Soil Bearing Capacity Prediction Using MATLAB ML and PLAXIS

A student-laptop system that predicts the ultimate and allowable bearing pressure of shallow footings with machine learning trained on finite-element analyses of the PLAXIS 2D type. 800 Latin-hypercube strip-footing cases were analysed with the project's own plane-strain FE code (6-node triangles, Mohr-Coulomb, K0 procedure, rough rigid footing): the collapse load by FE limit analysis (second-order cone programme) with a two-mesh Richardson correction, and the load-settlement curve by incremental elasto-plastic analysis. Seven algorithms were compared, each directly and as a physics-informed hybrid; the hybrid Gaussian process predicts q_u with 0.38 % mean error (90 % band x/÷1.010, 90.0 % coverage) and the 25 mm pressure with 1.28 %, against 2.5 % for Vesic / IS 6403 and +9.0 % bias for Terzaghi. The web application applies IS 6403 (FS = 3) and IS 1904 (25 mm), sizes footings, runs Monte Carlo reliability and a live FE check. Honest scope: PLAXIS 2D and MATLAB are NOT installed on the build PC - their files were written but never run; the training data are synthetic numerical analyses, not field load tests.

Category: Civil Engineering / Geotechnical Engineering / Foundation Engineering / Finite-Element Analysis / Machine Learning / Reliability-Based Design / MATLAB and PLAXIS (scripts) with Python

Civil Engineering Project C09 from ProTechSC Academy, with graphical abstract, video demonstration and project guidance.