Detection And Identification Of Transient Power Quality

With the development of the power system, a great deal of nonlinear, impact and load is taken to the grid, transient power quality problems are more and more serious, and it has been common concerned both by the power sectors and users. The traditional steady power quality analysis methods are not suiTablele for analyzing transient disturbance signals. As a result, recent power quality monitoring device can monitor only three transient disturbance signals, as voltage sag, voltage swell, and voltage interruption, powerless for the other disturbances, just as impulse transient, oscillatory transient, and voltage notching. Therefore, developing a new analysis method which can detect and identify the transient power quality disturbances rapidly and accurately is significant. This thesis designs a transient power quality on-line detection and identification system based on the virtual instrument, used for on-line detection and identification of the transient disturbance signals. Furthermore, a method for off-line classification and identification of the disturbance signals has been presented. The detailed contents of the thesis are arranged as follows:Firstly, the transient power quality on-line detection and identification platform is designed based on the virtual instrument, including the hardware system design, and data acquisition module, signal pre-processing module in the software system, which is the foundation for further research.Afterwards, a method for on-line detection and identification of recent six transient disturbance signals has been proposed based on the running discrete wavelet transform algorithm, the root mean square (RMS) algorithm, and the fuzzy logic. The transient disturbance signals are described by five index, including disturbance beginning time and ending time, disturbance duration, disturbance intensity and disturbance type. In the end, various transient disturbance signals generated by three-phase programmable AC power source are used for testing the system. The results demonstrate the good real-time performance and reliability of the system.Finally, a method based on sub-band feature extraction and Support Vector Machine with Binary Tree Architecture (SVM-BTA) is presented for power quality disturbances multi-classification. First, multiple feature vectors of the disturbance signals are extracted with frequency division method, which can not only reflect local feature of disturbance signals better, and also reduce the data size. And then, the SVM-BTA classifier is designed, which can classify steady and transient power quality disturbances rapidly and accurately. At last, simulation and experiment data under different signal to noise ratio (SNR) are used for testing the classifier, compared with other two methods at the same time. The results show that the method proposed has the merits of high identification accuracy, rapid classification speed and good anti-noise performance. Furthermore, classification of complex power quality disturbances which has not been resolved is discussed in the thesis.