基于LSSVM的木材干燥在线建模研究

Online modeling for wood drying based on least squares support vector machine

介绍了最小二乘支持向量机(LSSVM)回归原理,针对木材干燥系统的强耦合、强非线性等特点,提出以LSSVM方法建立木材干燥系统在线模型。模型以干燥实验获取的减速干燥阶段数据为样本,根据实际预测控制需要,建立木材干燥系统的在线预测模型,并进行在线预测。仿真结果表明,基于LSSVM的木材干燥在线模型能够实时反映系统当前状态,在线更新训练样本,滚动优化模型结构并预测系统下一步输出,模型结构简单,泛化能力强,预测精度高,能够满足实际干燥控制的需要。

This paper introduced principle of least squares support vector machine regression. An online wooa drying model based on LSSVM was built for drying process with severe nonlinear and coupling. The sample data was acquired by the experiment of wood drying with a small drying kiln. All decelerating wood drying online predict was processed for simulation. Result of simulation experiment shows that the online model of wood drying has a high predict precision, strong extensive use and simple frame, which can reflect real time system state, online predicting and updating model frame. Online LSSVM wood drying model is feasible and effective for wood drying model predicting control system.