Optimization of impact performance of composites using artificial neural networks and evolutionary algorithms

Optimization is becoming common practice in all the engineering applications of late. While, there are a lot of literature and research available on the optimization of certain designs but at times there are problems in engineering where the exact nature of variables that affect the cost function are not known. In such cases, it is very difficult to get a good optimized solution since it depends upon the understanding of the researcher which parameters to base the cost function upon. A procedure for the design optimization of the composite laminated structures under the low velocity impact loads has been presented in this study. The optimization process is performed using Differential Evolution (DE) algorithm. The Artificial Neural Network (ANN) was used to train the model which forms the cost function. The optimization problem of impact loads on composite laminates depends upon a multitude of variables and to identify the most influential variables, a sensitivity analysis approach was performed prior to the optimization process. The product of the cost and the amount of energy absorbed was selected as the single cost function while the parameters used for optimization included the number of layers, thickness of layers and the stacking sequence as well as the type of fiber used. The approach adopted here has been proved to be very versatile and can be applied for a number of optimization problems where the unavailability of the cost function and its dependence is not known.