In this paper, a modeling method of high dimensional piecewise affine models is proposed. Because the model is expressed by radial basis function networks, whose RBFs are located at the grid points of the orthogonal coordinate in the input space, the shape of the piecewise affine model becomes easily understood. The more the number of RBFs is, the higher the approximation capability of the model is. The algorithm to increase the number of RBFs was developed in focusing on the distribution of estimation errors. By approximating nonlinear processes with piecewise affine models, control theories using mixed logical dynamical systems can be applied.