Approximation of a class of distributed parameter systems using proper orthogonal decomposition

In this paper, approximation of the spatio-temporal response of a hyperbolic distributed parameter system with the use of the proper orthogonal decomposition method is discussed. Based on a simulation data set, representing the profile of a selected process variable, the model reduction procedure is performed. The procedure consists in the projection of the original data into the subspace represented by eigenvectors of the spatial covariance matrix, corresponding to its highest eigenvalues. Influence of the approximation order on the response approximation error and on the data compression ratio is also analyzed.