Research and applications of artificial neural network in pavement engineering: A state-of-the-art review

Given the great advancements in soft computing and data science, artificial neural network (ANN) has been explored applied to handle complicated problems field of pavement engineering. This study conducted a state-of-the-art review for surveying recent progress ANN application at different stages engineering, including design, construction, inspection monitoring, maintenance. focused on papers published over last three decades, especially studies since 2013. Through literature retrieval, total 683 this were identified, among which 143 selected an in-depth review. The architectures used these mainly included multi-layer perceptron (MLPNN), convolutional (CNN) recurrent (RNN) processing one-dimensional data, two-dimensional time-series data. CNN-based health monitoring attracted largest research interest due its potential replace human labor. While proved be effective tool material cost analysis, defect detection maintenance planning, it is facing huge challenges terms collection, parameter optimization, model transferability low-cost annotation. More attention should paid bring multidisciplinary techniques into engineering tackle existing widen future opportunities.