Comparative 1D–2D Convolutional Deep Learning for Intelligent Induction Motor Fault Diagnosis Based on Data Fusion

Reliable fault diagnosis is essential for maintaining the operational safety, efficiency, and availability of induction motors used in industrial electromechanical systems. This study presents a comparative convolutional deep-learning framework based on one-dimensional and two-dimensional convolutional neural networks for intelligent multiclass induction motor fault diagnosis through data fusion. Experimental measurements were collected under five operating conditions: Healthy operation, air-gap eccentricity, ball defect, inter-coil short circuit, and inter-turn short circuit. Time-domain vibration signals were segmented and transformed into statistical descriptors, including mean, standard deviation, variance, skewness, and kurtosis. These descriptors were fused with temperature, rotational speed, and three-phase current measurements to form the final diagnostic feature set. The resulting dataset was divided into 80% training and 20% testing subsets and evaluated using classification accuracy, precision, recall, F1-score, area under the receiver operating characteristic curve, Matthews correlation coefficient, specificity, and logarithmic loss. The 2D-CNN consistently outperformed the 1D-CNN during both training and independent testing. On the testing subset, the 2D-CNN achieved a classification accuracy of 92.4%, precision of 92.173%, recall of 92.526%, F1-score of 92.284%, AUC of 96.418%, MCC of 90.659%, and specificity of 98.134%, while attaining a low LogLoss of 0.241. In comparison, the 1D-CNN achieved 91.2% accuracy, 90.972% F1-score, and 95.732% AUC. The improved performance of the 2D architecture indicates that spatially organized feature representations can better capture interactions among vibration-derived descriptors and operating variables. The proposed framework provides an effective approach for data-driven induction motor condition monitoring and early fault identification.

Authors

Institutions

Publication Details

Journal
Engineering and Technology Journal
Published
2026-11-16
DOI
https://doi.org/10.30684/2412-0758.2430
Primary Topic
Machine Fault Diagnosis Techniques
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Comparative 1D–2D Convolutional Deep Learning for Intelligent Induction Motor Fault Diagnosis Based on Data Fusion

Israa Saad Ahmed, Ali Hussein Numan, Manal Hadi Jaber
Engineering and Technology Journal
Machine Fault Diagnosis Techniques
article

Comparative 1D–2D Convolutional Deep Learning for Intelligent Induction Motor Fault Diagnosis Based on Data Fusion

Israa Saad Ahmed, Ali Hussein Numan, Manal Hadi Jaber
article en

Abstract

Reliable fault diagnosis is essential for maintaining the operational safety, efficiency, and availability of induction motors used in industrial electromechanical systems. This study presents a comparative convolutional deep-learning framework based on one-dimensional and two-dimensional convolutional neural networks for intelligent multiclass induction motor fault diagnosis through data fusion. Experimental measurements were collected under five operating conditions: Healthy operation, air-gap eccentricity, ball defect, inter-coil short circuit, and inter-turn short circuit. Time-domain vibration signals were segmented and transformed into statistical descriptors, including mean, standard deviation, variance, skewness, and kurtosis. These descriptors were fused with temperature, rotational speed, and three-phase current measurements to form the final diagnostic feature set. The resulting dataset was divided into 80% training and 20% testing subsets and evaluated using classification accuracy, precision, recall, F1-score, area under the receiver operating characteristic curve, Matthews correlation coefficient, specificity, and logarithmic loss. The 2D-CNN consistently outperformed the 1D-CNN during both training and independent testing. On the testing subset, the 2D-CNN achieved a classification accuracy of 92.4%, precision of 92.173%, recall of 92.526%, F1-score of 92.284%, AUC of 96.418%, MCC of 90.659%, and specificity of 98.134%, while attaining a low LogLoss of 0.241. In comparison, the 1D-CNN achieved 91.2% accuracy, 90.972% F1-score, and 95.732% AUC. The improved performance of the 2D architecture indicates that spatially organized feature representations can better capture interactions among vibration-derived descriptors and operating variables. The proposed framework provides an effective approach for data-driven induction motor condition monitoring and early fault identification.

Engineering and Technology JournalVol. 45(1)
University of Technology - Iraq (IQ)
Machine Fault Diagnosis Techniques
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon