Reliability-based multi-objective optimization design of composite patch repair structure using artificial neural networks

Due to the parameter differences and the existence of random factors, the performance of composite repair structure shows a large dispersion. In order to obtain stronger, lighter and more reliable repair structure, multiobjective optimization design of carbon fiber reinforced polymers (CFRP) laminate repair structure is investigated by combining the reliability theory, artificial neural networks and the genetic algorithm. First, a 3D simulation model of the composite laminate patch repair structures is established using the 3D Hashin criterion and the cohesive region model. Then, the Latin Hypercube sampling (LHS) method is used to realize the random sampling, and the strength proxy model of the repair structure is established by using Back-propagation artificial neural network, further a multi-objective optimization model with tensile strength, reliability and weight as objective functions is built considering the design parameters and the random parameters. Finally, NSGAIIalgorithm is used to solve the multi-objective optimization problem, and a set of solutions on the Pareto front surface are obtained, also the optimal design parameters of the composite repair structure meets the requirements is obtained.