In　order　to　get　a　more　accurate　finite　element　analysis　model　of　bolt-ball　shell　for　health　monitoring,　an　element　with　adjustable　stiffness　model　was　used　to　reveal　the　semi-rigid　characters　of　node.　Then,　the　neural　network　technology　was　introduced,　and　a　network　input　parameter　CPFM　utilizing　limited　measuring　points　information　was　constructed.　A　new　method　for　the　recognition　of　rigid　factor,　in　reasonable　consideration　of　the　semi-rigid　character　of　joint,　was　put　forward.　An　experimental　modal　of　single-layer　latticed　cylindrical　shell　with　157　nodes　and　414　elements　was　used　in　shaking　table　test.　Based　on　the　basic　model　and　measured　modal　data,　a　finite　element　model　updating　was　carried　out.　The　result　shows　that　the　method　is　effective　and　the　true　dynamic　characters　of　the　shell　structure　can　be　reflected　better,　at　the　same　time　the　neural　network　can　be　simplified　by　applying　step-by-step　correction　algorithm.