Abstract: Due to the fact that the differential algebraic equations (DAEs) of considered chemical processes are often difficult to build, a new data-based modeling approach is proposed using ARX-neural network under partial least squares framework (ARX-NNPLS), where less specific knowledge of the process is required but the input and output data. To represent the dynamic and nonlinear behavior of the process, the ARX combined neural network structure is used in the PLS inner model between the input and output latent variables. In the proposed ARX-NNPLS model based dynamic optimization strategy, neither parameterization nor iterative DAEs solving is needed as the ARX-NNPLS model gives a proper representation for the dynamic behavior of the considered process, and the computing time is greatly reduced comparing to the conventional control vector parameterization method. To demonstrate the ARX-NNPLS model based optimization strategy, the polyethylene grade transition problem in gas phase fluidized-bed reactor is taken into account. The optimization results show that the final optimal trajectory of quality index determined by the new approach grows quicker to the target values and the computing time cost is much less.