Study On Short-Circuit Fault Current Prediction And Limitation Technology For Ultra High Voltage Grid

In a modern grid system of increasing size and load density,rising short-circuit currents can make it difficult to remove faults and destabilizing the grid,or worse,damage power installations and bring the regional grid to a standstill.As the UHV grid is a large-scale,long-distance,high-power transmission project,it can cause huge losses and serious consequences in the event of a short-circuit fault,and many grid companies are still facing problems with short-circuit currents exceeding the limits and difficulties in limiting them.Therefore,it is important to understand how the short-circuit fault current changes with the short-circuit fault time in the shortest possible time after a short-circuit fault occurs,so that the system can take the most appropriate technical measures to cut off the fault line or limit the short-circuit fault current.A study of short-circuit fault detection methods and short-circuit current overzero point estimation methods.A short-circuit fault detection method based on the sudden change of load angle method is proposed,which takes advantage of the fact that the load angle also changes abruptly when a short-circuit fault occurs and the value of the load angle is close to π/ 2,to accurately determine whether a shortcircuit fault has occurred in the system.A dichotomous-Newton method based shortcircuit current crossing point estimation method is proposed,which can accurately estimate the time when the short-circuit current crosses the zero point,and the accuracy and rapidity of the prediction is related to the iteration error.Prediction of short-circuit current amplitude following a short-circuit fault on an ultra-high voltage transmission line based on the particle swarm optimization incremental extreme learning machine algorithm.The selection of the excitation function of the extreme learning machine algorithm has an impact on the prediction accuracy,of which the highest accuracy is achieved when the Sine function is used as the excitation function.The selection of the particle swarm algorithm fitness function has an impact on the accuracy of the prediction,with the highest accuracy being achieved when the grey absolute correlation is used as the fitness function of the particle swarm algorithm.The particle swarm algorithm optimizes the values of input weights and implied layer deviations,and the incremental limit learning machine model optimizes the selection of the number of nodes in the implied layer to improve prediction accuracy with less float and greater robustness on the basis of effective prediction of short-circuit current magnitude.The particle swarm optimization incremental extreme learning machine algorithm can achieve fast and accurate prediction of short-circuit current magnitude for the full fault phase angle and full fault location,and the prediction accuracy is the highest compared with other prediction methods,and the prediction results are closer to the real value of shortcircuit current magnitude.Two methods are proposed for predicting short-circuit fault current parameters based on ultra-short data acquisition windows,namely the full data window curve fitting method and the sliding shift sampling method.The full-data window curve fitting method is more accurate in fitting the short-circuit current waveform,and the value of the total number of samples N has a greater impact on the accuracy of the prediction results.The full-data window curve fitting method has a higher prediction accuracy than the improved half-wave Fourier method,which requires only 1ms of sampling data to meet the requirements of both speed and accuracy.The sliding shift sampling method can accurately predict the short-circuit current parameters to be sought,and the predicted short-circuit current curve is in better agreement with the actual waveform curve,using the time past zero as the threshold setting criterion,with different thresholds predicting different times,but both can achieve fast prediction.Compared to the two methods,the sliding shift sampling method predicts a better match between the current waveform and the actual current waveform with the same sampling data window,and predicts the short-circuit current parameter to be sought more accurately;the sliding shift sampling method takes less time to predict with the same over-zero time accuracy.The impact of high frequency disturbances on the prediction results of both short circuit fault current prediction methods is significant.The sliding shift sampling method outperforms the full-data window curve fitting method in terms of noise tolerance,but the accuracy of both methods is severely affected when the noise intensity continues to increase.The sliding shift sampling method is more tolerant when the grid frequency is shifted.In this paper,a disturbance cancellation method based on trend filtering techniques is proposed.The trend filtering technique enables multiple trend filters to be applied to the sampled data at the initial stage of prediction to achieve rapid elimination of high frequency interference and white noise interference,improving the accuracy of prediction without affecting the rapidity of prediction.A topology model for a fault current limiter based on split reactors is proposed.A current limiter based on a double split reactor operating in parallel is designed to address the problems of high open circuit voltage requirements on the circuit breaker and the tendency to cause re-ignition and re-ignition of the circuit breaker.With a significant reduction of the short-circuit current amplitude,the open-circuit voltage of the circuit breaker operating in series with a bifurcated reactor is half of that of the circuit breaker operating in parallel with a bifurcated reactor,solving the technical problem caused by the high voltage on the circuit breaker.Through the analysis of the structural characteristics of the double split reactor series operation current limiter,the reactor coil structure is given,the calculation formula of the self and mutual inductance of the push wire turns is given,so that the calculation formula of the self and mutual inductance of the split reactor coil is derived,and finally the electromagnetic design process of the reactor is given.