MACHINE LEARNING VA SEYSMIK MA'LUMOTLAR BAZASI ASOSIDA KO'P QAVATLI BINOLAR UCHUN O'RGANUVCHI KONSTRUKTIV LOYIHALASH TIZIMI
Keywords:
Kalit so'zlar: machine learning, neyron tarmoq, seysmik bashorat, o'rganuvchi tizim, ma'lumotlar bazasi, ANN, SVM, transfer learning.Abstract
Annotatsiya: Ushbu maqola ko'p qavatli temir-beton binolar uchun o'rganuvchi (self-learning) konstruktiv loyihalash tizimini taqdim etadi. Tizim har yangi loyihalash tsiklidan keyin ma'lumotlar bazasini boyitadi va keyingi loyihalar uchun dastlabki yechimlarni yanada aniqroq tavsiya eta boradi. Sun'iy neyron tarmoqlar (ANN) va support vector machine (SVM) algoritmlariga asoslangan prediksiya modeli seysmik javob ko'rsatkichlarini analitik hisobdan 40 baravar tezroq baholash imkonini beradi. O'zbekiston sharoitlari uchun kalibrlangan 3,200 ta bino loyihasidan tashkil topgan ma'lumotlar bazasi asosida tizimning bashorat aniqligi 91.3% ni tashkil etdi. Natijada muhandis shartlarni kiritgach, tizim optimal sxemani tavsiya etadi va bu tavsiyaning seysmik xavfsizlik darajasi xuddi to'liq analitik hisobdan o'tilgandek bo'ladi.
References
1. Zhang Y. et al. Machine learning-based seismic response prediction for buildings // Engineering Structures. 2020. Vol. 220.
2. Mangalathu S., Jeon J.S. Classification of failure mode and prediction of shear strength for reinforced concrete beam-column joints using machine learning // Engineering Structures. 2018.
3. Sun H. et al. A transfer learning approach for structural health monitoring and damage detection // Structural Control and Health Monitoring. 2021.
4. Lundberg S., Lee S.I. A Unified Approach to Interpreting Model Predictions (SHAP) // NIPS. 2017.
5. KMK 2.01.03-19. Qurilishda seysmiklikka hisoblash. Toshkent, 2019.
6. Chollet F. Deep Learning with Python. Manning Publications, 2021.
7. Pedregosa F. et al. Scikit-learn: Machine Learning in Python // JMLR. 2011. Vol. 12. P. 2825–2830.
