Machine learning-assisted wood materials: Applications and future prospects

Abstract

Wood and wood-based materials, surpassing their conventional image as mere stems and branches of trees, have found extensive utilization in diverse industrial sectors due to their low carbon footprint. Nonetheless, maximizing wood utilization and advancing multifunctional wood materials face challenges due to resource-intensive conventional approaches. Integrating machine learning (ML) in wood mechanics has emerged as a promising avenue for deeper exploration of this remarkable material. By leveraging advanced computational techniques, researchers can delve into wood's intricate properties and behavior, unraveling the complex interactions between its chemical constituents, microstructures, and mechanical characteristics. Combined with imaging and sensor technologies, ML contributes to efficient, fast, and real-time health detection of wood materials. This review aims to illuminate the transformative impact of ML in unlocking the hidden potential of wood, fostering innovative applications, and facilitating sustainable engineering solutions. The basic workflow of ML and its typical applications in property prediction, defect detection, and optimized design of wood materials are discussed, thereby highlighting the challenges and the need for future research.