To　address　the　issue　of　insufficient　redundancy　during　structural　collapse,　this　paper　proposes　a　beam-column　joint　reinforced　with　corrugated　steel　plates.　Through　experimental　research,　the　failure　modes　and　load-displacement　relationships　of　this　type　of　joint　were　clarified.　Compared　to　traditional　joints,　this　innovative　design　effectively　enhances　both　the　peak　load　and　ultimate　displacement　before　and　after　fracture.　A　corresponding　finite　element　model　was　established,　and　the　simulation　results　aligned　with　the　experimental　findings,　followed　by　a　parametric　analysis.　According　to　the　analysis　results,　variations　in　the　dimensions　of　the　corrugated　steel　plates　primarily　affect　the　structural　stiffness　once　the　plates　become　effective,　as　well　as　the　ultimate　displacement　of　the　structure;　however,　their　impact　on　the　ultimate　bearing　capacity　can　be　considered　negligible.　Based　on　the　finite　element　simulation　results,　predictions　and　cross-validation　were　conducted　using　six　machine　learning　methods,　with　results　indicating　that　the　PSO-SVM　model　demonstrates　good　predictive　accuracy　and　robustness.　Additionally,　a　comprehensive　evaluation　index　for　joint　performance　was　proposed　to　assess　the　prediction　results　thoroughly,　thereby　optimizing　the　beam-column　substructure　and　determining　the　characteristic　parameters　of　the　joint.　©　2025