Using the ANN to Determine Effective Factors in Corner Break Distress in Jointed Concrete Pavements (JRCP)

Corner break is one of important distress in jointed reinforced concrete pavements (JRCP) which considerably influence riding quality and road smoothness which is defined as a difference in elevation across the joint. Since pavement distress prediction is crucial and important purpose in pavement management. Engineers need to be able to predict the distress upon easy-of-access data calls using a more comprehensive and applicable methods. As such, according to abilities and successful background of Artificial Neural Network (ANN), this method has been applied for corner break prediction purposes as the first step. The datasets of various categories, such as: design feature, slab concrete properties, climatic condition, base layer properties and sub-grade properties which were prepared using reliable database system. The system were trained and tested. Finally, ANN where evaluated based on two measurements, linear correlation coefficient and mean squared errors. The results showed that the correlation of aforementioned parameters varies between ۰.۷ and ۰.۹۵, beside very little amounts of errors. Upon the best amounts of these measures in testing process, optimized network were selected, showing that ANN can successfully predict corner break as an intelligent approach. Finally, input parameters of database were arranged stated by the results of sensitivity analysis of output parameters (Corner Break) to the input parameters in dataset. These results can be helpful for engineers in designing JRCP in future.

کلیدواژه ها:

Corner Break ، JRCP ، Artificial Neural Networks