Frame optimization using neural network / Mohd Suhairil Meon, Muhammad Azhan Anuar and Muhammad Zulkifli Mohamad

This research addressed the performance of Neural-network (NN) to the structural optimization concept of frame structure. The development of NN technology stemmed from the desire to create an artificial system that could perform "intelligent" tasks similar to those performed by the human brain. The optimization part is performed using Finite Element software. Stress and displacement constrains has been chosen as the optimum criteria. The optimized data taken from FE analysis had been trained through Back Propagation Neural-network technique (BPNN) in order to identify the capability of this strategy to predict the exact data. Three case study were performed with different complexity of structural configuration. Result indicates the Neural-network were found capable of predicting the exact solution with proper training but this ability depended on the complexity of the frame structural optimization itself.