花椒真空干燥特性分析及动力学模型研究

Vacuum drying characteristics and kinetics modeling study of Zanthoxylum bungeanum

为了探求花椒真空干燥特征,提高干燥速率,实现花椒真空规模化干燥。选取干燥温度为50、60、70℃,真空度为-0.06、-0.04、-0.02 MPa,装载量为30、40、50 g进行全面试验,考察3个因素对花椒干燥总时间和干燥速率的影响。利用7种经典干燥数学模型对9组实验数据进行非线性拟合,利用决定系数R2、残差平方和(SSE)、卡方χ2三个指标对拟合结果进行评价,选取其最优模型与建立的BP神经网络模型进行对比检验。结果表明,三次多项式模型的拟合R2值都达到0.999、SSE值最低达到了7.0E-4,χ2值最低达到3.08E-5是7种经典模型中描述花椒真空干燥动力学特性的最优模型。但三次多项式模型拟合检验的平均相对误差值为1.98%高于BP神经网络模型的平均相对误差值(1.13%),因此BP神经网络是更适合描述花椒干燥动力学特性的数学模型。

The vacuum drying technology was applied in this experiment to study vacuum drying characteristics of Zanthoxylum bungeanum to improve the drying rate and achieve large-scale vacuum drying production. Under the conditions of temperature（50,60,70 ℃）,vacuum（-0. 06,-0. 04,-0. 02 MPa）,and loading（30,40,50 g）comprehensive tests,the effects of three factors on total drying time and drying rate of Zanthoxylum bungeanum were investigated. Seven different drying models were used to simulate the observed drying data and the comparative results of the value of R2,SSE and χ2 showed that the cubic polynomial model was much more adequate model for describing the vacuum drying of Zanthoxylum bungeanum. BP neural network was established to compare with cubic polynomial model. The results show that the cubic polynomial model fits with R2 of 0. 999. The lowest value of SSE is 7 E-4,and the lowest value of χ2 is 3. 08 E-5. It is the best model in describing the dynamic characteristics of Zanthoxylum bungeanum vacuum drying among seven classical models. However,the average relative error of the cubic polynomial fitting test is 1. 98%,higher than the average relative error 1. 13% of BP neural network model. Therefore,BP neural network is the optimal mathematical model to describe the drying dynamics of Zanthoxylum bungeanum.