The goal of this paper is to extend recently developed work [Chen, J.; Sheui, R.-G. Ind. Eng. Chem. Res. 2002, 41 (9), 2226] to design the optimal trajectory for a batch process on the basis of an intelligent data-driven experimental design scheme. This method integrates orthonormal function approximation with soft-computing techniques. The continuous batch trajectories of process measurements represented by a set of orthonormal functions are mapped onto a finite number of coefficients in the function space. These coefficients capture the trajectory behavior of the process measurements. The optimal trajectory can be obtained as long as the locations of the coefficients are properly adjusted. An adjustment algorithm combining a neural network with a genetic algorithm is developed to search for optimal coefficients. The neural network is used to identify the relationship between the coefficients and the objective function and to predict the quality response. Once a suitable neural-network model is obtained, the genetic algorithm is used to explore the optimal conditions. There are two major advantages of the proposed method. First, it can deal with the dynamic set-point problem that cannot be handled well using traditional experimental design. Second, by a sequence of design experiments, the objective of the process performance can be gradually improved for the optimization of dynamic batch processes. Potential applications of the proposed method are shown through three detailed simulation studies.