Multi-objective optimisation of assembly fixturing layout for large composite fuselage panel reinforced by frames and stringers

To improve the rigidity of large composite fuselage panels and reduce the deformation caused by factors such as gravity during the assembly locating process, this paper presents a multi-objective optimisation method of the discrete fixturing layout based on the integration of a back-propagation neural network (BPNN) with the elitist nondominated sorting genetic algorithm (NSGA-II), which considers the characteristics of the composite fuselage panel, such as its complex structure and high tendency to undergo damage when subjected to force, in addition to the control the fixturing system cost. This method can simultaneously determine the optimal number and positions of tooling fixturing points. First, a binary integer variable is proposed to represent the fixturing layout, to realise the coupling representation of the number and positions of the fixturing points. To improve the optimisation efficiency, after comprehensively comparing the prediction accuracies of several commonly used surrogate models, a BPNN prediction model is constructed to describe the nonlinear mapping relationship between the fixturing layout and the maximum deformation (maximum Mises stress) of the composite fuselage panel based on the limited samples obtained by finite element analysis (FEA) and minimax Latin hypercube sampling (MLHS). Thereafter, the allowable deformation of the composite fuselage panel is considered the constraint; the number of fixturing points and the stress of the panel are used as the optimisation objectives. The Pareto solution set of the fixturing layout is determined using NSGA-II. Finally, a composite front fuselage panel case study of a wide-body airliner is considered to demonstrate the effectiveness of the proposed method.