Dtnamic State Estimation Of Synchronous Machines Based On Robust CKF Under Complex Noise Conditions

The phasor measurement unit(PMU)can directly be used to measure the phasor information in the electromechanical transient process,which provides a new development direction for the detection and control of power system.A generator is an important part of the power system.Obtaining the operation information of the generator in the electromechanical transient process is undoubtedly of great significance for power system security.However,due to factors such as natural environment and production environment,the measurement obtained by PMU inevitably has random noise and have bad data indeed.These factors can cause the dynamic state estimation result of generator to deviate significantly from the true value.As for the dynamic state estimation of the generator based on PMU measurement information,this paper systematically studies the dynamic state estimation method in the electromechanical transient process.The results are as follows:(1)Using PMU to directly measure the generator output voltage phasor,complete decoupling of generator and its external network,construct the generator dynamic state estimation model,calculate the system noise error,complete dynamic state estimation based on robust cubature Kalman filter and cubature Kalman filter and unscented Kalman filter for synchronous machines under Gaussian white noise in the MATLAB environment.The simulation results show that the filtering accuracy of cubature Kalman filter and robust cubature Kalman filter is higher than the unscented Kalman filter.(2)For the problem when PMU measures noise does not obey Gaussian white noise distribution and there are bad data,a robust cubature Kalman filter combined with robust M estimation theory is proposed based on cubature Kalman filter.The robust M estimation theory can detect the existence of bad data and eliminate its influence.The problem cubature Kalman filter lack adaptive adjustment capability when the statistical characteristics of the measurement noise deviate from the prior statistical characteristics is solved.In the MATLAB environment,the dynamic state estimation based on the robust cubature Kalman filter and the cubature Kalman filter for synchronous machine completed when the PMU measurement noise does not obey the Gaussian white noise distribution and there are bad data in measurement.The simulation results show that the robust cubature Kalman filter has better performance when dealing with continuous or single-point outliers.