Abstarct: Grid impedance estimation is useful in many power system applications such as power quality analysis of smart grids and grid security against natural disasters such as thunderbolt. In this thesis, the impedance of the grid in frequency range of 2-150 kHz baxsed on signal processing methods and signal injection is estimated. In grid impedance estimation baxsed on signal injection, for impedance estimation at frequency f, the voltage or current signal must have a component at this frequency. In addition, in noisy conditions, the accuracy of impedance estimation is directly dependent on energy of injected signal. In this thesis, the grid impedance estimation in a wide frequency range is done using rectangular injection pulse that it has a wide frequency band. Energy of rectangular injection pulse is dependent on two parameters, maximum peak amplitude and pulse width. Increasing the pulse width leads to increasing the energy of injection signal. However, it reduces the bandwidth of the pulse and therefore estimation cannot be done accurately in the frequencies with low energy. Although increasing the maximum peak amplitude of the rectangular injection pulse leads to increasing the accuracy of impedance estimation, but it reduces the grid power quality. In addition, the accuracy of impedance estimation can be increased using denoising the measured voltage and current signals instead of increasing the energy of injection signal. In this thesis, the stationary wavelet denoising algorithm is employed for denoising of measured signals. The bandwidth and pulse width of a rectangular injection pulse have inversely relation. In addition, each pulse is unable to impedance estimation in some frequencies. Therefore, in this thesis, a method for selecting several short-term injection signals in frequency range of 2-150 kHz is proposed. When the measurement duration is low, the whole transient state of the grid is not considered; hence the impedance estimation is not accurate. In addition, when the measurement duration is high, the effect of noise is increased, hence the estimation may not be correct. In this thesis, a method for determining the suitable measurement duration baxsed on time-frequency representation of impedance is proposed. In this thesis, a comprehensive method for impedance estimation using injection of several short term signals and denoising of measured signals and determining the suitable measurement duration is proposed. The proposed method is applied on several girds and the simulation results show the ability of the proposed method. Our researches show that, there is no method in the literature review in frequency range 2-150 kHz for result comparison. Finally for validation of the proposed method, the idea of the proposed method is tested on a real grid.