基于SVM的木材纤维饱和点近区段含水率预测模型研究

Moisture Content Prediction Model Near Lumber Fiber Saturation Point Based on SVM

[目的]研究木材纤维饱和点近区段含水率预测模型。[方法]电阻法测量木材纤维饱和点近区段含水率会出现测量值突然偏离真值的现象,即出现测量"盲点"。在研究检测原理的基础上,提出利用支持向量机方法对已测木材含水率、温度和湿度数据进行训练建模,通过模型预测得出纤维饱和点近区段含水率数值。[结果]支持向量机方法建立的模型能够预测木材纤维饱和点近区段含水率数值,模型泛化能力强,预测精度高,而且只需要少量样本数据就可以实现预测,很好地解决了电阻法在测量过程中的"盲点"问题。[结论]支持向量机预测模型提高了木材干燥过程中全量程含水率的检测精度,对木材干燥过程的含水率建模具有一定研究意义。

[ Objective ] To study the moisture content （ MC ） prediction model near lumber fiber saturation point （FSP）. [ Method ] It would be suddenly appeared the phenomenon that the measured values could deviate from the true value when resistance method is used to measure MC prediction near lumber FSP,which means the measuring blind point is emerged. Based on detection principle,this paper is proposing a method, which is based on support vector machine （SVM）, to train the model by the measured MC,temperature and humidity, and it could get the MC pre- diction near lumber FSP by this model. [ Result ] The model built by SVM, which maintain the character of stronger generalization ability,higher prediction accuracy and accomplishing prediction by less sample data,could predict MC prediction near lumber FSP precisely. It also solves the problem of blind point that appears measured MC by resistance method. [ Conclusion ] The prediction model built by SVM could improve the de-tection accuracy of MC in wood drying process. So it also has significance meaning to build the MC model in wood drying process by using this model.