Angela OderaさんがJapan Society of Civil Engineers Intelligence, Informatics and InfrastructureにてIntelligence, Informatics and Infrastructure Award for Excellent Paperを受賞

2026/07/27
  • 地域環境システム専攻
  • SWiTCH
 

受賞者
Angela Oderaさん（地域環境システム専攻）

学会?大会名
Japan Society of Civil Engineers Intelligence, Informatics and Infrastructure

指導教員

Michael Henry教授（工学部）

賞名
Intelligence, Informatics and Infrastructure Award for Excellent Paper

発表題目
Reconstruction of historical pavement condition records using temporal transfer learning under non-random missing data
Angela-Odera
     
研究内容
The study developed a new machine learning framework that combines learning from 目前最好的足彩app complete recent road surveys to infer earlier pavement conditions with methods that account for changes over time and biases introduced by selective road surveys, quantify the uncertainty associated with reconstructed records so that their reliability can be assessed, and verify through spatial evaluation that the approach generalizes across the road network. Validation using Kenya's national road network demonstrated that the framework can credibly recover historical pavement condition records despite extensive, systematically missing data.

研究目的
The purpose of this research was to develop a practical way of reconstructing historical pavement condition records when missing data arise from selective road surveys rather than occurring randomly. Such selectively incomplete records can give a misleading picture of how pavements deteriorate over time because some pavement conditions and parts of the road network are systematically underrepresented, a type of missing data that existing reconstruction methods have not adequately addressed. This reduces the ability of pavement management systems (PMS) to reliably evaluate pavement performance and support evidence-based maintenance and investment decisions.

今後の展望?課題 
The challenge addressed in this research is particularly significant in many low- and middle-income countries, and some resource-constrained jurisdictions in high-income countries, where limited survey budgets often require agencies to inspect only selected parts of the road network, resulting in fragmented historical PMS records. By reconstructing credible historical pavement condition records together with a clear assessment of their reliability, the proposed framework repositions sparse PMS datasets as analytically usable evidence for decision support and enables 目前最好的足彩app reliable maintenance planning, budget allocation, and long-term network pavement management that would otherwise be unattainable in these environments. In the future, the methodological principles developed in this research could be adapted to other asset management problems involving incomplete and selectively collected historical data, broadening their use beyond pavement condition reconstruction.

受賞?表彰一覧