基于RF-MLP的烘丝出口含水率预测

Prediction of moisture content at the outlet of cut tobacco drying based on RF-MLP

针对烟丝烘丝出口含水率预测精度不足导致水分控制参数难以准确设置的问题,提出了基于随机森林和多层感知器的烘丝出口含水率预测方法。首先,采用随机森林方法提出烘丝出口含水率的相关特征,构建含水率特征矩阵。其次,将特征矩阵作为多层感知器的输入,建立特征量与烘丝出口含水率的5层隐含层预测模型。最后,通过网格搜索算法进行预测模型的结构优化。结果表明,在烘丝含水率预测中,文中方法的模型决定系数达到了0.97,优于对比方法。文中方法提高了烘丝出口含水率预测精度,可辅助含水率控制策略的制定,进而提高烟丝品质。

Aiming at the problem that the moisture control parameters is difficult to set accurately due to the insufficient prediction accuracy of the moisture content of the outlet of the cut tobacco drying,a method of predicting the moisture content at the outlet of the cut tobacco drying based on random forest and multilayer perceptron was proposed.Firstly,the relative characteristics of the moisture content of the outlet of the cut tobacco drying are proposed by using the random forest method,and the moisture content feature matrix is constructed.Secondly,using the feature matrix as the input of the multilayer perceptron,a 5-layer hidden layer prediction model of feature variable and moisture content at the outlet of cut tobacco drying was established.Finally,the structure optimization of the prediction model is carried out by grid search algorithm.The results show that the model determination coefficient of the method of this paper reaches 0.97 in the prediction of the moisture content of cut tobacco drying,which is better than the comparison method.The method in this paper improves the prediction accuracy of the moisture content at the outlet of the cut tobacco drying,which can assist the formulation of the moisture content control strategy,thereby improving the quality of the cut tobacco.