AC Line Overvoltage Classification Recognition And Prediction Methods Research

Insulation breakdown is the mean cause to line trip and interruption of power supply, and most of the insulation failure is caused by overvoltage in power systems. Within the total line trip time of high voltage transmission line,40～70% persent can be attributed to lightning stroke. Different insulation failures have different mechanism, so the over-voltage control methods will be different according to the fault mechanism. So it is of great significance to make clearly identification to different overvoltage types.Along with the application of on-line monitoring device such as Rogowski coil, the transient waveform data after overvoltage happened on transmission line can be recorded and saved. But such big amount of waveform data need expert analysis and reliable analysis results before it can be directions to manual processing, which reduces the practical applicability of these on-line monitoring devices. If on-line feature extraction and classification can be done to the transient waveform data of fault line, not only the fault point can be found and achieve power supply restoration quickly, but also the over-voltage control measures can be carried based on the fault facts to improve reliability power supply. Therefore, it is meaningful to make accurate classification of over-voltage to insure the safe and reliable operation of power grid.The thesis is carried around the analysis of power line lightning electromagnetic transient, different characteristics of transient traveling wave on 500kV faulty power line were analyzed before over-voltage sorting. The power transmission line itself can be regard as a subsistent terminal of the lightning current, the transient traveling wave detected from power line and relay location differs according to different lightning stroke. The production mechanism of different lightning stroke and its form of expression on transient traveling wave is analyzed in this thesis, and proper signal processing methods are chosen to extract the feature of transient traveling wave on that basis to form the classification criterions of different over-lightning voltage. The lightning over-voltage classification methods discussed in this thesis includes the four following aspects:1. Lightning stroke and Ordinary short-circuit; 2. Back flashover and Shielding failure; 3. Direct lightning and Induced lightning; 4. Lightning failure and Lightning interference. Moreover, the internal overvoltage characteristic of 35kV feeder line is analysised, identification methods of internal overvoltage researched in this thesis include:1. Arc grounding and Single-phase grounding overvoltage; 2. Ferro-resonance and Single-phase grounding overvoltage.Finally, power line over-voltage level prediction methods based on artificial neural network is introduced. Multilayer feed forward neural network model is adopted in the over-voltage level prediction. The maximum values of power line over-voltage under different fault boundary conditions are calculated through traversal simulation to make up the total sample set for over-voltage level prediction. Direct lightning over-voltage of 500kV lines and closing over-voltage of 110kV cable hybrids lines has been done in this thesis so as to trace for the significant factors that affect power line over-voltage level.