Abstract：Considering the difficulty of keeping the accuracy of on-line product concentration measurement in practical process under scaling or scabbing conditions, Kernel partial least square (KPLS) algorithm is proposed to build soft-sensor model for the product concentration based on the analysis of nonlinear characteristic and factors affecting product concentration of the vertical tube falling-film evaporator. KPLS-based model realizes on-line concentration measurement by effectively extracting nonlinear eigenvector of concentration and relevant detection variables. Simulation results show that the KPLS-based model achieved better performance in accuracy than that in the conventional linear partial least squares-based model, and relative error for production concentration measurement can meet the practical requirements.
［1］姚骏, 张权, 李斌. 基于超声衰减的纸浆浓度测量技术的研究［J］.计量学报, 2008,29(1): 87-91.
［2］景步云, 谷波, 黎远光. 板式蒸发器仿真计算模型［J］. 系统仿真学报, 2003, 15(10): 1481-1483.
［3］Khademi M H, Rahimpour M R, Jahanmiri A. Simulation and optimization of a six-effect evaporator in a desalination process［J］.Chemical Engineering and Processing: Process Intensification, 2009,48(1): 339-347.
［4］张光新, 周泽魁. 烧碱蒸发过程出料浓度软测量技术研究［J］. 仪器仪表学报, 2004,25(z3): 675-676.
［5］Visakha K Nanayakkara, Yasuyuki Ikegami, Haruo Uehara. Evolutionary design of dynamic neural networks for evaporator control［J］. International Journal of Refigeration, 2002,25(6): 813-826.
［6］贾润达, 毛志忠, 常玉清. 湿法冶金萃取组分含量混合建模方法的研究［J］. 仪器仪表学报,2009,30(2): 267 -271.
［7］张飞, 童朝南, 李江昀. 基于KPLS和TD的热连轧精轧末架微调AGC［J］. 控制理论与应用, 2008,25(1): 87-90.
［8］Woo S H, Jeon C O,Yun Y S,et al. On-line estimation of key process variables based on kernel partial least squares in an industrial cokes wastewater treatment plant ［J］.Journal of Hazardous Materials,2009, 161(1): 538-544.
［9］Christian Sonntag,Sven Lohmann, Anna Vlker,et al. Analyzing safety properties of hybrid processing systems: A case study on an industrial evaporator ［J］. Journal of Process Control, 2008,18(9):885-895.
2011, Vol. 32 
