A low-order Kalman filter model based on modulation is presented to detect real-time disturbances of short-term power quality.αβ transform result of three-phase voltage is decomposed into the fundamental signal and modulation signal.Then two characteristics of modulation signal replacing each harmonic are taken as new state variables.It reduces model orders and improves speed of transient response.Considering the characteristic of filtering error which occur mutation when there is a transient signal,the method of filtering error ratio threshold is introduced to judge initiative and terminative time of transient.Then system noise matrix Q is adapted automatically in order to reduce model error and restrain radiation.Simulation results of Matlab show that the new method has higher accuracy of steady-state and faster speed of transient response than general Kalman filters and low-pass filters.

Aiming at the problem of voltage sag real-time detection,the time-varying Kalman filter with compensation component is presented.The model of Kalman filter with compensation component is constructed by decomposing sag voltage into steady component and compensation component.The compensation component which is to be controlled into effect by the compensation instruction of dynamic voltage restorer(DVR) is removed during steady period of voltage,and functioned when voltage sags occurring.The root mean square(RMS) and phase shift of voltage sags are confirmed by αβ/dq transformation.The residual error is adopted to detect the voltage step change and lock-delay method is proposed to overcome bouncing problem of compensation component functioning.Simulation results show that the proposed method is fast than Kalman filter and virtual three-phase dq transformation,meets the real-time request of DVR.

Voltage sag is one of important power quality problems,which can be mitigated efficiently by dynamic voltage restorer(DVR)which requires fast detection method.Dq transformation is applied widely in dynamic voltage restorer(DVR)as it is simple and easy to be implemented,but the low pass filter slows down its dynamic response speed.Aiming at this problem,the dynamic response feature of Dq transformation is analyzed.A modified dq transformation corrected by Kalman filter is presented.The d component is decomposing into decaying component and constant component.Kalman filter is used to estimate constant component as the corrected d component of dynamic process.It's suggested to start Kalman filter when d component variation exceeds the threshold and quit Kalman filter when the decaying component is less than the threshold.The proposed method is used to detect several types of voltage sag in the software Matlab6.5.Results show that the method is faster than dq transformation,and meets the real-time request of DVR.

In order to accurately detect real-time harmonic current signals in real power system,a random harmonic current detection method is proposed based on αβ transform and adaptive Kalman filter.The three-phase current signal is normalized and the three-phase asymmetry current with harmonic and noise signals is divided into two mutually perpendicular αβ component through αβ transform,in order to establish the state equation of Kalman filter,which has plus and minus sequence components of fundamental and harmonic current.It can detect any appointed harmonic by changing the parameters in the state equation.Kalman filtering algorithm is used to quickly and accurately detect the fundamental component and random harmonic component of unbalanced three-phase current.System noise matrix Q is auto-adapted in order to reduce model error and restrain the radiation of hormonics.Results of MATLAB simulation prove the effectiveness of the method.