Data-Driven Fault Diagnosis Techniques: Non-Linear Directional Residual vs. Machine-Learning-Based Methods
Linear dependence of variables is a commonly used assumption in most diagnostic systems for which many robust methodologies have been developed over the years. In case the system nonlinearities are relevant, fault diagnosis methods, relying on the assumption of linearity, might potentially provide u...
Main Authors: | , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2022-03-01
|
Series: | Sensors |
Subjects: | |
Online Access: | https://www.mdpi.com/1424-8220/22/7/2635 |
_version_ | 1797437697317404672 |
---|---|
author | Nicholas Cartocci Marcello R. Napolitano Francesco Crocetti Gabriele Costante Paolo Valigi Mario L. Fravolini |
author_facet | Nicholas Cartocci Marcello R. Napolitano Francesco Crocetti Gabriele Costante Paolo Valigi Mario L. Fravolini |
author_sort | Nicholas Cartocci |
collection | DOAJ |
description | Linear dependence of variables is a commonly used assumption in most diagnostic systems for which many robust methodologies have been developed over the years. In case the system nonlinearities are relevant, fault diagnosis methods, relying on the assumption of linearity, might potentially provide unsatisfactory results in terms of false alarms and missed detections. In recent years, many authors have proposed machine learning (ML) techniques to improve fault diagnosis performance to mitigate this problem. Although very powerful, these techniques require faulty data samples that are representative of any fault scenario. Additionally, ML techniques suffer from issues related to overfitting and unpredictable performance in regions which are not fully explored in the training phase. This paper proposes a non-linear additive model to characterize the non-linear redundancy relationships among the system signals. Using the multivariate adaptive regression splines (MARS) algorithm, these relationships are identified directly from the data. Next, the non-linear redundancy relationships are linearized to derive a local time-dependent fault signature matrix. The faulty sensor can then be isolated by measuring the angular distance between the column vectors of the fault signature matrix and the primary residual vector. A quantitative analysis of fault isolation and fault estimation performance is performed by exploiting real data from multiple flights of a semi-autonomous aircraft, thus allowing a detailed quantitative comparison with state-of-the-art machine-learning-based fault diagnosis algorithms. |
first_indexed | 2024-03-09T11:26:21Z |
format | Article |
id | doaj.art-3142ffb9df9646d9b5ecaad4f4bc6299 |
institution | Directory Open Access Journal |
issn | 1424-8220 |
language | English |
last_indexed | 2024-03-09T11:26:21Z |
publishDate | 2022-03-01 |
publisher | MDPI AG |
record_format | Article |
series | Sensors |
spelling | doaj.art-3142ffb9df9646d9b5ecaad4f4bc62992023-12-01T00:02:29ZengMDPI AGSensors1424-82202022-03-01227263510.3390/s22072635Data-Driven Fault Diagnosis Techniques: Non-Linear Directional Residual vs. Machine-Learning-Based MethodsNicholas Cartocci0Marcello R. Napolitano1Francesco Crocetti2Gabriele Costante3Paolo Valigi4Mario L. Fravolini5Department of Engineering, University of Perugia, Via G. Duranti 67, 06125 Perugia, ItalyDepartment of Mechanical and Aerospace Engineering, West Virginia University, Morgantown, WV 26506, USADepartment of Engineering, University of Perugia, Via G. Duranti 67, 06125 Perugia, ItalyDepartment of Engineering, University of Perugia, Via G. Duranti 67, 06125 Perugia, ItalyDepartment of Engineering, University of Perugia, Via G. Duranti 67, 06125 Perugia, ItalyDepartment of Engineering, University of Perugia, Via G. Duranti 67, 06125 Perugia, ItalyLinear dependence of variables is a commonly used assumption in most diagnostic systems for which many robust methodologies have been developed over the years. In case the system nonlinearities are relevant, fault diagnosis methods, relying on the assumption of linearity, might potentially provide unsatisfactory results in terms of false alarms and missed detections. In recent years, many authors have proposed machine learning (ML) techniques to improve fault diagnosis performance to mitigate this problem. Although very powerful, these techniques require faulty data samples that are representative of any fault scenario. Additionally, ML techniques suffer from issues related to overfitting and unpredictable performance in regions which are not fully explored in the training phase. This paper proposes a non-linear additive model to characterize the non-linear redundancy relationships among the system signals. Using the multivariate adaptive regression splines (MARS) algorithm, these relationships are identified directly from the data. Next, the non-linear redundancy relationships are linearized to derive a local time-dependent fault signature matrix. The faulty sensor can then be isolated by measuring the angular distance between the column vectors of the fault signature matrix and the primary residual vector. A quantitative analysis of fault isolation and fault estimation performance is performed by exploiting real data from multiple flights of a semi-autonomous aircraft, thus allowing a detailed quantitative comparison with state-of-the-art machine-learning-based fault diagnosis algorithms.https://www.mdpi.com/1424-8220/22/7/2635additive modelanomaly detectionmultivariate adaptive regression splinestime-dependent directional residualsnon-linear residual-based techniquefault isolation |
spellingShingle | Nicholas Cartocci Marcello R. Napolitano Francesco Crocetti Gabriele Costante Paolo Valigi Mario L. Fravolini Data-Driven Fault Diagnosis Techniques: Non-Linear Directional Residual vs. Machine-Learning-Based Methods Sensors additive model anomaly detection multivariate adaptive regression splines time-dependent directional residuals non-linear residual-based technique fault isolation |
title | Data-Driven Fault Diagnosis Techniques: Non-Linear Directional Residual vs. Machine-Learning-Based Methods |
title_full | Data-Driven Fault Diagnosis Techniques: Non-Linear Directional Residual vs. Machine-Learning-Based Methods |
title_fullStr | Data-Driven Fault Diagnosis Techniques: Non-Linear Directional Residual vs. Machine-Learning-Based Methods |
title_full_unstemmed | Data-Driven Fault Diagnosis Techniques: Non-Linear Directional Residual vs. Machine-Learning-Based Methods |
title_short | Data-Driven Fault Diagnosis Techniques: Non-Linear Directional Residual vs. Machine-Learning-Based Methods |
title_sort | data driven fault diagnosis techniques non linear directional residual vs machine learning based methods |
topic | additive model anomaly detection multivariate adaptive regression splines time-dependent directional residuals non-linear residual-based technique fault isolation |
url | https://www.mdpi.com/1424-8220/22/7/2635 |
work_keys_str_mv | AT nicholascartocci datadrivenfaultdiagnosistechniquesnonlineardirectionalresidualvsmachinelearningbasedmethods AT marcellornapolitano datadrivenfaultdiagnosistechniquesnonlineardirectionalresidualvsmachinelearningbasedmethods AT francescocrocetti datadrivenfaultdiagnosistechniquesnonlineardirectionalresidualvsmachinelearningbasedmethods AT gabrielecostante datadrivenfaultdiagnosistechniquesnonlineardirectionalresidualvsmachinelearningbasedmethods AT paolovaligi datadrivenfaultdiagnosistechniquesnonlineardirectionalresidualvsmachinelearningbasedmethods AT mariolfravolini datadrivenfaultdiagnosistechniquesnonlineardirectionalresidualvsmachinelearningbasedmethods |