Determining of signal derivatives with the use of regressive differential filters in identification problems

The paper presents a methodology of regressive differentiating filters design. Such filters ensure considerable elimination of measurement and quantization noise. The designed filters well approximate the characteristic of ideal filters. The advantage of the designed regressive differential filters is the possibility of obtaining the filtered signal of the same length as before the filtration. An example of solving the parameter identification problem in case of robot with four degrees of freedom has been also presented. The identification has been performed with the use of elaborated differential filters. The applied identification method does not require differential equations solving but only determining the appropriate derivatives.