A DPSO-BP NN modeling for predicting mechanical property: a case of 6181H18 aluminum alloy

Aluminum alloy is widely used in daily life due to their good properties. In order to get the change rule of the mechanical properties of 6181H18 aluminum alloy, a detecting particle swarm optimization (DPSO) algorithm was adopted to update weights and thresholds of back propagation neural network (BP NN) in an innovative way. In this way, a DPSO-BP NN prediction model was established to improve the prediction accuracy and was applied to predict the peak stresses of 6181H18 aluminum alloy. The results show that the predicted values obtained based on BP NN and DPSO-BP NN are both very close to the experimental ones and they can reflect the variation law of the stresses of 6181H18 aluminum alloy. It is confirmed that the DPSO-BP NN has a higher prediction accuracy by the mean relative error, standard residual, R-squared and root mean square error (RMSE). The established DPSO-BP NN prediction model owns better prediction capability compared with the traditional BP NN model. The results of this study can provide a scientific basis for the improvement of mechanical properties of alloy materials, and offer a technical reference for technical workers in related fields.