Model Reduction Using the Balanced Truncation Method and the Padé Approximation Method

Abstract

The objective of this article is the construction of a new model reduction method for the simplification of large-scale linear dynamical plants. In this method, the denominator of the lower order plant is determined by using the balanced truncation technique and the numerator is obtained by the Padé approximation method. This technique is given for the removal of existing limitations of the Padé approximation and the balanced truncation method. In this proposed method, it is ensured the stability and steady-state value of the original system are preserved in the reduced model. Thus the instability problem of the Padé approximation method and the steady-state error limitation of the balanced truncation method are circumvented. The advantages of the Padé approximation and the balanced truncation method are retained in the reduced model. For showing the accuracy and simplicity of the proposed method, standard numerical examples are simplified. This method is also compared to the recent and standard methods of the model diminution with the help of MATLAB.