Performance and vulnerability based seismic design using artificial intelligence : case of framed structures

Simple item page
dc.contributor.advisorChikh, Benazouz
dc.contributor.advisorMebarki, Ahmed
dc.contributor.authorBenbokhari, Abdellatif
dc.date.accessioned2026-10-29T14:28:40Z
dc.date.available2026-10-29T14:28:40Z
dc.date.issued2025-03-23
dc.description181 p. ; ill ; 30 cm.+ (1cd-rom). Bibliogr. p.171-181. + Annexes.
dc.description.abstractSeismic design is a crucial step in ensuring structural stability and the capacity to withstand seismic forces. For critical and strategic buildings, such as hospitals, performance during an earthquake is vital. This is where performance-based seismic design (PBSD) comes into play. PBSD is known for its complexity and often relies on Nonlinear Time History Analysis (NLTHA), which can be time-consuming. The process of finding optimal seismic designs can take hours or even days. In this thesis, a new approach is proposed that replaces the use of structural software and numerical methods like NLTHA for determining structural performance during an earthquake. The proposed AI-based approach is divided into two main steps. The first step focuses on identifying the best Machine Learning (ML) model capable of predicting seismic responses, including Maximum Inter-story Drift Ratio (MIDR), Maximum Base Shear (MBS), and Maximum Roof Drift Ratio (MRDR). Dimensionality reduction algorithms are employed to enhance ML predictions by reducing input features and reorienting the original axes to new ones. The second step involves using Multi-Objective Optimization (MOO) to find the best design that satisfies a dual performance check under two seismic intensity levels, taking into account damage grade and initial construction cost. To validate this approach, three special moment frames (SMFs) were selected. Both Force-Based Design (FBD) and the AI-based approach were used to design these SMFs. A comparative analysis of the design behavior was performed to evaluate the benefits of the proposed approach relative to FBD. The results indicated that the proposed approach effectively reduced the initial cost, improved the damage grade ratio, and lowered the probability of exceeding damage levels. Additionally, the proposed method ensured that the inter-story drift (ISD) did not exceed the fixed performance threshold for extensive earthquake levels, unlike the FBD design. Overall, the proposed approach successfully balanced cost and performance in structural design while reducing processing time. This fast and accurate method allows for quicker and more efficient designs without the need for extensive resources.
dc.identifier.citationBenbokhari, Abdellatif. Performance and vulnerability based seismic design using artificial intelligence : case of framed structures. Th. Dct., Alger, ENSTP-FJ, 2025.
dc.identifier.urihttp://dspace.enstp.edu.dz/handle/123456789/1312
dc.language.isoen
dc.publisherENSTP-FJ
dc.subjectPerformance-based seismic design
dc.subjectArtificial Intelligence
dc.subjectMulti-objective optimization
dc.subjectSeismic response prediction
dc.subjectNonlinear time history analysis
dc.subjectPushover analysis
dc.titlePerformance and vulnerability based seismic design using artificial intelligence : case of framed structures
dc.typeThèse de Doctorat
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
W3147-25_Dct 17.pdf
Size:
56.07 MB
Format:
Adobe Portable Document Format
Description:
Download
License bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
license.txt
Size:
1.71 KB
Format:
Item-specific license agreed to upon submission
Description:
4
item.page.filesection.subject
Download
Collections
Thèses de Doctorat