پیش بینی فضای خالی مخلوط های آسفالتی در جاده های تحت سرویس با استفاده از شبکه های عصبی مصنوعی

The void of the asphalt mixes is one of the crucial parameters in the design and performance of asphalt mixes in the design construction and maintenance of pavements. Changes in this parameter after road construction under traffic and over time, cause changes in the performance of asphalt mixtures. Therefore, predicting the void of asphalt mixes in the roads under service is an essential requirement for assessing asphalt performance. In this research, an AAN-based model with the high accuracy of R2 =0.97 has been developed to predict the void of asphalt mixes using feed-forward Artificial Neural Networks with the Levernberg-Marquardt Back Propagation training algorithm. The LMBP algorithm is a dynamic technique that combines the speed of the Gauss-Newton method with the convergence guarantee of the Steepest Decent method. Furthermore, the method of adjusting the training parameters of the ANN models is presented to increase the possibility of achieving higher accuracies in the process of reinitializing the weights and retraining the ANN models.