Aug 2026· IEEE Sensors Journal· Vol 26, pp. 24002-24013· 0 citations· 46 references
Abstract
This article proposes a cable overheating state recognition method based on a metal–oxide–semiconductor (MOS) gas sensor array. Given the limited number of original experiments and the temporal misalignment caused by heating delay, gas diffusion, and sensor–response hysteresis, this work designs a response-aligned sliding-window strategy. Specifically, a reference sensor channel is used to locate the response onset, and an offset window is then constructed to capture the rising response stage with stronger state discrimination. Multidimensional features were extracted, and both traditional machine learning models and time-series classification models were systematically evaluated. MiniRocket achieved the best offline accuracy and F1-score, while compact traditional models such as K-nearest neighbors (KNNs) and support vector machines (SVMs) also achieved accuracies above 98%. Considering offline recognition performance, online stability, and STM32 resource constraints, selected models were deployed on an STM32 platform for online recognition every 15 s. Pressure experiments further showed that the deployed model can correctly identify heating states within the evaluated pressure range of 91.2–101.3 kPa. The proposed method provides an early warning approach for cable overheating and shows potential for embedded engineering applications.
Unplanned failures of induction motors impose serious operational and financial penalties on industrial facilities, yet the fault signatures that precede such failures are detectable well in advance through careful sensor instrumentation and data-driven analysis. This paper presents an end-to-end Internet-of-Things (IoT) predictive maintenance scheme based on two off-the-shelf sensors: a DS18B20 one-wire digital thermometer and 2 piezo vibration sensor modules, with an ESP32 edge device for running the full machine learning pipeline offline, independent from any cloud services. Four operating scenarios are considered: healthy condition, BPFO (bearing outer race fault), misaligned shaft, and rotor imbalance. Based on 1-second sampling intervals, 17 descriptors are derived, including statistics in the time domain, Fourier harmonic peaks, energy ratios between different frequency bands, and temperature gradients measured across all sensors. A dual-stage feature selection method using mutual information (MI) score and Random Forest mean decrease impurity (MDI) ranking reduces the number of features to the 10 most relevant descriptors, reducing the computational complexity by 41% at the expense of 5.6% F1-macro. On a balanced 600-sample synthetic dataset, the resulting Random Forest classifier attains 87.3% hold-out accuracy, 91.0±1.9% five-fold cross-validation accuracy, and a macro area-under-the-ROC-curve of 0.980. End-to-end inference takes just 39 ms on the ESP32, easily meeting the 200 ms requirement for real-time alerting.
Akash Mastud, Dhiraj Vaidya, Azaroddin Sayyed et al.· International Conference on...· 0 citations
Internal short-circuit (ISC) faults in lithium-ion batteries shorten service life and may cause severe safety issues such as thermal runaway. Therefore, this study proposes a purely data-driven method based on terminal voltage during charging. The analysis focuses on the stable mid-to-late stage of low-rate constant-current charging. Multiple discrete time instants within 4200 s–6000 s are selected, and the voltage difference between them is used as the diagnostic feature. A random forest (RF) model is developed, trained, and validated. To improve performance, Bayesian optimization tunes RF hyperparameters, and particle swarm optimization is applied for feature selection. Diagnostic thresholds are established based on model outputs to enable ISC fault identification and prediction. Support vector regression and Gaussian process regression are used for comparison. Results show that the optimized RF model achieves a mean relative error of 4.254%, a root mean square error of 32.688 Ω, and a coefficient of determination (R2) of 98.42%, significantly outperforming the comparison methods.
S. Duan, Yizhen Qu, Ye Liu et al.· Engineering Research Express· 0 citations
Main steam isolation valves face severe challenges during monitoring under high-temperature, high-pressure, and noisy environments. These challenges include weak fault features, data blocking in multi-channel acquisition, and high false alarm rates. To address these issues, this paper proposes a dedicated intelligent fault diagnosis software based on Model-Based Design (MBD) and the PyQt5 framework. The software innovatively employs a QThread pool architecture to achieve non-blocking synchronous acquisition across 26 channels, effectively resolving resource contention under high-throughput data transmission. For diagnosis, the system integrates dual lightweight algorithms (SVM and LSTM) to balance computing constraints with precise status determination and trend prediction, while introducing a spatial weighted algorithm based on time-domain deviation for accurate fault localization. Crucially, a linear regression dynamic threshold model corrected by temperature and pressure is established to mitigate false alarms under non-stationary operating conditions. Validation using full-power operation data from a nuclear power plant demonstrates that the software limits synchronous acquisition delay to within 50ms and achieves a fault determination accuracy of 94%. Furthermore, the conditioncorrected mechanism reduces the false alarm rate by over 98% (to 0.07%) and shortens the average fault investigation cycle by 90%. This system effectively overcomes the lag in traditional monitoring, marking a critical shift from “passive repair” to “proactive prediction” in steam valve maintenance.
Yanjie Zhong, Renpu Fu, Haiyang Wang et al.· 2026 6th International Confe...· 0 citations
In power systems, failures of medium-voltage switchgear contacts caused by high temperatures occur from time to time. Traditional temperature measurement methods are applied in enclosed and strong electromagnetic interference environments, which have certain limitations. This paper designs an intelligent monitoring system for the contact temperature of medium-voltage switch cabinets based on ZigBee technology and AI anomaly prediction. The system uses the CC2530 as its main control chip, adopts the DS18B20 digital temperature sensor to collect contact temperatures, and solves the problem of difficult power supply at the high-voltage end through a CT self-powered and battery switching solution. Based on the Z-Stack protocol stack, it realizes node networking and data transmission. Finally, LabVIEW is used to develop a host computer monitoring platform, in which a lightweight AI temperature anomaly prediction model is embedded. This model uses historical temperature data of switchgear contacts to train the Gradient Boosting Decision Tree (GBDT) algorithm, and can real-time predict potential sudden temperature rises within 5-10 minutes.
Pengfei Kang, Xuewen Zong, Xianmei Zhang et al.· International Conference on...· 0 citations
AC series arc faults in household appliances pose significant safety risks but remain challenging to detect due to load-dependent signatures and complex interactions in mixed-branch operations. This paper presents a practical, real-time arc fault detection system that integrates a comprehensive two-branch household dataset, lightweight handcrafted features, and classical machine learning for deployment on an STM32F407 microcontroller (MCU). Current signals were sampled at 10 kHz across 32 scenarios, encompassing 13 singleload and 19 two-branch mixed-load cases. Each signal window is represented by a 133-dimensional feature vector, comprising five statistical descriptors and 128 FFT magnitude coefficients, optimized for resource-constrained environments. Among various evaluated classifiers, Random Forest was selected for its superior trade-off between recognition performance and deployment efficiency. The proposed model achieved 89.18% accuracy, 91.34% recall, and 86.96% F1-score. Furthermore, a leave-one-scenario-out (LOSO) evaluation was employed to systematically analyze the masking effect in multi-branch environments, identifying challenges in which high-power loads obscure weak arc signatures. The complete firmware requires 896 KB of Flash and 102 KB of RAM, with a total processing time of only 7.02 ms per cycle, significantly within the update interval and confirming its feasibility for low-cost, real-time protection devices.
Tan Tong Sy, Toan Le Duc, Dao Le Xuan et al.· E3S Web of Conferences· 0 citations
Metal oxide surge arresters (MOAs) are core overvoltage protection devices in power systems. However, conventional single-parameter MOA fault diagnostic methods exhibit low accuracy and limited anti-interference capability in complex operating conditions. In this article, a distributed in situ measurement technique and a lightweight analytical criterion for MOA fault diagnosis are proposed. First, a measurement system is designed based on a distributed parameter equivalent circuit model, which deploys eight embedded sensors within the valve column along the axial direction. This architecture enables synchronous acquisition of the longitudinal leakage current vector and temperature profile. Second, experiments are performed under three typical conditions of normal operation, surface fouling, and valve disk failure to reveal the distribution characteristics of temperature and leakage current. Third, an analytical criterion is proposed, which integrates multiple feature parameters, including leakage current inhomogeneity coefficient, full current, and maximum temperature rise. Last, the K-means clustering method is adopted to determine thresholds for three typical operating conditions. The experimental results show that the proposed criterion achieves an accuracy of 96.67 %, outperforming single-feature and traditional machine learning methods. The research provides a new approach for MOA fault diagnosis.
Sisheng Wu, Shengwen Shu, Zheping Lin et al.· IEEE Transactions on Instrum...· 0 citations