To achieve the online monitoring of the buckling deformation of the beam structure, a shape sensing method for monitoring the buckling deformation of FRP composite beam structure is proposed. Firstly, according to the higher order shear deformation theory, a method for describing the deformation field of laminate composite beam structure is proposed. Based on the von Karman strain gradient theory, the section strain expressions are derived. Then, the least square function is employed to formulate the displacement reconstruction model. In this model, the displacement interpolation functions are constructed by using the quartic non-uniform rational Bsplines (NURBS) basis function as the shape function, and the theoretical section strain field formulation is derived. Based on a small number of discrete point strain measurement values, a nonlinear strain decoupling method is proposed, and the transformation relationship between surface strains and measured section strains is established. Finally, to evaluate the accuracy and effectiveness of the proposed method, a simply supported beam test platform is established with 25 layer carbon fiber composite beams as samples, and its numerical calculation and experimental demonstration are carried out. Results show that the proposed model and the reconstruction displacement error is less than 8% under different axial loads.