Structural parameter identification and inversion algorithm design of fair-faced concrete shell for construction precision control

Fair-faced concrete shell structure is widely used in large public buildings because of its aesthetic value and mechanical efficiency, but its construction accuracy control faces multiple challenges such as geometric nonlinearity, time-varying characteristics of materials and strict requirements on apparent quality. Traditional methods have some problems, such as lagging parameter identification, uncontrollable error transmission and relying on experience adjustment, which are difficult to meet the requirements of high-precision construction. In order to solve the above problems, this study constructs a multi-source data inversion framework integrating strain, displacement and surface topography, and puts forward a hybrid model of "orthogonal design-neural network-meso-finite element" to solve the coupling inversion problem of creep and elastic Poisson's ratio. Through multi-objective optimization modeling, the surface topography error is included in the objective function as a penalty term to realize the cooperative control of structural mechanical behavior and geometric accuracy. The staged decoupling strategy is adopted to identify the elastic parameters and creep parameters in stages to improve the inversion stability. The inversion algorithm based on improved particle swarm optimization (PSO) is developed, and the proxy model is introduced to replace the finite element calculation, and the efficiency is improved through dynamic inertia weight adjustment, boundary constraint and early stop mechanism. Case study shows that the relative errors of the algorithm in identifying key parameters such as elastic modulus, Poisson's ratio and creep coefficient are all below 6%, the root mean square error of displacement prediction is only 0.38mm, the improvement rate of average surface roughness is 35.7%, the single inversion time is shortened from 98 minutes to 12 minutes, the qualification rate of key indicators in the whole project is significantly improved, and the rework cost is reduced by 75%. The research results promote the transformation of construction control from experience-driven to data-driven, and provide effective technical support for the high-precision construction of fair-faced concrete shell structure.