Wavelet based forecasting methodologies have found increasing use in different areas of engineering. The ability of the wavelet transforms to deal with non-stationary characteristics of data has enabled researchers to develop efficient hybrid models for electrical load forecasting. Wavelet pre-processing is generally found to increase the accuracy of the forecast engines. However, selection of mother wavelet, number of levels of decomposition and border distortion effects are issues of concern in wavelet filtering of signals. This paper presents a study of these issues in filtering a load signal. The filtering quality is analyzed in terms of the Root Mean Square Error (RMSE).

wavelets, filtering, electrical load, threshold, denoising.