Aiming at typical pH neutralization process

a kernel learning dynamic model based on kernel recursive least squares（ KRLS） is proposed. The KRLS method uses sparse algorithm based on approximate linear dependence technology

which reduces computational complexity and storage capacity. It can be applied to the training of large scale data sets and modeling of dynamic time-varying process. The kernel learning method is applied to two-output pH neutralization process with buffered stream. In order to validity the effectiveness of the proposed kernel-based modeling method

under the same conditions

compared with kernel partial least square（ KPLS）

kernel principal component analysis-support vector machine（ KPCA-SVM）

extreme learning machine with kernel（ KELM）

extreme learning machine（ ELM）

support vector machine（ SVM） methods

etc.Experimental results show that as an online adaptive method

KRLS method has high dynamic modeling precision

this study lay the foundation for controlling of pH neutralization process.