Camellia oleifera seeds with high moisture content are prone to decay, leading to oil rancidity and seriously affecting subsequent processing and tea oil quality. Thus, an efficient drying method is needed. Hot air drying with ultrasonic pretreatment is a potential drying method for camellia oleifera seeds. In this paper, the effects of key parameters such as ultrasound power, ultrasound pretreatment time, and hot air drying temperature on the moisture ratio of camellia oleifera seeds, as well as the changes in microstructure of both seed shell and seed kernel, and the main quality parameters such as peroxide value and acid value of tea oil were explored. Results showed that the shorter the ultrasound pretreatment time, the greater the ultrasound power, the higher the temperature, the faster the drying rate, and the shorter the drying time. The drying time and specific energy consumption was reduced by up to 26.7 % and 16.8 % with the ultrasonic pretreatment, respectively. The optimal ultrasonic pretreatment time is 2 min, and the ultrasonic power is 300 W. A Genetic Algorithm Optimized Backpropagation Artificial Neural Network (GA-BP-ANN) drying model was proposed to predict the changes in moisture ratio during the drying process of camellia oleifera seeds to provide guidance for the development of intelligent drying system. The experimental results showed that the proposed drying model had a good predictive performance, with an RMSE of 0.0076969 and an R² of 99.909 %, which can accurately estimate the effect of ultrasonic pretreatment on hot air drying. The research results of this study are expected to reduce post-harvest losses of camellia oleifera fruits and provide theoretical support for the development of drying systems of camellia oleifera seeds and similar agricultural products.

油茶种子水分含量高，容易腐烂，导致油脂酸败，严重影响后续加工和茶油品质。因此，需要一种有效的干燥方法。超声波预处理热风干燥是一种潜在的油茶种子干燥方法。本文研究了超声功率、超声预处理时间、热风干燥温度等关键参数对油茶种子含水率的影响以及种子壳和种子仁微观结构的变化，并主要研究探讨了茶油过氧化值、酸价等质量参数。结果表明，超声波预处理时间越短，超声波功率越大，温度越高，干燥速度越快，干燥时间越短。通过超声波预处理，干燥时间和单位能耗分别减少了 26.7% 和 16.8%。最佳超声波预处理时间为2 min，超声波功率为300 W。提出了遗传算法优化反向传播人工神经网络（GA-BP-ANN）干燥模型来预测油茶干燥过程中水分比的变化。为种子智能干燥系统的发展提供指导。实验结果表明，所提出的干燥模型具有良好的预测性能，RMSE为0.0076969，R ²为99.909%，能够准确估计超声波预处理对热风干燥的效果。 本研究成果有望减少油茶果实采后损失，为油茶种子及同类农产品干燥系统的开发提供理论支持。