Research On Propagation Characteristics Of Voltage Sag Based On Data Driven And Flow Calculation

With the development of new energy sources and the access of sensitive equipment,the grid environment has become increasingly complex,and power quality problems have also increased.Among them,the voltage sag accounts for the largest proportion,and it has the characteristics of great harm,inevitable and unpredictable.At the same time,with the electronicization of source-grid-load power,the impact of sags has become wider,and the spread of voltage sags in the power grid will cause a large number of sensitive equipment to malfunction and bring economic losses to users.Realizing the real-time extraction of the propagation trajectory of the voltage sag and mining the propagation characteristics hidden in it will help grid companies formulate governance plans,reduce user losses and resolve disputes.There are two shortcomings in the traditional research on the propagation characteristics of voltage sags.On the one hand,under the background of increasingly complex power grid environment,it is difficult for methods based on mechanism analysis to extract sag characteristics comprehensively and accurately.On the other hand,most of the existing research methods are based on isolated time sections,combined with typical working conditions and typical faults using mechanism analysis research methods,which cannot fully reflect the time series evolution characteristics of the system dynamic characteristics,and cannot fully and accurately reveal the propagation process of sag events.In response to the above problems,this article proposes a data-driven and streaming calculation-based voltage sag propagation characteristics research method,which is mainly divided into the following three steps:First,the use of deep learning bidirectional long-term short-term memory network and attention mechanism to construct voltage Sag classification and recognition model;then,based on Spark Streaming,build a multi-point parallel pattern recognition framework for regional power grids,integrate the trained voltage sag classification and recognition model,and design a sliding window-based propagation trajectory extraction algorithm to achieve sag Real-time extraction of the propagation trajectory;finally,the propagation trajectory data is preprocessed,and a multi-dimensional association rule algorithm is used for mining,so as to obtain the voltage sag propagation characteristics.In this paper,the IEEE 14-node system is used as a model and simulated in PSCAD to verify the three steps respectively.The experimental results show that the proposed method can extract the sag propagation trajectory in real time when a sag occurs in the regional power grid,and effectively mine the sag propagation characteristics through the accumulated sag propagation trajectory data.