A knowledge-driven decision-support reference architecture for integrating cost estimation and optimization in open-cut pipeline construction

This is the authors' abstract. We don't add key points for this paper.

Open-cut pipeline construction (OCPC) involves complex cost behavior driven by uncertain and interdependent factors, making it a challenging domain for decision-making. Although machine learning, simulation, and optimization methods have been widely studied individually, their integration into coherent knowledge-driven decision-support systems remains limited. This study derives a structured and conceptual pathway linking cost estimation to cost optimization under uncertainty and uses it as the organizing framework for a systematic review. The pathway comprises seven interconnected domains – cost estimation, fuzzy logic, numerical modeling, machine learning, simulation, sensitivity analysis, and optimization – organized as a knowledge processing sequence rather than independent methods. A systematic analysis of 288 studies screened from 1,030 Web of Science records shows that existing research lacks structured integration: over 90% of studies examined in this review apply no more than two methods, and even the corpus’s most frequent pairing – cost estimation with optimization, 47 papers – occurs at 59% of the frequency that statistical independence would predict (φ = − 0.47). The study specifies the dependencies between methods as a knowledge-based reference architecture – specifying its parameter, model, and rule knowledge base; an ordered, forward-chained inference sequence; the input–output representation; and the decision logic separating system-supported analysis from planner judgment – and identifies five missing inferential links. The contribution is a measured map of the integration gap and a specified reference architecture for developing hybrid expert systems that unify predictive modeling, uncertainty reasoning, and optimization for OCPC decision support.