Role of AI in Construction Planning and Forecasting
Vuppalapadu, Sandeep Teja Reddy (2025)
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2025091524679
https://urn.fi/URN:NBN:fi:amk-2025091524679
Tiivistelmä
The construction industry has traditionally depended on established planning and forecasting techniques, such as the Critical Path Method (CPM), Program Evaluation and Review Technique (PERT), and statistical forecasting. While effective in structured environments, these methods often struggle to adapt to the increasing complexity, uncertainty, and dynamic nature of modern construction projects. The aim of this thesis is to investigate how Artificial Intelligence (AI) can be applied to enhance construction planning and forecasting, particularly in the areas of project scheduling, risk management, and cost and workforce prediction. By analyzing both conventional and AI-driven approaches, this work seeks to demonstrate the potential of AI as a transformative tool for improving efficiency, reliability, and decision-making in construction management.
The research process consisted of a literature-based study that systematically examined AI algorithms and their application in construction. The methods reviewed include evolutionary algorithms such as Genetic Algorithms (GA) and Ant Colony Optimization (ACO) for project scheduling, predictive analytics and digital twins for risk management, and machine learning models including Gradient Boosting, Long Short-Term Memory (LSTM) networks, and Back Propagation (BP) neural networks for forecasting. Case studies were analyzed to provide practical insights into the implementation and outcomes of these AI techniques in real-world projects. Materials for the thesis comprised published scientific research, industry reports, and documented case studies that highlight the integration of AI with existing construction planning tools.
The results show that AI-based methods significantly improve scheduling resilience, risk detection, and forecasting accuracy compared to traditional techniques. Genetic algorithms and ACO have demonstrated superior performance in generating optimized and resilient project schedules, reducing both time delays and cost deviations. In risk management, predictive analytics and digital twin simulations provided earlier detection of hazards and better preparedness strategies, while computer vision supported real-time safety monitoring. Forecasting studies revealed that models such as XGBoost, LSTM, and optimized BP neural networks achieved high predictive accuracy for workforce needs and project costs, while also uncovering hidden correlations useful for long-term planning. Overall, the findings confirm that AI is not only feasible but already impactful in construction planning and forecasting, though challenges remain in data acquisition, computational resources, and the need for skilled expertise.
The research process consisted of a literature-based study that systematically examined AI algorithms and their application in construction. The methods reviewed include evolutionary algorithms such as Genetic Algorithms (GA) and Ant Colony Optimization (ACO) for project scheduling, predictive analytics and digital twins for risk management, and machine learning models including Gradient Boosting, Long Short-Term Memory (LSTM) networks, and Back Propagation (BP) neural networks for forecasting. Case studies were analyzed to provide practical insights into the implementation and outcomes of these AI techniques in real-world projects. Materials for the thesis comprised published scientific research, industry reports, and documented case studies that highlight the integration of AI with existing construction planning tools.
The results show that AI-based methods significantly improve scheduling resilience, risk detection, and forecasting accuracy compared to traditional techniques. Genetic algorithms and ACO have demonstrated superior performance in generating optimized and resilient project schedules, reducing both time delays and cost deviations. In risk management, predictive analytics and digital twin simulations provided earlier detection of hazards and better preparedness strategies, while computer vision supported real-time safety monitoring. Forecasting studies revealed that models such as XGBoost, LSTM, and optimized BP neural networks achieved high predictive accuracy for workforce needs and project costs, while also uncovering hidden correlations useful for long-term planning. Overall, the findings confirm that AI is not only feasible but already impactful in construction planning and forecasting, though challenges remain in data acquisition, computational resources, and the need for skilled expertise.