The Study On Detection And Location Methods Of Power Quality Disturbances

With the rapid development of electric power industry, widespread use of new electronic devices and nonlinear loads, power quality problems have become more and more serious. To reduce the impact of power quality disturbances and improve the power quality level, the paper takes power quality disturbances as the research object to achieve detection, location and classification of power quality disturbances with the help of mathematical morphology, discrete orthogonal S transfom, GK clustering algorithm.Firstly, do research on the pretreatment method of power quality. The simulation models are established by Matlab to acquire time-domain waveform of transient power quality disturbances. The signals collected by the information center of power grid are often mixed with a amount of noise, however, single structure elememt has good filter characteristics only for certain type of noise. In order to reduce noise interference on subsequent detection, the paper applies morphological filter with multiple structure elements which obtained by single structure element through parallel combination for noises removing.Secondly, a method based on discrete orthogonal S transform is proposed to achieve the detection and location of transient power quality. The method creats a series of orthogonal basis functions by means of linear combination of original fourier basis functions in band limited subspaces with frequency variable, time variable and width of the frequency. As a result, the method has no redundant points at all and achieves the utmost resolution of time-frequency. Using modular matrix average of amplitude quadratic sum of the linearly independent time-frequency from the inner products between a time series and the basis functions, transient power quality disturbances and voltage sag, transient pulse with harmonics are detected in location of starting time and deadline, as well as the frequency components. Simulation results show that the calculation speed of proposed method can be faster than S transform with high accuracy, which leads to better real-time performance and harmonic immunity.Finally, GK clustering algorithm is applied to classify power quality disturbances. Based on the disturbances parameters initialization of fuzzy partition matrixariance matrix, similarity measurement function which is made of clustering covariance matrix is calculated to obtain the similarity among various disturbances, Then, the clustering center vector matrix is iterated and adjusted, as well as the membership matrix. When the object function composed of similarity measurement function and membership matrix reaches the minimum, iteration finishes and classification completes. Extract S transform feature parameters of twelve power quality disturbances as the input of GK clustering algorithm, set the appropriate numbers of clustering to vadicate, the clustering partitioning results show that the proposed method has high accuracy and less sensitive to noises.