中文摘要：

针对甲醇生产过程中高度的非线性和时变性,采用精确在线支持向量机模型预测粗甲醇的转化率.在线支持向量机模型一般采用单一的核函数,混合核函数可以弥补单一核函数的不足,提高模型的泛化能力和学习能力.为了使模型的预测精度进一步提高,在混合核函数的基础上运用在线误差校正方法.将基于混合核函数和误差校正的在线支持向量机建模方法应用在煤制甲醇数据上,通过与传统支持向量机和准确在线支持向量机模型对比,仿真实验和分析结果表明改进的在线支持向量机模型比传统支持向量机预测精度高,能够实现粗甲醇转化率的实时预测,从而更好的指导甲醇生产.

英文摘要：

For the high nonlinearity and time variation of methanol production process, an online crude methanol conversion rate prediction model based on online support vector machine is here used. Online support vector machine models generally use a single kernel funetion. Mixed kernel funetion can make up for the lack of a single kernel function, and improve the ability of generalization and learning In order to improve a better prediction accuracy, error correction methads based mixed kernel function is used. Online support vector machine which combined mixed kernel function and error correction methods is used in coal methanol data, simulation experiments and the results show that a higher prediction precision than the traditional support vector regression, which can achieve the real-time of crude methanol conversion rate and better guide the methanol production.