*Kyoto University
In this paper, we propose the normalized parameter sensitivity plot as a new tool for analyzing the difficulty of parameter estimation in modeling. The plot visualizes the contribution of the identification inputs to the parameter estimation accuracy and the relative relationship between the estimation accuracy of multiple parameters in the frequency domain. This relative relationship sets a limit on practical estimation accuracy, and the plot is useful for evaluating this limit. As an application example, a model of a rechargeable battery for an electric vehicle is analyzed, and it is shown that the proposed approach can obtain information useful in the selection of model structure and planning of identification experiments.