PHYSICS-DATA MODELING FOR ENHANCED PERFORMANCE OF AUTOMATED CONSTRUCTION SYSTEM

Ensuring both operational efficiency and structural safety in Automated Construction Systems (ACS) remains a critical challenge, particularly under complex high-rise construction environments characterized by uncertain loads and limited sensing precision.To address this issue, this study develops a hybrid physics-data-driven modeling framework to improve jacking efficiency and structural reliability of ACS under such complex conditions.The framework integrates a Physics-Informed Neural Network (PINN) for predicting hydraulic cylinder strokes with a Generative Adversarial Network (GAN)based anomaly detection model for structural safety evaluation.These models are embedded within a multi-objective optimization strategy, which is solved using the Non-dominated Sorting Genetic Algorithm II (NSGA-II) and further refined through the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS).Validation using real-world ACS monitoring data demonstrates that the proposed framework can significantly enhance jacking efficiency while maintaining structural safety within acceptable thresholds.The findings underscore the potential of combining physical modeling, machine learning, and evolutionary optimization to enable intelligent control of complex construction machinery.