The　crystal　plasticity　constitutive　model　involves　numerous　parameters　with　strong　coupling　effects,　and　how　to　accurately　and　efficiently　model　is　a　serious　challenge.　To　address　this　challenge,　this　study　proposes　an　advanced　parameter　identification　method　for　the　crystal　plasticity　constitutive　model　based　on　a　hybrid　surrogate　model.　The　numerical　integration　method　of　crystal　plastic　constitutive　is　optimized　to　improve　the　efficiency　of　parameter　identification.　The　Backtracking　Line　Search　(BLS)　algorithm　with　the　idea　of　Dogleg　method　are　introduced　to　optimize　the　Newton-Raphson　(N-R)　iterative　method,　reducing　iteration　counts.　A　novel　method　to　identify　the　parameters　of　the　constitutive　model　is　proposed.　A　CPFEM　(Crystal　Plastic　Finite　Element　Method)　model　is　established　to　simulate　the　macroscopic　mechanical　response　of　materials.　The　efficient　analysis　of　Sobol　＇s　global　sensitivity　is　achieved　by　calculating　the　coefficients　of　the　SPCE　(Sparse　Polynomial　Chaos　Expansion)　model,　obtaining　the　key　parameters　as　the　parameter　identification　target.　The　constitutive　model　parameters　are　accurately　and　efficiently　identified.　The　SPCE　model　further　serves　as　the　regression　part　of　the　Kriging　model　to　construct　a　more　accurate　SPCE-Kriging　hybrid　surrogate　model,　which　replaces　the　finite　element　analysis　to　save　the　calculation　cost　and　enhance　the　calculation　efficiency.　The　research　results　show　that　the　proposed　method　can　effectively　improve　the　identification　efficiency　of　the　constitutive　model　parameter　under　the　premise　of　ensuring　the　calculation　accuracy,　and　can　provide　technical　ideas　for　the　mesoscopic　simulation　of　material　plastic　deformation.