Research On The Detection Methods Of Wind Farm Grid Voltage Faults

Currently, the requirements of the wind generators operating on the grid are becoming increasingly strict. One of the important requirements is that the wind generators must have the ability of low voltage ride through. The prerequisite of realizing low voltage ride through is that grid voltage positive and negative sequence's amplitude, phase, and frequency offset must be detected accurately, therefore rapid and accurate detection of the above information has important significance.When the voltage imbalance happens, the negative sequence voltage appears. It can cause wind turbine oscillation and overcurrent, so the grid voltage positive and negative sequence must be detected rapidly in order to adjust the control strategy. The existing detection methods have slow detection speed and poor anti-harmonics. Aimed at this problem, a new method based on complex least squares is presented. This method which can detect the grid voltage positive and negative sequence's amplitude and phase accurately adds covariance resetting technology to the classical least squares. The simulation results show that the grid voltage's amplitude and phase information can be detected in half a grid period even when the grid voltage contains high order harmonics.Grid voltage frequency is very important because most of control strategies need to use it. However, the current detection methods can not meet voltage imbalance and harmonics which are the typical grid faults, so the paper introduces a new phase locked loop structure based on adaptive linear optimal filter that can meet the typical grid faults. Simulation results show that compared with other synchronization methods, the new one is more generic and robust.The above two methods can only detect a part of grid fault information, and they can't take into account of all the typical faults, including non-sinusoidal, asymmetric, phase jump, amplitude jump, frequency offset and random noise. To solve this problem, a new rapid detection method based on extended complex kalman filter is proposed. This method can detect grid voltage positive and negative sequence's amplitude, phase, and frequency offset accurately and rapidly when the voltage contains high order harmonics and random noise.The C language is used to program the three methods on TMS320F2812, and various grid faults which the experiments need are generated by programmable AC source. Through these experiments, the generality of the above three methods is tested. The experimental results verify the correctness and validity of the methods.