基于PEGO-Kriging模型的闽北乌龙茶烘焙机优化与试验

Optimization and Testing of Baking Machine for Northern Fujian Oolong Tea Based on PEGO-Kriging Model

摘要:

为提升闽北乌龙茶烘焙机性能,本文提出PEGO(Parallel efficient global optimization）算法对Kriging代理模型进行迭代寻优,以优化烘焙机结构并开展试验验证。采用最优拉丁超立方试验设计,以导流板倾斜角、排风口孔径、托盘数和侧边梯度孔孔数差值为优化变量设计30组样本数据,并进行数值模拟得到箱体温度标准差作为响应;根据样本数据构建描述优化变量与响应之间映射关系的Kriging代理模型,为求解闽北乌龙茶烘焙机结构参数的最优解集奠定基础;提出PEGO算法对Kriging代理模型进行迭代优化,得出最优参数组合并进行Spearman相关性分析。试验结果表明,托盘数与箱体温度标准差相关系数为-0.83,达到高相关程度;侧边梯度孔孔数差值与箱体温度标准差相关系数为-0.017,影响可忽略;最优参数组合为导流板倾斜角8.30°、排风口孔径25 mm、托盘数9个、侧边梯度孔孔数差值3个;相较于原型烘焙机,闽北乌龙茶新型烘焙机箱体温度标准差下降86.49%,温度相对标准偏差降幅最大达73%,速度相对标准偏差降幅最大达43%,烘焙机内部受热更均匀;新型烘焙机叶片含水率为3.88%,相较于原型烘焙机下降13.78%,烘焙机性能得到有效提升。研究结果可为闽北乌龙茶烘焙机结构优化与性能提升提供重要理论基础和技术支持。

Abstract:

Aiming to enhance the performance of the Northern Fujian Oolong tea baking machine, the parallel efficient global optimization (PEGO) algorithm for iterative optimization of a Kriging surrogate model was proposed to optimize the structure of a baking machine and conduct experimental validation. An optimal Latin hypercube design was employed, with 30 sample sets generated using the following optimization variables: the inclination angle of deflector plate, the aperture of exhaust vent, tray count, and difference in the number of holes on the side gradient. Numerical simulations were conducted to obtain the standard deviation of chamber temperature as the response variable. Based on the sample data, a Kriging surrogate model describing the mapping relationship between optimization variables and response was constructed, laying the foundation for solving the optimal parameter set of the Northern Fujian Oolong tea roaster. The PEGO algorithm was proposed to iteratively optimize the Kriging surrogate model, yielding the optimal parameter combination and performing Spearman correlation analysis tests. Results indicated that the correlation coefficient between tray count and the standard deviation of chamber temperature was -0.83, indicating a high correlation;the correlation coefficient between difference in the number of holes on the side gradient and the standard deviation of chamber temperature was -0.017, with negligible influence;the optimal parameter combination was: the inclination angle of deflector plate was 8.3°, the aperture of exhaust vent was 25 mm, tray count was 9, and difference in the number of holes on the side gradient was 3. Compared with the original roaster, the novel baking machine for Northern Fujian Oolong tea reduced the standard deviation of chamber temperature by 86.6%, with the maximum reduction in relative temperature standard deviation reaching 73% and relative velocity standard deviation reaching 43%, indicating more uniform internal heating. The moisture content of the blades in the new baking machine was 3.88%, a decrease of 13.78% compared with the prototype baking machine, demonstrating effective performance enhancement. The research result can provide a crucial theoretical foundation and technical support for structural optimization and performance enhancement of the Northern Fujian Oolong tea baking machine.