Authors

Abstract

During the paddy rice drying process, the uneven spatial distribution of pore spaces within drying chambers poses a significant challenge to accurate porosity characterization and results in inefficient energy utilization. To address this issue, this study proposes a porosity prediction model based on Support Vector Regression (SVR), aimed at effectively monitoring porosity variations during drying and enhancing energy efficiency. Using MATLAB based image processing, the porosity of paddy rice was quantitatively extracted. A Response Surface Methodology (RSM) was then employed to analyze the influence of geometric characteristics, moisture content, and grain bulk height on porosity during drying. To further improve the predictive performance, the SVR model was optimized using the Snake Optimizer (SO) algorithm. The resulting SO-SVR model was evaluated against porosity values derived from image analysis. Experimental results demonstrate that the SO-SVR model achieves high accuracy, with a Root Mean Square Error (RMSE) of 0.0095 and a coefficient of determination (R²) of 0.9913. Compared to standard SVR and BP neural network models, the proposed model reduces RMSE by 0.0867 and 0.1663, and increases R² by 0.0449 and 0.1102, respectively. These findings indicate that the SO-SVR model provides a reliable and efficient approach for predicting paddy rice porosity during drying, offering valuable support for energy-saving and intelligent drying system design.

Abstract in Chinese

针对稻谷干燥过程中，干燥机内稻谷孔隙空间分布不均难以精准解析进而影响稻谷的干燥能耗的问题，建立一种基于支持向量回归（SVR）的稻谷孔隙率预测模型，高效检测稻谷干燥过程中孔隙率的变化，以达到节能效果。本文用MATLAB对稻谷图像处理得出稻谷孔隙率，用响应面（RSM）分析稻谷几何参数、含水率、粮堆厚度在稻谷干燥过程中对孔隙率的影响程度，利用蛇算法（SO）对支持向量回归（SVR）模型进行优化，建立稻谷孔隙率预测模型，并与基于图像处理后的稻谷孔隙率进行分析对比。结果表明： SO-SVR模型的均方根误差（RMSE）为0.0095、决定系数（R2）为0.9913，相对于SVR和BP算法的RMSE降低了0.0867和0.1663；R2提升了0.0449和0.1102。实验数据表明该模型的预测误差较小，具有更高的准确性，可以有效预测稻谷在干燥过程的孔隙率。