Optimization of PCA-RBF Based Soft-Sensing Model
Principal Component Analysis (PCA) can effectively extract the characteristic information of data and eliminate co-linearity in variables. Radial Basis Function Neural-Network has the properties of fast learning and Global Convergence. This article puts forward a soft-sensing modeling method combining PCA and RBFNN, which selects the RBF centers by means of Data Clustering based on Quanta Particle Swarm Optimization. The transform function in Hidden Layer Unit is Gauss Transform. The weight from hidden to output layer is also optimized by QSPO. The obtained soft-sensing is verified in measuring the oxygen content in CCR heating furnace.
soft-sensing PCA Neural-Network QSPO RBF
Zhang Hong Xu Wenbo Liu Fei
School of Mechanical Engineering, Southern Yangtze University Wuxi 214122,China School of information Engineering, Southern Yangtze University Wuxi 214122,China Institute of Automation, Southern Yangtze University Wuxi 214122, China
国际会议
杭州
英文
1376-1381
2006-10-12（万方平台首次上网日期，不代表论文的发表时间）