In order to study the deformation law of the form and position errors between the constrained surface and related surface of the thinwalled deformed engine casing in the assembled state, a restraint measure method was proposed. The finite element method was used to simulate the actual assembly of the casing, and the influence of different clamping parameters on the flatness of the constraint surface was analyzed. Through the intuitive analysis,it was determined the main influencing factor was the platen position. By using the established model, the variation of the flatness and parallelism of the relevant surfaces under the restraint state were analyzed. The prediction models of the platen position and the constraint surface flatness were established by using the neural network, and the global minimum value of the constraint surface flatness and the corresponding platen position were solved by using the genetic algorithm, so that the deformation of the whole casing could be effectively controlled.