KPCA-KPLS方法在pH中和过程建模中的应用

针对化工过程中的具有严重非线性、不确定性、时变性的复杂pH中和过程系统建模问题,提出一种基于核主元分析(KPCA)与核偏最小二乘(KPLS)相结合的建模方法.在高维特征空间内,该方法通过KPCA有效地提取输入数据的非线性主元,利用KPLS方法将输入变量投影在潜在变量上,再用输入与输出变量之间的协方差信息提取潜在特征建立pH中和过程模型.为验证其有效性,将KPCA-KPLS方法应用到弱酸强碱中和过程、强酸强碱中和过程实例中,并与核偏最小二乘、核主元分析_支持向量机(KPCA-SVM)、核极限学习机(KELM)、极限学习机(ELM)、最小二乘支持向量机(LSSVM)、SVM等方法进行比较.实验结果表明:KPCA-KPLS方法具有很高的动态建模精度.

Research on Modeling Method for pH Neutralization Process Using KPCA-KPLS

In terms of pH neutralization process modeling problem with serious nonlinearity, uncertainty and time variability in chemical process, a modeling method via kernel principal component analysis(KPCA)and kernel partial least squares(KPLS)is proposed. This method via KPCA extracts nonlinear principal element of input data in the high-dimensional feature space effectively, and then use the KPLS method project input variable to the latent variables, and extract potential characteristics of the covariance information between input and output variables to establish the pH process model. In order to validity the effectiveness of the proposed kernel-based modeling method, KPCA-KPLS method is applied to the pH neutralization processes instances which including weak acid-strong base process and strong acid-strong base process, and is compared with kernel principal component analysis-support vector machine(KPCA-SVM), kernel extreme learning machine(KELM), extreme learning machine(ELM), least squares support vector machine(LSSVM), support vector machine(SVM)methods etc. The experimental results show that the KPCA-KPLS method has a high dynamic modeling accuracy.