Development of user-friendly kernel-based Gaussian process regression model for prediction of load-bearing capacity of square concrete-filled steel tubular members

A Machine Learning (ML) model based on Gaussian regression, using different kernel functions, is introduced in this paper to assess the load-carrying capacity of square concrete-filled steel tubular (CFST) columns. The input data used to develop the prediction model, which consists of 314 datasets i...

Mô tả chi tiết

Lưu vào:
Hiển thị chi tiết
Tác giả chính: Tien-Thinh, Le, Minh Vuong, Le
Định dạng: Bài Báo
Ngôn ngữ:English
Nhà xuất bản: Materials and Structures 54(2), 59 2021
Chủ đề:
Truy cập trực tuyến:https://link.springer.com/article/10.1617%2Fs11527-021-01646-5
https://dlib.phenikaa-uni.edu.vn/handle/PNK/1429
Từ khóa: Thêm từ khóa
Không có từ khóa, Hãy là người đầu tiên đánh dấu biểu ghi này!
Mô tả
Tóm tắt:A Machine Learning (ML) model based on Gaussian regression, using different kernel functions, is introduced in this paper to assess the load-carrying capacity of square concrete-filled steel tubular (CFST) columns. The input data used to develop the prediction model, which consists of 314 datasets including the structural geometrical parameters and the mechanical properties of the materials, was collected from available resources in the literature. The performance of the prediction model has also been validated by comparing with: (i) other ML models such as Artificial neural network, Support vector machine, etc.; and (ii) existing formulations in the literature for predicting load-carrying capacity of square CFST columns (including several codes such as EC4, AISC and ACI). The obtained results showed that the proposed model has outperformed them. The drawbacks of the model have been investigated by studying the influence of the input variables, together with uncertainty analysis providing 68, 95, and 99% prediction confidence intervals. Finally, a user-friendly interface has been developed to facilitate the application of the proposed model, providing the prediction value as well as confidence levels.