Research On New Methods Of Power System Relay Protection And Failure Detection

Power system relay protection are important guarantee for the safe operation of power network. With the wide application of computer technology to power system, the methods research in faults detection and relay protection has become more and more important. Consequently,this dissertation fully studies the differential protection of electric transformer and detection of faults in non-direct earth neutral system, then the corresponding improved methods are presented according to the problems existing in present methods, some new methods are puted forth to accomplish automatic protection and fault detection.Against the spectrum leakage of Short Time Fourier Transform(STFT), a new algorithm structure consisting of parallel filter and Bartlett window is presented in the dissertation. The dissertation theoretically analyzes the filter implementation of STFT, and the response features of window function STFT are modulated. According to the outcome of the study, the source of the error is spectral leakage, an inevitable result of the problem that the signal frequency migrate central frequency of the filter. Rapid decaying of the sidelobe of window function can help reduce such long-range leakage, but rectangle window doesn't belong to this type of window function. 50Hz and 100Hz frequency coexist in the inrush current of the transformer, which result in serious leakage error of main lobe when the other window functions process this type of signals. Separate the main components of frequency and adopt STFT of Bartlett, it can overcome the defect of sidelobe of rectangle window, This new algorithm structure can obviously restrain spectral leakage and is suitable for the differential protection of transformer.Identify inrush current is the key problem in transformer differential protection. An improved algorithm based on holocycle fourier algorithm and a new method to identify inrush current based on wavelet energy are presented in the dissertation. Studying the features of modern transformer core materials, the dissertation constructs a transformer model based on hysteresis loop, analyzes the relation between the parameters of inrush as well as of the transformer's iron core and the closing initial phase angle, and comes to a conclusion that a dead zone of harmonic restraint exists within the closing initial phase angle in a specific scope; And the improvement in the core materials reduces inrush peak value, but the principle of harmonic restraint can still work; Inrush current contains a large amount of decaying DC component. To solve the contradiction between decaying DC component and STFT, an improved holocycle fourier algorithm is put forth and applied in transformer differential protection. The dissertation studies the wavelet theory from the aspect of time-frequency analysis, particularly analyzes the multisolution analysis and Mallat wavelet decomposition method, and comes up with a new method, based on db6 wavelet decomposition, this new method adopts the energy ratio of 4 and 6 scale as a criterion to identify inrush current and internal transformer faults.A major problem in electric power system is how to detect faults in small current's ground system, method of detecting fault grounded point besed on electromagnetic field, and another method of selecting the fault line are put forth in the dissertation. The dissertation analyzes the features of the current and voltage in case of single-phase-to-ground faults from aspect of steady state, then analyzes the distribution rules of electromagnetic field generated by the current and voltage of the perfect branch or fault branch. After that, a new method is put forward to detect new ground spots by using fifth harmonic and zero sequence electric field and magnetic field. Its validity has been testified by the simulative experiment. Compared with the commonly used method of fault line selection and fault distance measure, This method connects the features of fifth harmonic under static state with the cross electromagnetic field, and achieve outdoor detection of ground faults point. The dissertation simulates the transient wave patterns of the current and votage in grounding and collects the feature value with db13 wavelet, then puts forth a wavelet energy analysis method to handle Single-phase-to-ground fault. With a strong robustness, this method uses the energy in 6 scale of the load current directly as its criterion to verdict grounded phase and grounded branch.Apply the new methods into practice, a differential protector of power transformer is realized in this dissertation. As a fairly complicated element protection device, transformer differential protection device is one of the essential carriers to detect the hardware structure of protection devices and the rationality of its principle. The dissertation introduces the whole process of researching and developing the device and combines its hardware principle, its software design, the protection method and the algorithm. The device has been examined by National Relay Quality Supervision Centre and Stable operation on transformer substation, hydro power station and pumping station.