A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Improving damage detection by combining multiple classifiers in different feature spaces
https://orcid.org/0000-0001-8377-8149
https://orcid.org/0000-0003-2233-2383
Highlights Algorithm parameter setting improves the damage detection performance. Artificially generated damaged samples allow algorithm setting and validation. Classifiers in different feature spaces show distinct sensitivity to damage. Fusing more classifiers from distinct feature spaces enhances damage detection. Thresholds on consecutive outputs ensure a prompt detection reducing false alarms.
Abstract Vibration-based damage identification methods have been explored in several engineering fields. However, the effective damage-sensitivity of vibration-based features and the balance between false alarms and readiness of the anomaly detection are still open issues. Here, a cost-effective damage detection methodology based on a Deterministically Generated Negative Selection Algorithm is implemented and tested over a wide dataset generated through the numerical simulation of a three-storey concrete building in different damage scenarios. The approach allows training the classifiers only based on undamaged samples and ensures a prompt response to damage outbreak by fusing the classification conducted in parallel on different feature spaces, exploiting their distinct sensitivity to damage. A strategy to improve the reliability of the alarm is implemented by introducing counters and thresholds on consecutive classification outputs. A satisfactory trade-off between false and true alarms is achieved by comparing their probabilities, obtained through Markov chains, for different values of such thresholds.
Improving damage detection by combining multiple classifiers in different feature spaces
https://orcid.org/0000-0001-8377-8149
https://orcid.org/0000-0003-2233-2383
Highlights Algorithm parameter setting improves the damage detection performance. Artificially generated damaged samples allow algorithm setting and validation. Classifiers in different feature spaces show distinct sensitivity to damage. Fusing more classifiers from distinct feature spaces enhances damage detection. Thresholds on consecutive outputs ensure a prompt detection reducing false alarms.
Abstract Vibration-based damage identification methods have been explored in several engineering fields. However, the effective damage-sensitivity of vibration-based features and the balance between false alarms and readiness of the anomaly detection are still open issues. Here, a cost-effective damage detection methodology based on a Deterministically Generated Negative Selection Algorithm is implemented and tested over a wide dataset generated through the numerical simulation of a three-storey concrete building in different damage scenarios. The approach allows training the classifiers only based on undamaged samples and ensures a prompt response to damage outbreak by fusing the classification conducted in parallel on different feature spaces, exploiting their distinct sensitivity to damage. A strategy to improve the reliability of the alarm is implemented by introducing counters and thresholds on consecutive classification outputs. A satisfactory trade-off between false and true alarms is achieved by comparing their probabilities, obtained through Markov chains, for different values of such thresholds.
Improving damage detection by combining multiple classifiers in different feature spaces
https://orcid.org/0000-0001-8377-8149
https://orcid.org/0000-0003-2233-2383
Highlights Algorithm parameter setting improves the damage detection performance. Artificially generated damaged samples allow algorithm setting and validation. Classifiers in different feature spaces show distinct sensitivity to damage. Fusing more classifiers from distinct feature spaces enhances damage detection. Thresholds on consecutive outputs ensure a prompt detection reducing false alarms.
Abstract Vibration-based damage identification methods have been explored in several engineering fields. However, the effective damage-sensitivity of vibration-based features and the balance between false alarms and readiness of the anomaly detection are still open issues. Here, a cost-effective damage detection methodology based on a Deterministically Generated Negative Selection Algorithm is implemented and tested over a wide dataset generated through the numerical simulation of a three-storey concrete building in different damage scenarios. The approach allows training the classifiers only based on undamaged samples and ensures a prompt response to damage outbreak by fusing the classification conducted in parallel on different feature spaces, exploiting their distinct sensitivity to damage. A strategy to improve the reliability of the alarm is implemented by introducing counters and thresholds on consecutive classification outputs. A satisfactory trade-off between false and true alarms is achieved by comparing their probabilities, obtained through Markov chains, for different values of such thresholds.
Improving damage detection by combining multiple classifiers in different feature spaces
Barontini, Alberto (author) / Masciotta, Maria Giovanna (author) / Amado-Mendes, Paulo (author) / Ramos, Luis F. (author) / Lourenço, Paulo B. (author)
Engineering Structures ; 299
2023-10-17
Article (Journal)
Electronic Resource
English
Shape Classification Using Multiple Classifiers with Different Feature Sets
| British Library Conference Proceedings | 2012
| British Library Online Contents | 2017
Feature Selection for Modular Neural Network Classifiers
| British Library Online Contents | 2002