Smart material estimation for the engineering, procurement, and construction (EPC) sector

Artikel i vetenskaplig tidskrift, 2025

This study presents a novel AI-based approach that integrates deep learning techniques for symbol and text recognition with predictive modeling based on historical project data. The aim is to automate and enhance material cost estimation and procurement in Engineering, Procurement, and Construction (EPC) projects. Unlike existing methods, our approach combines data extraction from Piping and Instrumentation Diagrams (P&IDs) with predictive modeling to improve estimation accuracy. In addition, we introduce methods such as tiling and augmentation to optimize the accuracy of symbol recognition in complex and noisy industrial diagrams. We also present methods for managing diverse symbology, improving annotations, and handling background noise in actual industrial blueprints. Furthermore, we apply domain-specific knowledge rules while utilizing available historical data repositories from past engineering projects. Our findings suggest significant potential for engineering time and cost savings in large-scale EPC projects, supported by empirical analysis of development costs in relation to engineering hours saved.