Wheat Moisture Content Detection Method Based on Data-Driven Microwave Multifrequency Optimization Under Multiparameter Influence
Abstract
Non-destructive measurement of wheat moisture content is of paramount importance for minimizing grain loss and guaranteeing quality, yet its accuracy is easily compromised by various external environmental factors. To compensate for this, this research develops a microwave detection platform for wheat MC, and explores the impacts of external parameters on MC detection by adjusting the antenna distances, sample thicknesses and frequency. Through machine learning, this research established a wheat MC prediction model using the RF algorithm, and proposed a dual feature parameter optimization method based on data-driven methods that combines clustering analysis with swarm intelligence algorithms, achieving frequency subset screening applicable to different sample thicknesses. Based on the data characteristics, an optimal four-frequency subset with the highest sensitivity was identified, which effectively improved the MC detection accuracy (R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>=0.9905, RMSE=0.9722, MSE=0.9452). The methodology provides an innovative solution for implementing microwave technology in grain product inspection, effectively balancing measurement accuracy with operational efficiency.