This study presents a new application approach to evaluate the parameters of fragment coming down curves and acquire the corresponding fitting curves. The approach is based on a system identification method known as the impulse response sequence algorithm to fit the fragment coming down curves. The impulse response sequence algorithm generates a state space description of the fragment coming down curves directly from its available input-output data. Comparing to the static mathematical curve model, the dynamic state space model gives more exact fit for all types of fragment coming down curves. This paper demonstrates the feasibility of decomposing the fragment coming down curves into two groups of signal components such as smoothed components and oscillation components. According to the characteristics of each component, the unwanted components are removed to construct the mean-curve with very little or no distortion corresponding to the original signal. Results obtained from the mathematical calculations and the measurements of real fragment coming down curves show its applicability.