The non-stationary signals are the signals with statistical properties that change with time. The spectral properties of the non-stationary signals can be analyzed by estimating the instantaneous frequency (IF). The separation of the individual mono-components from the multi-component signal is essential for the IF estimation. In this paper, we propose a methodology for the separation of the mono-components from a multi-component non-stationary signal based on a dynamic Q-value-based wavelet transform (DQVWT) method. The windowing of the time domain signal with a moving Gaussian function and the separation of components using an array of tunable Q wavelet transform (TQWT) blocks results in the mono-component separation followed by the IF computation. The proposed methodology is applied to signals that consist of linearly frequency modulated (LFM) mono-components and non-LFM (NLFM) mono-components. The IF estimation by the proposed method is compared with the existing TQWT-based filter bank (TQWT-FB) method and the Fourier Bessel/time order method. The proposed method has been applied to estimate the IF of the fundamental frequency component of the speech signal. The performance is analyzed in terms of mean square error (MSE) and the proposed method has shown better performance than the other compared methods.