The ANN model has great potential in predicting durability of geasynthetics
because of its unique learning. training and prediction characteristics. By contrast to the regression method
the ANN method doesn’t demand the regulation of the experiment data. Based on 31 groups of experimental data obtained from aging tests of non-woven polyester
a BP neural network model for predicting durability of geosynthetics is developed. The model proposed has four parameters (the temperature
relative humidity
aging time and ultraviolet radiation) at input layer and two parameters (strength and percentage elongation) at output layer. The comparison between the prediction values and experimental results shows that the model proposed has high accuracy for predicting durability of geocynthetics.