Optimization of Planting and Harvesting Patterns in Wood Farming Using Wood Electrical Resistance Data and Artificial Intelligence Algorithms

Wood farming, as a strategic approach to supplying raw materials for the pulp and paper industry while reducing pressure on natural forests, increasingly requires novel optimization methods for planting and harvesting patterns. This paper investigates the integrated application of wood electrical resistance data and artificial intelligence algorithms to optimize harvest timing, assess internal stem quality, and reduce premature harvesting. Electrical Resistivity Tomography (ERT) serves as a non-destructive technology that enables in-situ evaluation of living tree internal conditions without damage. This study reviews advanced machine learning models including Support Vector Regression (SVR), Backpropagation Neural Networks optimized with the Improved Harris Hawk Optimization (IHHO-BP), and Convolutional Neural Networks (CNN) for analyzing resistivity data, estimating relative water content, detecting internal anomalies, and classifying stem quality. Results from previous studies indicate that the Privileged Information SVR (PI-SVR) model reduces moisture estimation error to below ۶% using environmental and physiological privileged information. Furthermore, TinyML-based systems combined with ERT can detect ۱۵ out of ۱۶ simulated anomalies. The IHHO-BP model improves mean absolute error by at least ۵۱% compared to baseline models. This integrated approach can increase profitability, reduce premature harvesting, and promote sustainable wood resource management.

کلیدواژه ها:

Wood farming ، harvest optimization ، electrical resistivity tomography ، machine learning ، artificial neural network ، metaheuristic algorithm.