Mahalanobis-Taguchi System Applications in Fault Diagnosis and Quality Improvement

The Mahalanobis-Taguchi System (MTS) is a hybrid multivariate analysis framework that combines Mahalanobis distance metrics with Taguchi’s robust design principles to detect anomalies, classify fault conditions and identify critical features for quality enhancement. Originally conceived for binary classification problems, MTS has been extended to handle multiclass scenarios through adaptive measurement scales and feature-selection strategies. In fault diagnosis applications, it is widely used to monitor vibration or sensor data from rotating machinery, power systems and logistics equipment, offering real-time detection of incipient failures even in highly imbalanced datasets. In quality improvement contexts, MTS assists in optimising manufacturing processes—such as electronics assembly or tablet production—by pinpointing influential process parameters and reducing inspection burdens while maintaining high predictive accuracy. Recent developments have focused on refining threshold selection for classification, integrating advanced signal-processing techniques and extending the method to symbolic or interval data. Together, these advances underscore the global relevance of MTS for predictive maintenance, process stability and cost-effective quality control across diverse industrial and environmental domains.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Mahalanobis-Taguchi System Applications in Fault Diagnosis and Quality Improvement publication trend

The graph below shows the total number of articles in mahalanobis-taguchi system applications in fault diagnosis and quality improvement across all publications each year (not limited to Nature Index journals).

Technical terms

Mahalanobis distance: A multivariate measure of the distance between a point and a distribution, used to detect deviations from normal operating conditions.

Taguchi method: A statistical approach for robust design that identifies key factors affecting process performance and minimises variation.

Threshold value: A cut-off on the Mahalanobis distance scale that separates normal from abnormal observations.

One-class classification: A modelling technique trained only on normal data to identify anomalies or faults without needing examples of every abnormal condition.

References

1. Modified Mahalanobis Taguchi System for Imbalance Data Classification. Computational Intelligence and Neuroscience (2017).
2. Adaptive Multiclass Mahalanobis Taguchi System for Bearing Fault Diagnosis under Variable Conditions. Sensors (2018).
3. Applying the Mahalanobis–Taguchi System to Improve Tablet PC Production Processes. Sustainability (2017).
4. Classification Performance of Thresholding Methods in the Mahalanobis–Taguchi System. Applied Sciences (2021).
5. Evaluation of One-Class Classifiers for Fault Detection: Mahalanobis Classifiers and the Mahalanobis–Taguchi System. Processes (2021).
6. On the Influence of Reference Mahalanobis Distance Space for Quality Classification of Complex Metal Parts Using Vibrations. Applied Sciences (2020).
7. Mahalanobis-Taguchi System for Symbolic Interval Data Based on Kernel Mahalanobis Distance. IEEE Access (2020).
8. Anomaly Detection in a Logistic Operating System Using the Mahalanobis–Taguchi Method. Applied Sciences (2020).