Application Of Statistical Signal Processing Techniques In The Detection And Tracking Of Power Quality Disturbance

With the development of national economy the enhancement of people's living standard, an increasing number of electrical and electronic devices are put into use in control and automation industry. As a result, customer's demand for power qualities, especially such issues as voltage quality and harmonics, is leveling up. Power quality issues have drawn a wide attention among utilities and customers in numerous countries. The level of power quality has a direct impact on the industrial and commercial systems, even the well-being of the whole national economy. Therefore, the enhancement of power quality and mitigation of harmonics have become the key technique problems in the transmission and distribution system, which is extremely important for a developing country like China from both realistic and strategic perspectives.Among the existing power quality analysis algorithms, Fourier Transformation based techniques are suitable only for deterministic and stationary signals (like harmonics) while gives an insufficient description of non-stationary signals. This is because that Fourier Transformation is integrated along the whole time domain. Thus the time varying information of non-stationary signals is lost in the transformation. Although the multi-resolution analysis of wavelets is able to detect and locate the short-duration power system disturbances, it has difficulties in determining the exact type of disturbance. Therefore, for the analysis and tracking of non-stationary power quality disturbances, this thesis introduced statistical signal processing methodologies into the treatment of abovementioned problems. Statistical signal processing techniques are intrinsically suitable for dealing with non-stationary signals and possess a good random noise rejection property in the same time.This thesis develops a Sigma Point Kalman filtering and a systolic array based adaptive QR-decomposition Least Mean Square algorithm on the basis of Kalman filtering and adaptive filtering theory for the online tracking the detection of two of the most common phenomena– voltage flicker and harmonics in power quality disturbances. Simulations conducted in MATLAB and LabVIEW have revealed the superiority of proposed algorithms.