在微波加热过程中加热介质在不同温度阶段有不同的内部特性，传统的温度预测方法难于同时对加热介质低温段与高温段温度取得满意的预测结果。为此提出了一种基于ANFIS 的分段温度预测模型，该方法建立基于K均值聚类法的温度划分机制，并采用不同结构的ANFIS预测加热介质不同温度阶段的温度。低温阶段构建常规ANFIS预测温度，高温阶段利用减法聚类能从数据中确定模糊规则的特性构建ANFIS预测温度。仿真结果表明，与采用单一结构的ANFIS和BP（back propagation）神经网络的预测结果相比，ANFIS分段温度预测模型可同时在加热介质低温段与高温段取得较好的预测结果，模型效率可达到97．41％，显著提高了预测准确率，这有助于提高实际微波加热过程的生产效率和安全性。%During the microwave heating process, materials in different temperature regions have different internal characteristics. Using traditional temperature forecasting methods, it is difficult to obtain satisfactory prediction re⁃sults for both low⁃and high⁃temperature sections in a medium. To solve this problem, this study proposes a new tem⁃perature⁃sectioned forecasting model based on the ANFIS ( adaptive neuro⁃fuzzy inference system) . For this meth⁃od, we established a temperature⁃division mechanism based on K⁃means clustering. Additionally, we used an AN⁃FIS with different structures to forecast the temperature of the heated medium at different stages. We also construc⁃ted a conventional ANFIS to predict a material�s low temperature and a subtraction⁃clustering ANFIS that determines the fuzzy rules from data to predict a material�s high temperature. Simulation results demonstrate that the proposed method achieves satisfactory results for both low⁃and high⁃temperature sections when compared to ANFISs and BP ( back propagation) networks with a single structure. Model efficiency can reach 97.41% and the prediction accura⁃cy is significantly improved. The proposed model can improve the efficiency and safety of the microwave heating process.
展开▼
