Fatigue behaviors prediction method of welded joints based on soft computing methods

Abstract

In allusion to the highly nonlinear, multivariable, strong coupling and interaction of various factors of the welded process, it is difficult to predict the fatigue behaviors of welded joints. We make use of the capabilities and advantages of rough set theory, ant colony algorithm and BP neural network, a novel fatigue behaviors prediction method (RST–ACO–BPNN) of welded joints based on RST, ACO and BPNN is proposed in this paper. The proposed RST–ACO–BPNN method utilizes the knowledge reduction ability of rough set theory for dealing with the original fatigue sample data, the minimum fatigue feature subset is obtained. The ant colony algorithm with the ability of strong global search was used to optimize the weights of BP neural network for obtaining the optimized BP neural network model. Then the minimum reduced subset was inputted into the optimized BP neural network model to construct the novel fatigue behaviors prediction model of welded joints by the continuous training and adjusting. To verify the correctness and validity of the novel RST–ACO–BPNN prediction model by applying in aluminum alloy welded joints. The simulation results show the proposed method can predict effectively the fatigue behaviors of aluminum alloy welded joints. And the RST–ACO–BPNN prediction method is provided with the merits of building model easily, simple structure, high precision and good generalization. Consequently, the prediction method can provide an effective approach to predict the fatigue behaviors of welded joints.