Implementation of wavelet transformation in analyzing and reconstructing a noisy sinusoidal signal

Noise is the main issue that commonly occurs in taking and processing the signal. The noise may interfere with the first vibration mode frequency based on the results of Fast Fourier Transform. Previous study carried out using signals with sinusoidal functions has been able to minimize the noise in a signal. However, there is limited information about the time or location the frequency value of a signal takes place. For this reason, wavelet transforms are reapplied in reconstructing signals to provide information on the frequency and time. In this study, signal data with sinusoidal functions is used with a specified frequency of 0.25 Hz. The signal data is subjected to noise with a range of standard deviation 0.1 - 1.0. Wavelet analysis used is Haar wavelet, while the signal filtering process used high-pass and low-pass filter. The filtering lengths used are 8 (23) and 32 (25), so it will produce 8*8 and 32*32 Haar wavelet analysis matrics. The results show that the specified frequency value can be determined using signal analysis with Haar wavelet analysis and signal reconstruction. The higher the scale paramater value (j) is used in signal analysis and filtering, the higher range of noise values can be reduced. The results also indicated that the signal frequency can be obtained when there is a similar pattern between the reconstruction and original signal.