基于KPLS的丙烯爆炸极限非线性预测研究

准确预测丙烯爆炸极限是保证丙烯酸安全生产的有效手段.丙烯爆炸混合气各成分体积分数之间呈现非线性关系,采用KPLS算法对丙烯爆炸极限进行预测,并利用丙烯酸实际数据进行仿真.仿真结果表明,该方法能够有效预测丙烯爆炸极限.与PLS算法相比,预测结果具有显著的跟踪能力,并确定了2.56%~9.25%的丙烯爆炸区域,解决了生产过程中丙烯爆炸问题.

Nonlinear Prediction Research on Explosion Limits of Propylene Based on Kernel Partial Least Squares

Prediction on the propylene explosion limit is an effective way to ensure the acrylic acid safety in production. The propylene and mixture volume concentrations have nonlinear relationship between them. Therefore， kernel partial least squares method is used to predict the limits of propylene explosion and the simulation is carried out by the actual data. Simulation results showed that the algorithm was feasible and effective. Compared with those of partial least squares， the prediction results of the proposed method have more remarkable tracking capability. Meanwhile， the propylene explosion region which is from 2.56% to 9.25% is determined. So the problem of propylene explosion can be solved.