Harris Hawks optimization algorithm for band combination selection from near-infrared spectroscopy to predict the tensile strength of Acer mono Maxim

Tensile strength is an important indicator of the mechanical properties of wood. It should be tested non-destructively to avoid damaging the wood. This paper develops a predictive model for the tensile strength of Acer mono Maxim. by using near-infrared spectral data and a partial least squares regression (PLSR) model. To address the limitations of existing models, such as weak interpretability and susceptibility to overfitting, a Newly Improved Harris Hawks Optimization (NIHHO) algorithm is proposed to select wavebands associated with tensile strength. This algorithm employs Latin hypercube sampling and a V-type transfer function to generate initial binary waveband combinations, ensuring uniformity in band selection. Additionally, game-theoretic and adaptive flight strategies are introduced to enhance the diversity of feature wavelength band combinations while balancing global exploration with local development. The experiment was conducted using spectral data from 60 A. mono samples. The results indicate that the combination of feature wavelength bands selected by the NIHHO algorithm demonstrated superior predictive capability. Specifically, an R 2 value of 0.99 was achieved, with a root mean square error of 3.77 MPa and a mean absolute error of 3.02 MPa. The values obtained exceed those from Genetic Algorithms, Particle Swarm Optimization, and Ant Colony Optimization. These findings demonstrate that applying NIHHO to enhance the PLSR model effectively identifies spectral bands highly correlated with tensile strength, thereby improving detection accuracy. This advancement will provide a more reliable quality basis for engineering material selection.