Structural dynamic characteristics of the aerospace structures are essential to obtain the structural responses due to dynamic loads during its mission. The structural dynamic parameters of the aerospace structures are frequencies, associated mode shape and damping. Usually finite element (FE) model of the aerospace structures are generated to estimate the frequencies and the associated mode shape. These FE models are validated by modal survey/ground resonance tests (GRT) to ensure its completeness and correctness. The modelling deficiencies, if any, in these FE models have to be corrected. This paper describes the method to locate and correct/update the FE mode: using the difference in the curvature of the mode shape between the FE mode shape and the test mode shape based on artificial neural network (ANN). The Latin hypercube sampling (LHS) technique is used for generating the training sets for ANN. Latin hypercube sampling is to generate a distribution of plausible collections of parameter values from a multi-dimensional distribution.