講演情報
[AI-11]Development of a Machine Learning-based Prediction Model for Forecasting the Pavement Surface Deterioration of Ghanaian Trunk Roads
*Belinda Mamphey1, Kazuya Tomiyama1 (1. Kitami Institute of Technology (Japan))
キーワード:
International Roughness Index、pavement condition score、machine learning、Ghanaian trunk roads
Road traffic is the most common transportation modes in Ghana, which handles majority of the passenger traffic, movement of goods and services in the country. The trunk road network falls under the supervision of the Ghana Highway Authority and constitutes a small percentage of the total road network. However, the trunk roads serve a crucial role since they are used for long distance travels and haulage purposes. The critical roads within the trunk road network in Ghana usually deteriorate sooner than expected. Although the government allocates funds each year for road maintenance, a significant section of the road networks within the country are in poor conditions. This problem results from inadequate budget funding, misplaced maintenance intervention and improper prioritisation among others. In Ghana Highway Authority, the road asset management system does not currently employ any prediction models. In line with this background, the main objective of this study is to develop a prediction model for forecasting the deterioration rate in terms of pavement conditions of Ghanaian trunk roads by incorporating machine learning-based methods. For this purpose, this study analyses the administrative data of pavements such as International Roughness Index (IRI), in-service year, pavement condition score and road categorization acquired by the Ghana Highway Authority for more than ten years. The final purpose of this study is to forecast the deterioration rate of a pavement in terms of pavement condition index (PCI) by developing a machine learning-based model using the acquired data. The outcome of this study will help to improve the pavement management system in Ghana by introducing a more proactive, cost effective and data driven approach.
