A machine learning modelling for the seismicity in the region of Greece from 2000 and thereafter
Abstract In the present article, a set of artificial neural network models, following the Feed Forward Neural Network (FNN) method, are formulated for the spectrums of acceleration, velocity and displacement of a single degree of freedom system. Earthquake accelerations from strong ground motions re...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2025-05-01
|
| Series: | Discover Applied Sciences |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s42452-025-07012-2 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850268944489775104 |
|---|---|
| author | Ambrosios-Antonios Savvides Leonidas Papadopoulos Dimitrios-Panagiotis Serris |
| author_facet | Ambrosios-Antonios Savvides Leonidas Papadopoulos Dimitrios-Panagiotis Serris |
| author_sort | Ambrosios-Antonios Savvides |
| collection | DOAJ |
| description | Abstract In the present article, a set of artificial neural network models, following the Feed Forward Neural Network (FNN) method, are formulated for the spectrums of acceleration, velocity and displacement of a single degree of freedom system. Earthquake accelerations from strong ground motions recorded in Greece from the year 2000 and hereinafter, are obtained and the equation of motion is solved, for various values of eigenperiod T and damping ratio $$\xi$$ ξ , in order to obtain the aforementioned spectrum. Subsequently, a data vector of about 10000 in size was obtained and the FNN models have been obtained with input values T and $$\xi$$ ξ , and output values to be the spectral acceleration ( $$S_a$$ S a ), spectral velocity ( $$S_v$$ S v ) and spectral displacement ( $$S_d$$ S d ). The aforementioned spectrums have been formulated for all 3 spatial components of the earthquake acceleration: direction East–West, direction North–South, vertical direction Z. It has been demonstrated that, the convergence of the iterative procedure of supervised learning is substantially fast, since it requires only a small amount of epochs in the order of magnitude of 30 and the corresponding relative round mean square error (RRMSE) is less than 0.01. In the East–West and North–South directions the acceleration spectrum is substantially approaching the regulatory Eurocode 8 design spectrum for the peak ground acceleration of 0.16 g and the maximum values of the spectrum are slightly less than the regulatory estimation. In the vertical direction Z, the peak ground acceleration is substantially lower than the 0.16 g however, the amplification in the largest values is very evident which leads the largest values to be again slightly smaller than the regulatory prediction. Subsequently, the proposed models are fast converging, easy to enrich with new recorded earthquakes, reliable in relation with the regulatory predictions and can be implemented to obtain the estimations that are important in earthquake engineering design by taking into account the whole seismicity over a broad area like the country of Greece. |
| format | Article |
| id | doaj-art-5203b996d3d9492bb43d2e575fe0f8ae |
| institution | OA Journals |
| issn | 3004-9261 |
| language | English |
| publishDate | 2025-05-01 |
| publisher | Springer |
| record_format | Article |
| series | Discover Applied Sciences |
| spelling | doaj-art-5203b996d3d9492bb43d2e575fe0f8ae2025-08-20T01:53:19ZengSpringerDiscover Applied Sciences3004-92612025-05-017614310.1007/s42452-025-07012-2A machine learning modelling for the seismicity in the region of Greece from 2000 and thereafterAmbrosios-Antonios Savvides0Leonidas Papadopoulos1Dimitrios-Panagiotis Serris2School of Civil Engineering, National Technical University of AthensSchool of Civil Engineering, National Technical University of AthensDivision of Aeronautical Engineering, Technical Mechanics, Construction Tests, Infrastructure Works, Hellenic Air Force AcademyAbstract In the present article, a set of artificial neural network models, following the Feed Forward Neural Network (FNN) method, are formulated for the spectrums of acceleration, velocity and displacement of a single degree of freedom system. Earthquake accelerations from strong ground motions recorded in Greece from the year 2000 and hereinafter, are obtained and the equation of motion is solved, for various values of eigenperiod T and damping ratio $$\xi$$ ξ , in order to obtain the aforementioned spectrum. Subsequently, a data vector of about 10000 in size was obtained and the FNN models have been obtained with input values T and $$\xi$$ ξ , and output values to be the spectral acceleration ( $$S_a$$ S a ), spectral velocity ( $$S_v$$ S v ) and spectral displacement ( $$S_d$$ S d ). The aforementioned spectrums have been formulated for all 3 spatial components of the earthquake acceleration: direction East–West, direction North–South, vertical direction Z. It has been demonstrated that, the convergence of the iterative procedure of supervised learning is substantially fast, since it requires only a small amount of epochs in the order of magnitude of 30 and the corresponding relative round mean square error (RRMSE) is less than 0.01. In the East–West and North–South directions the acceleration spectrum is substantially approaching the regulatory Eurocode 8 design spectrum for the peak ground acceleration of 0.16 g and the maximum values of the spectrum are slightly less than the regulatory estimation. In the vertical direction Z, the peak ground acceleration is substantially lower than the 0.16 g however, the amplification in the largest values is very evident which leads the largest values to be again slightly smaller than the regulatory prediction. Subsequently, the proposed models are fast converging, easy to enrich with new recorded earthquakes, reliable in relation with the regulatory predictions and can be implemented to obtain the estimations that are important in earthquake engineering design by taking into account the whole seismicity over a broad area like the country of Greece.https://doi.org/10.1007/s42452-025-07012-2Machine learningNeural networksAcceleration spectrumEarthquake engineering |
| spellingShingle | Ambrosios-Antonios Savvides Leonidas Papadopoulos Dimitrios-Panagiotis Serris A machine learning modelling for the seismicity in the region of Greece from 2000 and thereafter Discover Applied Sciences Machine learning Neural networks Acceleration spectrum Earthquake engineering |
| title | A machine learning modelling for the seismicity in the region of Greece from 2000 and thereafter |
| title_full | A machine learning modelling for the seismicity in the region of Greece from 2000 and thereafter |
| title_fullStr | A machine learning modelling for the seismicity in the region of Greece from 2000 and thereafter |
| title_full_unstemmed | A machine learning modelling for the seismicity in the region of Greece from 2000 and thereafter |
| title_short | A machine learning modelling for the seismicity in the region of Greece from 2000 and thereafter |
| title_sort | machine learning modelling for the seismicity in the region of greece from 2000 and thereafter |
| topic | Machine learning Neural networks Acceleration spectrum Earthquake engineering |
| url | https://doi.org/10.1007/s42452-025-07012-2 |
| work_keys_str_mv | AT ambrosiosantoniossavvides amachinelearningmodellingfortheseismicityintheregionofgreecefrom2000andthereafter AT leonidaspapadopoulos amachinelearningmodellingfortheseismicityintheregionofgreecefrom2000andthereafter AT dimitriospanagiotisserris amachinelearningmodellingfortheseismicityintheregionofgreecefrom2000andthereafter AT ambrosiosantoniossavvides machinelearningmodellingfortheseismicityintheregionofgreecefrom2000andthereafter AT leonidaspapadopoulos machinelearningmodellingfortheseismicityintheregionofgreecefrom2000andthereafter AT dimitriospanagiotisserris machinelearningmodellingfortheseismicityintheregionofgreecefrom2000andthereafter |