基于支持向量机的氨含量预测模型

Prediction Model of Ammonia Concentration Based on Support Vector Machine

氨合成反应器出口氨含量与其影响因素间存在较强的非线性关系,为其建模,可预报氨含量,进而指导生产、优化反应器的操作。本文运用具有较强的非线性拟合能力和基于结构风险最小化原则的支持向量机,建立了氨含量的预测模型,验证表明,该模型具有较强的拟合和预测能力。

There was complex nonlinear relation between the ammonia concentration at the exit of the ammonia synthesis reactor and its influencing factors. Founded a model between them, it could predict the ammonia concentration, then guided production and optimize the operational parameters of the reactor. Support vector machine （SVM） was applied to model the ammonia concentration prediction, because it was strong capability of nonlinear expression, and it was favorable general performance based on structural risk minimization principle. The experiment indicated that the mean fitting relative error and mean prediction relative error of the model was all small.