OPTIMIZED DEEP LEARNING WITH OPPOSITION-BASED ANT LION APPROACH FOR CRACK IDENTIFICATION OF THICK BEAMS

Abstract

Crack identification in thick beams has improved increasing considerations from the scientific and building areas since the unpredicted structural failure may cause disastrous, catastrophic and life trouble. The goal of the present examination is to predict the unknown crack location and its depth in thick beams from the information of frequency data obtained from experimental examination. The effectiveness of the proposed strategy is approved by numerical simulations in view of experimental data for a cantilever beam, free-free beam and simply supported beam. With the improvements in delicate figuring, optimization strategies are acknowledged to be an extremely proficient instrument to offer an answer for crack identification issue. In the simulation modeling, the parameters, for example, shift; modal assurance criterion (MAC) and stiffness, are predicted by utilizing optimized deep learning neural network (ODNN) approach in view of crack location and size. To improve the weight in DLNN, the opposition-based ant lion (OAL) is used by minimizing the mean square error (MSE) rate. The result shows that the proposed model achieves the optimal performance compared with existing techniques.