提出基于三段膜分离过程的智能模型,并应用它在线分析炼厂气氢回收过程中的关键性能参数.首先,应用网格搜索和交叉验证,结合贝叶斯估计得到最小二乘支持向量机的sig2和gam参数的最优值;然后,建立基于最小二乘支持向量机的三段氢回收膜分离过程模型;最后,基于Matlab2010a软件平台和现场数据编程建模,对炼厂气氢回收过程中的关键性能参数进行在线预测分析.仿真结果表明,模型正确合理、预测速度快,其预测值和实际测量值基本吻合,误差小,可以很好地反映出三段膜膜组件良好的分离性能,对气体膜分离过程中的参数在线检测和过程实时优化控制具有一定的指导意义.
A three-stage intelligent model of a gas membrane separation process has been proposed, and applied to analyze the key performance parameters of a hydrogen recovery membrane separation process in real time. Firstly, combined grid search and cross validation with Bayes estimation were used to obtain the optimal value of two important parameters (i. e. , sig2 and gam ) of the least squares support vector machine~ a three-stage model of the hydrogen recovery membrane separation process based on least squares support vector machine was built. Finally,a modeling program was wrote based on Matlab2010a and field data, and the key performance parameters for the hydrogen recovery membrane separation process were predicted and analyzed on-line. The simulation results show that the model is reasonable, its convergence speed is very fast, and the predicted results of the model are in good agreement with the measured values with reasonable errors. The model well reflects the good separation performance of the membrane module of the three-stage membrane process. This study has a great significance for future research on the on-line detection of important performance parameters and their optimal control in gas membrane separation processes.