Application of multiple moving approximation with polynomials in curve smoothing

The paper has characterized the method of multiple moving least-square approximation with polynomials, known as the Savitzky-Golay filter. This method enables smoothing the measurement series, decomposition and separation of disturbances, generation of derivatives as well as approximate integration of measurement series. The smoothing properties of the method as well as the possibilities of separation and decomposition of disturbances are shown on the examples of treatment of the selected indicator graph. The attention was paid to the need for additional application of special filters in case of abnormal impulse deviations. The examples show that in case of the analyzed curve, the results of smoothing by means of several wave filters from Wavelet Explorer package, are worse than those received with the methods of moving approximation. On the derivatives of smoothened curve by means of wavelet filters, there are significant oscillations, what is typical of the whole-interval approximation with spline functions. One should highlight the high simplicity of algorithms of multiple moving approximation and associated with this, high speed of operation, what particularly predisposes this method of data processing in an on-line mode. The authors presented their own proposals of programs for measurement data processing with the method of multiple moving approximation with polynomials, with the possibility of selection of an approximation polynomial to the fifth degree. The programs were developed in the Excel and Delphi environment.