The　angle　steel　members　in　the　transmission　tower　are　mostly　connected　by　bolts.　The　bearing　capacity　of　angle　steel　bolted　connections　is　related　to　many　factors　(e.g.,　material　parameters,　geometry　of　angel　steel,　and　distribution　of　bolts)　and　is　difficult　to　fit　into　conventional　formulas.　This　paper　utilizes　machine　learning　techniques　to　predict　the　bearing　capacity　of　angle　steel　bolted　connections　with　different　design　parameters.　Experimental　studies　and　numerical　simulations　were　performed　to　investigate　the　failure　modes　and　bearing　capacities,　and　meanwhile　to　establish　a　database　for　machine　learning.　Three　different　machine　learning　models,　i.e.,　Back　Propagation　Neural　Network　(BPNN),　Random　Forest　(RF),　and　Gradient　Boosting　(GB),　were　trained　to　predict　the　bearing　capacity　of　steel　bolted　connections.　The　predicted　accuracy　of　machine　learning　models　was　compared　with　existing　industrial　standards.　The　results　show　that　the　failure　of　angle　steel　bolted　connections　mainly　happened　in　the　hole　wall.　The　bearing　capacity　of　angle　steel　bolted　connections　is　mainly　influenced　by　the　bolt　end　distance　and　bolt　edge　distance.　In　the　machine　learning　prediction　of　bearing　capacity,　the　R-2　correlation　coefficients　for　RF　and　GB　machine　learning　models,　compared　to　the　original　data,　are　all　above　0.98.　The　predicting　errors　of　machine　learning　models　are　within　10　%,　while　the　maximal　predicting　error　of　existing　industrial　standards　are　38.2　%similar　to　58.2　%.　This　demonstrates　the　feasibility　of　the　proposed　machine　learning　technology　in　accurately　predicting　the　bearing　capacity　of　angle　steel　bolted　connections.