Research On Voltage Disturbance Detection Based On Differential Entropy And Dynamic Time Warping

Electricity is closely related to human, the stand or fall of power quality directly affect the power grid operation and the human production and living, especially the influence of high precision instruments and the power electronic equipment is very serious. Power quality is mainly manifested in three aspects: voltage, frequency and phase. Fluctuations of voltage and frequency will directly affect the almost all electric equipment. Voltage phase change can affect most of the normal operation of the electric equipment. Therefore, the real-time detection and monitoring of power quality is necessary. And some corresponding compensation measures should be taken as needed.First of all, after contrastive research and analysis with several common voltage detection algorithm which mainly includes: RMS algorithm, defect voltage algorithm,the fundamental component algorithm, the fundamental component algorithm based on Fourier, neural network algorithm and dqO transformation algorithm based on reactive power, adopted dqO transformation algorithm as the basis of this thesis.Through simulation and contrastive analysis of Butterworth filter, Bessel filter,Chebyshev filter, Elliptic filter, transfer function filter and the traditional morphological filter,a new type of morphological filter, which combines the sliding mean and morphological gradient to improve the traditional morphological filter,is presented in this thesis.On the basis of dqO transformation and the morphological filter,this thesis begins to study the dynamic voltage disturbance location and classification. Through researching and analyzing the traditional disturbance positioning technology,a new improved differential entropy disturbance begin-end time positioning algorithm,which combines the new morphological filter, is presented. It not only reduces the complexity of the algorithm but also improves the precision and efficiency of positioning. On this basis, combining with dynamic time distortion (DTW) algorithm and the improvement of the traditional identification technology,a new improved disturbance type identification algorithm,which combines differential entropy disturbance begin-end time positioning and dqO transformation, is presented. It not only effectively improves the recognition precision but also reduces the workload of MCU.Finally,physical verification experiments are carried out on the several kinds of improved algorithm in this thesis on the basis of STM32 and LABVIEW. Then research and analyze the result of the experiment, so as to verify the validity and feasibility of the algorithms.