Investigation of forecast models for Marshall experimental results

Today, asphalt roads are exposed to increasing traffic loads in recent times. It is important to obtain a quality and healthy road coverage when considering the conditions in which our freight and passenger transportation is carried out by roads. One of the most important aspects of asphalt road design is the determination of the optimum percentage of bitumen. The Marshall stability test is utilized for the percentage of bitumen. In our work, instead of the long and laborious Marshall experiment process, alternative statistical methods and artificial intelligence methods were used. Models were developed for Marshall stability predictions which are Marshall experiment results. Prediction models are constructed using the results of previous experiments. These parameters are the bitumen penetration (P), weight of the sample in the weather (H), the temperature (C), the bitumen weight (G), the sample heights (Y), the bitumen percentage (W), weight of the sample in water (S), the stability (ST). Regression and Artificial Neural Networks (ANN) techniques are being used when creating models. For the performance comparison of the techniques used the correlation coefficient (R), the mean percentage errors (MPE) and the mean square errors (MSE) are used. It is seen that the model with regression is the model with the highest performance value HPCYWSST. The correlation coefficient of the HPCYWSST model is 0,57125, the mean square errors is 14841,82, the mean percentage errors is 9,5858 percent. It is seen that HCPGSST is the model with the highest performance value from the models generated by random selection of data sets with ANN. The correlation coefficient of the HCPGSST model is 0,75168. The mean square errors of the HCPGSST model is 9616,43. The mean percentage error of the HCPGSST model is 7,657227. As a result of these comparisons, it can be seen that the models formed by ANN have reached to the results with smaller errors from regression models.