Enhancing Reliability in Redundant Homogeneous Sensor Arrays with Self-X and Multidimensional Mapping
Mechanical defects and sensor failures can substantially undermine the reliability of low-cost sensors, especially in applications where measurement inaccuracies or malfunctions may lead to critical outcomes, including system control disruptions, emergency scenarios, or safety hazards. To overcome t...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-06-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/25/13/3841 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Mechanical defects and sensor failures can substantially undermine the reliability of low-cost sensors, especially in applications where measurement inaccuracies or malfunctions may lead to critical outcomes, including system control disruptions, emergency scenarios, or safety hazards. To overcome these challenges, this paper presents a novel Self-X architecture with sensor redundancy, which incorporates dynamic calibration based on multidimensional mapping. By extracting reliable sensor readings from imperfect or defective sensors, the system utilizes Self-X principles to dynamically adapt and optimize performance. The approach is initially validated on synthetic data from tunnel magnetoresistance (TMR) sensors to facilitate method analysis and comparison. Additionally, a physical measurement setup capable of controlled fault injection is described, highlighting practical validation scenarios and ensuring the realism of synthesized fault conditions. The study highlights a wide range of potential TMR sensor failures that compromise long-term system reliability and demonstrates how multidimensional mapping effectively mitigates both static and dynamic errors, including offset, amplitude imbalance, phase shift, mechanical misalignments, and other issues. Initially, four individual TMR sensors exhibited mean absolute error (MAE) of 4.709°, 5.632°, 2.956°, and 1.749°, respectively. To rigorously evaluate various dimensionality reduction (DR) methods, benchmark criteria were introduced, offering insights into the relative improvements in sensor array accuracy. On average, MAE was reduced by more than 80% across sensor combinations. A clear quantitative trend was observed: for instance, the MAE decreases from 4.7°–5.6° for single sensors to 0.111° when the factor analysis method was applied to four sensors. This demonstrates the concrete benefit of sensor redundancy and DR algorithms for creating robust, fault-tolerant measurement systems. |
|---|---|
| ISSN: | 1424-8220 |