An underlying experimental design algorithm for obtaining optimal batch profiles in a sequential fashion is addressed. The proposed design technique, composed of the hybrid function approximation and the Taguchi orthogonal array, is developed to determine the new design profiles in the next run. This hybrid type of function approximation allows the global and local approximations to construct the profiles in the whole design space and some local regions, respectively, for a wide range of the batch profiles. It can convert the batch profiles into a set of function coefficients. The optimal profile can be obtained if the location of the function coefficients is properly adjusted in the function space. The Taguchi approach is used to design experiments and analyze the outcomes of each experimental design before conducting the next new run. To reduce the number of experiments in each run, a search procedure and a forbidding strategy are proposed according to the effect of each coefficient. They adjust the design coefficients and freeze the undesigned coefficients. Without prior knowledge of the batch processes, the proposed method using information from the previous batches can update and modify the profiles that will be applied to the subsequent experiments. The performance of the proposed method is illustrated through two end-point optimization problems, including a nondifferential system and a fed-batch process. Comparisons with some other optimization methods are also made.