Study On Voltage Sag Source Identification And Location

Voltage sag is the most typical and most frequent problem of power quality.Voltage sag will cause problems such as equipment outage,factory shutdown,residential power interruption,etc.,directly and indirectly bring economic losses,and cause serious social impact.Therefore,it is necessary to analyze voltage sags in detail.The popularization of intelligent monitoring equipment and the development of intelligent algorithm in modern power system bring new ideas to the monitoring and analysis of voltage sag,which makes it possible to identify and locate the sag events quickly,eliminate faults quickly,and focus on the high incidence areas so as to reduce or even avoid the generation of sag.Firstly,this paper focuses on the problem of distinguishing the characteristics of different types of sag sources.The electrical characteristics of voltage sag caused by single-phase short circuit,two-phase short circuit,two-phase grounding,three-phase short circuit,induction motor starting and no-load transformer switching are theoretically derived and simulated.The direct grounding system and the non effective grounding system are considered respectively in the theoretical analysis and simulation,and the positive and negative zero sequence composite network is used for calculation,and the analyzed sag source is simulated in Matlab/Simulink.The comparison highlights the characteristics of all kinds of sag sources,which provides theoretical and simulation basis for identification and location of sag sources.Then,aiming at the identification of voltage sag sources including lightning,the voltage sag caused by lightning fault is analyzed separately.Combined with the simulation model and the measured data of the sag monitoring system,five sag waveform characteristic indexes are proposed for the waveform characteristics of short circuit fault,lightning fault,no-load transformer switching and motor starting,and a sag source identification classifier based on particle swarm optimization decision tree support vector machine is designed.The effectiveness of the algorithm is verified by the actual monitoring data of the power grid sag terminal.The accuracy of sag classification is more than 90%,and the performance of the algorithm is compared with other classifiers.The results show that the performance of the proposed classification algorithm is better.Finally,aiming at the problem that most of the existing sag location algorithms require too much monitoring points or cannot locate accurately,the electrical characteristics of the sag caused by the fault are analyzed,and the node ground fault feature database is established by using the distribution grid simulation model to train the transition resistance estimation model based on the limit learning machine.According to the current fault information,the transition resistance estimate is obtained to calculate the distance between the current fault information and the section characteristic line,and then the section where the current fault is located is screened out.Based on admittance matrix equation,the optimal search model of fault point is established,and the specific location of fault point is estimated in the range of fault feature.Based on the IEEE-33 node model of typical distribution grid,the proposed algorithm is verified,and the better location result is obtained,which proves that it has higher engineering practicability.The fault identification method proposed in this paper has high identification accuracy and better performance in practical application.The fault location method proposed in this paper only needs the power supply node as the monitoring point for the common radial distribution grid.It can directly use the power quality monitoring terminal to locate the single-phase ground fault in the grid,which is more practical than the full information fault location method.