Theory And Method Of Extracting Information And Diagnosing Fault For Insulator With Map And Bispectrum

About the quantities of electrostatics and electricity of insulators, two aspects are in-depth researched and discussed:the calculation and improvement of the error of the discrete value of the quantities of electrostatic field; finding the electrical quantities sensitive either to fault or to contamination and diagnosing insulators.First, the calculation and improvement of the error of numerical solution of the quantities of electrostatic field are researched and discussed. Calculating the quantities of the electrostatic field of insulator often uses software ANSYS that has two problems:the error of numerical solution, and only bounded field. To the error of ANSYS solution of electrostatic field of insulator, there are few papers to research because it is difficult to find the analytical solution of the electrostatic field of a actual insulator. Thus a single ideal pin insulator is constructed, and the electric potential and the electric field intensity of any point in insulator are calculated with software ANSYS and analytical method. Two conclusions are reached:the sharper the negative electrode is, the bigger the error of the electrical potential and electric field of any point in insulator are; the more the point in insulator close with negative sharp electrode, the bigger the error of the electric potential of the point is, and the more the point in insulator close with positive electrode or negative electrode, the bigger the error of the electric field of the point is. To calculate the unbounded electrostatic field of insulator strings, some papers present methods so as to apply ANSYS to the unbounded electrostatic field of insulator strings. To decrease error more, this thesis presents that combining map and ANSYS solve unbounded electrostatic field with high accuracy. The method uses a complex exponential function to establish a map between unbounded and bounded electrostatic fields of two-dimension. Then ANSYS is used in bounded fields. As the boundary value problems are same, the electric potential of each point in bounded fields equals the electric potential of each mapping point in unbounded fields. An example indicates that the accuracy obviously increase when solving unbounded field with map and ANSYS.Secondly, finding the electrical quantities sensitive either to fault or to contamination and diagnosing insulators are researched and discussed. To the higher incorrect and missing judgement of instruments using leakage current (LC) to diagnose insulator and monitor contamination, having consulted many correlative papers the thesis presents that that the essential reason for higher incorrect and missing judgement is that the instruments have not found the eigenvalue only related to the monitored state and not related to other states. To find the sensitive eigenvalues of both faulty insulator and insulator contamination, based on experimental 198+17+78+14 LC time series(198+17 LC time series belong to perfect contaminated insulators; 78+14 LC time series belong to faulty contaminated insulators) of light discharge of single insulator which is at the state of various dirtiness, humidity, and temperature, and is applied non-sinusoidal voltages at different virtual values, this thesis reachs the following conclusions:Fundamental resistance is the sensitive eigenvalue for faulty insulator. By 198+78 LC time series, the scopes of the fundamental resistance of perfect and faulty insulators are 222.39-2841.93MΩand 0.23～4.97MΩ, respectively. The bispectrum amplitude eigenvalueβof LC time series is sensitive eigenvalue for faulty insulators. It is validated in theory that the bispectrum of the LC time series is the eigenvalue sensitive to insulators, but not sensitive to insulator contamination, humidity and temperature. It is validated in datum that by 198+17+78+14 LC time series, the scopes ofβvalue of perfect and faulty insulators are 1.2290×10-3～1.5026×10-1, and 7.5520×10-5～1.2660×10-3. The wave crest factor of LC time series is sensitive eigenvalue for light contaminations. The waveform factor of 1562.5-25000Hz high frequency current component of LC time series is sensitive eigenvalue for heavy contaminations. These conclusions result from the contrast of many quantities of each LC time series, and have practical significance. A three-dimensional vector, which consists of theβof bispectrum, Shannon entropy, and wave crest factor is the sensitive eigenvector for faulty insulator. The eigenvector can divide all the insulators into two regions not only non-overlapping but far apart regions:the region of perfect contaminated insulators; the region of faulty contaminated insulators. Using above a three-dimensional vector as input vector, any of three classifiers, Euclide distance judgment, grey relational analysis of B-mode, and BP artificial neural network, can exactly distinguish between perfect and faulty insulators at various dirtiness, humidity, temperature and the virtual values of the voltage involving harmonics. This conclusion has both academic and practical significances.