI) in lithium-ion battery (LIB) systems is essential for ensuring safety, optimal performance, and lifespan, particularly as LIBs power critical applications such as electric vehicles and renewable energy storage. Traditional model-based diagnostics face limitations in handling sudden and compound faults due to modeling uncertainties and transient dynamics. This study proposes an experimental/synthetic hybrid, data-driven FDI framework that leverages supervised machine learning (ML) integrated with an experimentally validated second-order electro-thermal battery model to generate a mixed experimental–synthetic dataset covering healthy operation and diverse fault conditions, including cell, connection, and sensor faults. Time-domain statistical features are extracted from current, voltage, and temperature measurements, and nature-inspired wrapper-based algorithms are employed for feature selection, while multiple ML classifiers—Random Forest (RF), Decision Trees (DTs), K-Nearest Neighbors (KNN), and Naive Bayes (NB)—are comparatively evaluated on a six-class diagnosis problem. The best-performing configurations achieve near-perfect multi-class accuracy of approximately 99.80–99.87% with low variance across repeated train–test splits, demonstrating high robustness and stability overall. Rather than identifying a single universally optimal model, the results indicate a practical trade-off between diagnostic performance and implementation cost. Within this group, the RF classifier combined with Particle Swarm Optimization (PSO) for feature selection (RF–PSO) attains slightly higher accuracy and stability than the other top combinations, while DT–PSO and KNN–MPA (Marine Predators Algorithm) yield much smaller models, shorter inference times, and more compact feature subsets, which are attractive for constrained embedded implementations and large pack deployments. The proposed framework thus enables fast, real-time diagnosis of complex multi-fault scenarios at the cell or module level in series–parallel LIB pack architectures and allows practitioners to select the most suitable classifier–optimizer pair according to application-specific priorities on accuracy, robustness, computational cost, and diagnostic granularity.
State of Charge (SoC) estimation is an indispensable feature in Battery Management Systems (BMS) and is an important function to be implemented for Electric Vehicles (EVs), Renewable Energy Storage (RES) and Portable electronics. For the nonlinearity of lithium-ion Batteries, the current model-based ones generally suffer from uncertain parameters and lose accuracy under various operating conditions. This paper discusses four traditional machine learning algorithms: Extreme Gradient Boosting (XGBoost), Random Forest (RF), Extra Trees (ET), Support Vector Regression (SVR) and explores them in the context of a common evaluation methodology with regard to SoC estimation. In order to make a fair comparison between the machine learning algorithms, a set of common methodology, which involves preprocessing, feature engineering, splitting into train and test set, and hyperparameters tuning are evaluated in all these machine learning techniques. Various performance parameters have been taken into account for assessing the performance of the actual prediction and to decide the extent to which the selected models are appropriate to be utilized in actual BMS applications.
Sudharshana, Subramanya Bhat, Bhushith M K· 2026 International Conferenc...· 0 citations
Accurate prediction of lithium-ion battery state of charge (SOC) is critical for ensuring safety, durability, and efficient energy management in electric vehicles (EVs). This study presents a systematic comparative analysis of supervised machine learning (ML) models for SOC prediction using the publicly available SiCWell battery dataset. Five regression-based algorithms—Generalized Linear Model (GLM), Gradient Boosting Machine (GBM), Extreme Gradient Boosting (XGBoost), Distributed Random Forest (DRF), and Extremely Randomized Trees (XRT)—are implemented within the H2O framework. The analysis considers six physically measurable battery features derived from voltage, resistance, charge throughput, and current direction. Model performance is evaluated using multiple regression metrics, including MAE, MSE, RMSE, RMSLE, residual deviance, and R², under a consistent train–test evaluation protocol. Results indicate that GLM achieves low prediction errors across both training and testing datasets, suggesting that the selected feature set exhibits a strong linear relationship with SOC within the evaluated operating range. However, ensemble models demonstrate higher variance and sensitivity to hyperparameter configurations, leading to reduced generalization performance. Rather than asserting a universally superior model, this study highlights how dataset characteristics, feature linearity, and model complexity influence SOC prediction accuracy. The findings provide insights into selecting appropriate ML models for battery management systems while emphasizing the importance of dataset transparency, interpretability, and generalization analysis.
M. S. Reddy, M. Monisha· Scientific Reports· 0 citations
An ensemble subspace k-nearest neighbour model for fault detection and classification of switch open-circuit and short-circuit faults and indicates that the proposed model provides an effective and computationally efficient solution for reliable fault diagnosis in electric vehicle motor drive systems.
Masadi Prashanth Kumar, Srikanth Velpula, Chidurala Saiprakash· Transactions of the Institut...· 0 citations
This study addresses the challenge of accurate and real-time battery health diagnostics in autonomous systems. Conventional voltage-based methods often fail to capture complex electrochemical dynamics, limiting their reliability in safety-critical applications. To overcome this limitation, this work proposes an integrated framework combining alternating current impedance spectroscopy with machine learning techniques for enhanced battery state estimation. Electrochemical impedance features are extracted under varying operating conditions, including state of charge, temperature, and cycling behavior, and are used as inputs to a data-driven diagnostic model. Experimental results demonstrate that the proposed method achieves a battery health prediction accuracy of 93.2%, reduces false alarm rates by approximately 40% compared to traditional methods, and maintains a response time within 500 ms. Additionally, a human–machine interface is developed to translate diagnostic results into intuitive visual outputs for decision support in autonomous cockpits. The proposed approach provides a reliable and efficient solution for advanced battery management systems in intelligent transportation environments.
Jiang Hong, Shi Meiqi, Gu Di et al.· Clean Energy· 0 citations
Renewable energy systems, including solar photovoltaic arrays and wind turbines, operate under highly variable environmental and operating conditions. Factors such as changing irradiance, temperature fluctuations, wind variability, and component aging make Fault Detection and Diagnosis (FDD) particularly challenging. Therefore, developing reliable, accurate, and interpretable diagnostic methods is essential to ensure system efficiency, safety, and long-term operation. Traditional model-based approaches, which rely on physical system models, offer clear interpretability and solid theoretical foundations. However, their effectiveness can be limited by modeling inaccuracies and difficulties in capturing complex nonlinear behaviors. On the other hand, data-driven and Artificial Intelligence (AI) techniques have demonstrated strong capabilities in pattern recognition and fault classification, but often face challenges related to data dependence, limited transparency, and reduced robustness under unseen conditions. This paper provides a comprehensive and structured review of FDD techniques for renewable energy systems, covering model-based, signal-based, data-driven, and hybrid approaches. A unified perspective is presented to clarify the strengths, limitations, and application domains of each category. Particular attention is given to recent advances in hybrid methods that combine physical modeling and AI, including feature fusion, ensemble learning, attention-based models, and transfer learning. Moreover, advanced signal processing techniques are discussed for their role in extracting meaningful features from noisy and non-stationary data. Rather than ranking methods by headline accuracy, which has become saturated and is only weakly comparable across heterogeneous datasets, the review adopts a critical, deployment-oriented perspective that emphasizes cross-condition robustness, standardized benchmarking, and the constraints of real-world deployment. The review also highlights the growing importance of digital twin technology as a promising framework for next-generation FDD systems, enabling real-time monitoring, adaptive learning, and predictive maintenance. Furthermore, Explainable AI is explored as a key direction for improving the transparency and trustworthiness of AI-based diagnostic models. Finally, the paper identifies major challenges and open research issues, such as data scarcity, generalization among different operating conditions, computational efficiency, and system reliability. Future research directions are outlined toward developing more robust, adaptive, and interpretable FDD solutions that can operate effectively in dynamic and uncertain environments.
Marouane Marzouk, Majdi Mansouri, Ahmed Anis Kahloul et al.· IEEE Access· 0 citations
Results show that using statistical vibration features with ensemble classifiers is a good way to diagnose multi-class bearing faults and establishes a comprehensive benchmark for ML- and DL-based rolling bearing FDD.
M. I. Quamar, Abdulrazaq Nafiu Abubakar, Ali Nasir· Journal of Vibration Enginee...· 0 citations