ارائه چارچوب پشتیبان تصمیم مبتنی بر هوش مصنوعی توضیحپذیر برای هشدار زودهنگام مدیریتی در کنترل پروژههای ساختمانی
ارائه چارچوب پشتیبان تصمیم مبتنی بر هوش مصنوعی توضیحپذیر برای هشدار زودهنگام مدیریتی در کنترل پروژههای ساختمانی
Seyyed Azim Hosseini1 Mohammad Mahdi Chegini2
1) 1. Department of Civil Engineering, ST.C., Islamic Azad University, Tehran, Iran
2) 2. PhD Student, Construction Engineering and Management, Department of Civil Engineering, Islamic Azad University, South Tehran Branch, Tehran, Iran
IOI Code
:
AAHC-ACZBKG
Publication :
National Conference on Management, Commercialization, Knowledge-Based Economy and Investment for Production(impeb.ir)
Abstract :
Construction project control requires timely interpretation of schedule, cost, progress, risk, change, procurement, and resource information. Earned Value Management (EVM) provides standardized diagnostic and forecasting functions, including trend analysis and estimates at completion, but conventional EVM has limited ability to integrate heterogeneous contextual drivers, nonlinear interactions, and explainable multivariate risk. This conceptual paper proposes an explainable-AI decision-support framework for managerial early warning in construction project control. The method is explicitly a targeted narrative literature synthesis and framework-development exercise rather than a PRISMA systematic review or an empirical model-validation study. The framework s novelty is positioned in an end-to-end governance chain that connects status-date project data to a specified two-period breach-prediction task, calibrated probability and uncertainty, local model explanation, governance-oriented probability-impact prioritization, an approved project-control action playbook, human approval, and intervention-aware feedback. The primary use case is binary classification at each project reporting period: estimating the probability that a project will cross an organization-defined red schedule-control threshold within the next two reporting periods. A transparent 12-period EVM scenario is used only to illustrate control logic, thresholding, data alignment, and managerial communication; it is not field data and no trained-model, SHAP, or validation statistics are fabricated. The framework further requires project-wise and time-wise out-of-sample validation, comparison with non-AI EVM warning rules, managerial early-warning metrics such as lead time and false alarms per project, and explicit separation between predictive attribution and causal evidence. AI is therefore positioned as a governed decision-support component that augments, rather than replaces, project-manager accountability.
Keywords : construction project control explainable artificial intelligence earned value management early warning human in the loop governance risk prioritization