Research On Low Frequency Oscillation Mode Identification Based On Wide-area Multi-channel Measurement System

With the rapid development of power industry,low frequency oscillation has become one of the hot topics in power system stability researchs.Accurate and rapid acquisition of oscillation mode characteristics is an important prerequisite for the analysis of oscillation signals.At present,the wide-area measurement signal analysis method can directly extract the characteristic parameters of the oscillating signal such as amplitude,phase,frequency and damping ratio,which provides the possibility of oscillation mode and damping analysis,but the existing signal analysis method is mostly anti-Gaussian.Colored noise and multi-channel measurement system output signal calculation efficiency needs to be improved.Aiming at this deficiency,this paper proposes a low-frequency oscillation mode identification method with strong adaptability.This method is based on wide-area multi-channel measurement signal,has strong anti-noise ability(including anti-colored Gaussian noise)and excellent parallelism Calculate ability.Firstly,the characteristic analysis of the low-frequency oscillating signal of the wide-area measurement system is carried out,and it is found that it is a multi-channel signal with mixed noise(including both Gaussian white noise and Gaussian colored noise).Based on this,a single channel pattern identification method with strong anti-noise ability is proposed.In this method,Resonance-based signal sparsity decomposition(RSSD)is chosen as the main means of signal preprocessing,and Prony is chosen as the main method of pattern recognition to form a single-channel pattern recognition method with stronger adaptability-RSSD+Prony method,and its effectiveness is analyzed in multi-channel signals.The numerical simulation of single-channel signal shows that the RSSD+Prony algorithm can accurately obtain the oscillation mode information,and the sensitivity to Gaussian colored noise is lower than Prony algorithm and morphological filtering+Prony algorithm;when the simulation signal is multi-channel,the accuracy of the method Great discounts,and the phenomenon of missing patterns,so the effectiveness is not guaranteed.Secondly,for the problems of single channel processing methods for multi-channel low-frequency oscillation signals,a pattern recognition method for multi-channel signal multi-channel processing in the background of Gaussian colored noise is proposed.,and the application flow in the power system is given.In this method,the RSSD is used as the pre-processing method,and the Independent component algorithm(ICA)is selected for pattern identification.The main principle and implementation steps of the method are emphatically studied.The validity of the method is verified by using single channel and multi-channel numerical signals in terms of rapidity,accuracy,reliability,etc.The numerical simulation results show that RSSD+ICA has good anti-noise,high computational efficiency and strong parallelism.Finally,the overall process of algorithm verification under the power system standard example is proposed,and the interface of the implementation is designed.In this process,the effectiveness of the RSSD+ICA algorithm is verified by using the 4-machine 2-area and the 8-machine 36-node example.The simulation results show that RSSD+ICA is a kind of low-frequency oscillation mode identification method with strong adaptability.It has strong anti-noise ability(including anti-Gaussian colored noise)and excellent parallel computing ability based on wide-area multi-channel measurement signals.It can significantly improve the speed and accuracy of the dominant mode identification and has practical value.