An advantage of using fibre-reinforced composites over conventional materials is that they can easily reduce the mass of structures by adapting the stiffness to the requirements of many practical applications. Thus, they are often used in the various fields related to aeronautics, space, or cars manufacturing. Depending on the size of the structure we want to manufacture, different techniques exist. Among those techniques, we have the resign transfer molding (RTM) process which is more suitable for low to medium volume [
]. In the RTM process some parameters have a significant impact on the production. In general, optimization of the RTM process, use an iterative procedure based on the RTM simulation by finite elements method and an optimization tool. This procedure converge when the error between the response of the simulated model and the experimental test is very low. This procedure is easy to implement but, it’s time-consuming and expensive even with a high computing machines. In this article, an hybrid strategy based on the genetic algorithms (GA) and an artificial neural networks (ANN) will be introduced to reduce the time-consuming in the RTM process. This approach starts by a mapped solutions (created by simulation results) using a small niching parameter of the GAPS program (genetic algorithm with parallel selection [
]), then an ANN is trained to create a RTM process model. Once this Meta-model is created (in general nonlinear) we run the GAPS with the objective function evaluated from the Metamodel response in order to locate the global solution that optimize the RTM process design. In some cases, design variables can be continuous or discrete variables. In such case binary and real coding are used simultaneously. The efficiency of the above strategy will be illustrated through a numerical examples.