摘要
由于变压吸附(PSA)过程中吸附剂的参数不易获得,对沼气净化PSA过程中产品气甲烷浓度的机理建模是比较困难的。提出了基于最小二乘支持向量机(LS-SVM)甲烷浓度的建模方法。分析比较了系统辨识、RBF神经网络与LS-SVM模型,结果表明,运用LS-SVM方法建立的模型在精度上明显优于其他两种的方法,从而验证了LS-SVM在PSA过程产品浓度建模中是有效的。
In the pressure swing adsorption (PSA) process, the parameters of the adsorbent were not easy to acquire, so it's hard to build the mechanism model on PSA process for biogas purification. A method of establishing a methane concentration model based on least square support vector machine (LS-SVM) was proposed, and a comparison of the system identification, RBF neural network and LS-SVM models was made. The results indicated that the LS-SVM method was nmch better than the other two methods, which vei~.fied its effectiveness for modeling methane concentration in PSA process.
出处
《天然气化工—C1化学与化工》
CAS
CSCD
北大核心
2013年第1期36-38,50,共4页
Natural Gas Chemical Industry
基金
国家自然科学基金项目(21046004)
关键词
沼气提纯
变压吸附
PSA
LS-SVM
系统辨识
建模
biogas upgrading
pressure swing adsorption
PSA
LS-SVM
system identification
modeling